diff --git a/.env.docker.example b/.env.docker.example
index cf109b1..ca5a646 100644
--- a/.env.docker.example
+++ b/.env.docker.example
@@ -20,5 +20,10 @@ MONITOR_DB_PATH=/app/data/monitor.db
 # 如需东方财富 suggest 补全，请在真实 .env 中配置，不要提交真实 token
 EASTMONEY_TOKEN=
 
+# ====== 美股数据源 ======
+# 美股行情主源为 yfinance（无需 key）；Finnhub 作为备选源及美股新闻源。
+# 未配置时仅用 yfinance，美股个股新闻会跳过。https://finnhub.io 免费注册获取 key
+FINNHUB_API_KEY=
+
 # ====== 可选推送配置 ======
 SERVERCHAN_KEY=
diff --git a/.gitignore b/.gitignore
index c4b7850..03c1062 100644
--- a/.gitignore
+++ b/.gitignore
@@ -64,6 +64,7 @@ logs/
 # Local runtime data
 data/*.db
 data/agent_memory/
+data/strategy_analysis_cache.json
 data/user_settings.json
 
 # Temporary files
diff --git a/README.md b/README.md
index c0c6a37..99b029a 100644
--- a/README.md
+++ b/README.md
@@ -56,6 +56,14 @@ OpenAshare 试图把这些碎片收进一个可控的工作台：
 - 研究进度事件流，适合较长的分析任务
 - 盘前、盘中、午间、盘后不同研究节奏提示
 
+### 多市场支持
+
+- **A股**：上交所 / 深交所 / 创业板 / 科创板（数据源 Ashare + akshare）
+- **港股**：如 `00700.HK`（数据源腾讯）
+- **美股**：如 `AAPL`、`NVDA`、`BRK.B`（主源 yfinance，备选 Finnhub）
+- 三市场统一搜索：在同一界面里搜 `茅台`、`腾讯`、`苹果` / `AAPL` 均可
+- 美股技术分析、指标、信号与 A股 完全对齐；个股新闻走 Finnhub（需配置 `FINNHUB_API_KEY`），A股 特有的资金流/龙虎榜对美股自动跳过
+
 ### 新闻与热点
 
 - 个股新闻与全局市场消息浏览
diff --git a/api/main.py b/api/main.py
index a816107..c900791 100644
--- a/api/main.py
+++ b/api/main.py
@@ -263,6 +263,44 @@ def _warm_read_caches() -> None:
         hotspot_service.list_hotspots(limit=10)
     except Exception:
         logger.exception("Warmup failed for hotspots")
+    _warm_stock_pool()
+
+
+def _warm_stock_pool() -> None:
+    """后台预热股池中部分个股的行情分析（不含 AI），让用户首次点击命中缓存。
+
+    通过 WARMUP_STOCK_LIMIT 控制预热数量（默认 12，设为 0 可关闭）。
+    """
+    try:
+        limit = int(os.getenv("WARMUP_STOCK_LIMIT", "12"))
+    except ValueError:
+        limit = 12
+    if limit <= 0:
+        return
+
+    try:
+        from ashare.stock_pool import load_stock_pool
+
+        pool = load_stock_pool()
+    except Exception:
+        logger.exception("Warmup failed to load stock pool")
+        return
+
+    # 多个别名可能指向同一代码，按出现顺序去重后截断。
+    seen: set[str] = set()
+    codes: list[str] = []
+    for code in pool.values():
+        if code and code not in seen:
+            seen.add(code)
+            codes.append(code)
+        if len(codes) >= limit:
+            break
+
+    for code in codes:
+        try:
+            stock_service.get_stock_analysis(code, include_ai=False)
+        except Exception:
+            logger.warning("Warmup failed for stock %s", code, exc_info=True)
 
 
 @asynccontextmanager
@@ -847,7 +885,13 @@ def list_hotspots(request: Request, limit: int = Query(10, ge=1, le=20)) -> Resp
 def list_global_news(request: Request, limit: int = Query(20, ge=1, le=50)) -> Response:
     try:
         payload = news_service.get_global_news(limit=limit)
-        return _cached_json_response(request, payload, max_age=60, stale_while_revalidate=180)
+        return _cached_json_response(
+            request,
+            payload,
+            max_age=60,
+            stale_while_revalidate=180,
+            no_store=not bool(payload),
+        )
     except Exception as exc:
         logger.exception("get_global_news failed")
         return JSONResponse(content=[], headers={"Cache-Control": "no-store"})
diff --git a/api/schemas.py b/api/schemas.py
index a139b9a..09e52a3 100644
--- a/api/schemas.py
+++ b/api/schemas.py
@@ -122,6 +122,16 @@ class HotspotRelatedStock(BaseModel):
     reason: str
 
 
+class HotspotHeatBreakdown(BaseModel):
+    trading_activity: float = 0
+    discussion_heat: float = 0
+    news_heat: float = 0
+    alert_count: int = 0
+    rank_hits: int = 0
+    attention_level: Literal["focus", "caution", "watch"] = "watch"
+    basis: List[str] = Field(default_factory=list)
+
+
 class HotspotItem(BaseModel):
     topic_name: str
     heat_score: float
@@ -130,6 +140,7 @@ class HotspotItem(BaseModel):
     trend_direction: Literal["up", "down", "flat"] = "flat"
     ai_summary: Optional[str] = None
     source: str = "derived"
+    heat_breakdown: HotspotHeatBreakdown = Field(default_factory=HotspotHeatBreakdown)
 
 
 class HotspotHistoryPoint(BaseModel):
@@ -270,6 +281,7 @@ class StrategyHoldingAnalysis(BaseModel):
     action_reason: str = ""
     trigger_hits: List[str] = Field(default_factory=list)
     alerts: List[str] = Field(default_factory=list)
+    analysis_status: Literal["live", "cached", "degraded", "local"] = "live"
 
 
 class StrategyTodoItem(BaseModel):
diff --git a/api/services.py b/api/services.py
index 109072f..9c602c8 100644
--- a/api/services.py
+++ b/api/services.py
@@ -1,6 +1,7 @@
 from __future__ import annotations
 
 import copy
+import json
 import sqlite3
 import time
 import re
@@ -24,6 +25,7 @@
     AgentResponse,
     AgentHistoryTurn,
     GlobalNewsItem,
+    HotspotHeatBreakdown,
     HotspotDetailResponse,
     HotspotHistoryPoint,
     HotspotItem,
@@ -549,17 +551,32 @@ def get_stock_analysis(
         basic_data = bundle.analysis.get("基础数据", {})
         latest = bundle.dataframe.iloc[-1]
         # K 线图默认展示最近约 5 年日 K（~250 交易日 * 5），方便观察中长期趋势
-        chart_df = bundle.dataframe.tail(1250).copy().reset_index()
+        # 按列向量化提取，避免对上千行使用 DataFrame.iterrows（每行构造 Series，开销大）。
+        chart_df = bundle.dataframe.tail(1250)
+        chart_dates = [str(idx)[:10] for idx in chart_df.index]
+
+        def _column_values(name: str) -> list:
+            if name not in chart_df.columns:
+                return [None] * len(chart_df)
+            return [_safe_float(value) for value in chart_df[name].to_numpy()]
+
         chart_series = [
             {
-                "date": str(row[chart_df.columns[0]])[:10],
-                "open": _safe_float(row.get("open")),
-                "high": _safe_float(row.get("high")),
-                "low": _safe_float(row.get("low")),
-                "close": _safe_float(row.get("close")),
-                "volume": _safe_float(row.get("volume")),
+                "date": date,
+                "open": open_,
+                "high": high,
+                "low": low,
+                "close": close,
+                "volume": volume,
             }
-            for _, row in chart_df.iterrows()
+            for date, open_, high, low, close, volume in zip(
+                chart_dates,
+                _column_values("open"),
+                _column_values("high"),
+                _column_values("low"),
+                _column_values("close"),
+                _column_values("volume"),
+            )
         ]
         indicators = {
             key: _safe_float(latest.get(key))
@@ -616,7 +633,7 @@ def get_stock_analysis(
             chart_series=chart_series,
             metadata={
                 "data_points": len(bundle.dataframe),
-                "source": "akshare",
+                "source": "yfinance/finnhub" if infer_market(quote.stock_code) == "us" else "akshare",
                 "generated_at": quote.timestamp.isoformat(),
             },
         )
@@ -716,7 +733,7 @@ def get_global_news(self, limit: int = 20) -> List[GlobalNewsItem]:
 
         if not records:
             try:
-                records.extend(self._global_news_from_alerts())
+                records.extend(self._global_news_from_alerts(min_impact=2))
             except Exception:
                 pass
 
@@ -737,7 +754,7 @@ def get_global_news(self, limit: int = 20) -> List[GlobalNewsItem]:
                 result.append(GlobalNewsItem(**item))
             except Exception:
                 pass  # skip invalid item
-        self._response_cache.set(cache_key, result, 90)
+        self._response_cache.set(cache_key, result, 90 if result else 15)
         return result
 
     def get_context_news_for_query(self, query: str, limit: int = 6) -> List[GlobalNewsItem]:
@@ -845,13 +862,14 @@ def _normalize_global_news_dataframe(self, data_frame: Optional[pd.DataFrame], s
             )
         return normalized
 
-    def _global_news_from_alerts(self) -> List[Dict[str, Any]]:
+    def _global_news_from_alerts(self, min_impact: int = 3) -> List[Dict[str, Any]]:
         items: List[Dict[str, Any]] = []
         for alert in self.state_store.get_recent_alerts(limit=120):
             title = alert.get("title", "")
             summary = alert.get("summary", "")
             topic_meta = self._classify_global_topic(title, summary, alert.get("source") or "monitor")
-            if topic_meta["impact_level"] < 3:
+            impact_level = max(int(topic_meta["impact_level"]), int(alert.get("priority", 1)))
+            if impact_level < min_impact:
                 continue
             items.append(
                 {
@@ -864,7 +882,7 @@ def _global_news_from_alerts(self) -> List[Dict[str, Any]]:
                     "topic": topic_meta["topic"],
                     "region": topic_meta["region"],
                     "sentiment": _sentiment_from_summary(summary),
-                    "impact_level": topic_meta["impact_level"],
+                    "impact_level": min(impact_level, 5),
                     "url": alert.get("raw_payload", {}).get("url"),
                     "related_symbols": topic_meta["related_symbols"],
                     "raw_payload": alert.get("raw_payload", {}),
@@ -1239,6 +1257,15 @@ def list_hotspots(self, limit: int = 10) -> List[HotspotItem]:
         topic_counter: Counter[str] = Counter()
         topic_stocks: Dict[str, List[HotspotRelatedStock]] = defaultdict(list)
         topic_reasons: Dict[str, List[str]] = defaultdict(list)
+        topic_breakdowns: Dict[str, Dict[str, float]] = defaultdict(
+            lambda: {
+                "trading_activity": 0.0,
+                "discussion_heat": 0.0,
+                "news_heat": 0.0,
+                "alert_count": 0.0,
+                "rank_hits": 0.0,
+            }
+        )
 
         for alert in alerts:
             title = alert.get("title", "")
@@ -1246,10 +1273,19 @@ def list_hotspots(self, limit: int = 10) -> List[HotspotItem]:
             stock_name = alert.get("stock_name", "")
             stock_code = alert.get("stock_code", "")
             priority = max(1, int(alert.get("priority", 1)))
+            event_type = str(alert.get("event_type") or "")
             for sector_name, sector in sectors.items():
                 if not self._sector_keywords_match(sector["keywords"], title, summary, stock_name, stock_code):
                     continue
-                topic_counter[sector_name] += priority
+                breakdown = topic_breakdowns[sector_name]
+                breakdown["alert_count"] += 1
+                if event_type == "fund_flow" or any(token in title + summary for token in ["资金流", "净流入", "净流出", "放量", "成交"]):
+                    breakdown["trading_activity"] += priority * 1.8
+                    breakdown["discussion_heat"] += priority * 0.35
+                else:
+                    breakdown["discussion_heat"] += priority
+                    breakdown["news_heat"] += priority * 0.7
+                topic_counter[sector_name] = self._hotspot_total_score(breakdown)
                 topic_reasons[sector_name].append(title or summary)
                 topic_stocks[sector_name].append(
                     HotspotRelatedStock(
@@ -1269,7 +1305,11 @@ def list_hotspots(self, limit: int = 10) -> List[HotspotItem]:
                     " ".join(global_news.related_symbols),
                 ):
                     continue
-                topic_counter[sector_name] += max(2, global_news.impact_level)
+                impact = max(2, global_news.impact_level)
+                breakdown = topic_breakdowns[sector_name]
+                breakdown["discussion_heat"] += impact
+                breakdown["news_heat"] += impact * 1.2
+                topic_counter[sector_name] = self._hotspot_total_score(breakdown)
                 topic_reasons[sector_name].append(global_news.title)
                 topic_stocks[sector_name].extend(sector["stocks"][:2])
 
@@ -1281,7 +1321,13 @@ def list_hotspots(self, limit: int = 10) -> List[HotspotItem]:
                 for sector_name, sector in sectors.items():
                     if not self._sector_keywords_match(sector["keywords"], rank_name, rank_code):
                         continue
-                    topic_counter[sector_name] += 1
+                    rank_value = _safe_float(rank.get("rank")) or 50
+                    trading_boost = max(1.0, 6.0 - min(rank_value, 50.0) / 10.0)
+                    breakdown = topic_breakdowns[sector_name]
+                    breakdown["trading_activity"] += trading_boost
+                    breakdown["discussion_heat"] += 1
+                    breakdown["rank_hits"] += 1
+                    topic_counter[sector_name] = self._hotspot_total_score(breakdown)
                     topic_reasons[sector_name].append(f"热度排名 {rank.get('rank', '未知')}")
                     topic_stocks[sector_name].append(
                         HotspotRelatedStock(
@@ -1302,21 +1348,88 @@ def list_hotspots(self, limit: int = 10) -> List[HotspotItem]:
             if not related_stocks or score <= 0:
                 continue
             top_reason = next((item for item in topic_reasons[topic_name] if item), f"{topic_name} 相关消息密度提升")
+            breakdown = topic_breakdowns[topic_name]
+            heat_breakdown = self._build_hotspot_breakdown(breakdown, topic_reasons[topic_name])
             items.append(
                 HotspotItem(
                     topic_name=topic_name,
                     heat_score=float(score),
-                    reason=f"{top_reason[:40]}，累计热度 {int(score)}",
+                    reason=(
+                        f"{top_reason[:40]}，综合热度 {int(score)}"
+                        f"（交易 {breakdown['trading_activity']:.1f} / 讨论 {breakdown['discussion_heat']:.1f}）"
+                    ),
                     related_stocks=related_stocks,
-                    trend_direction="up" if score >= 4 else "flat",
-                    ai_summary=f"{topic_name} 近期被多条消息和异动共同触发，可优先跟踪相关代表股。",
+                    trend_direction="up" if score >= 8 else "flat",
+                    ai_summary=self._build_hotspot_summary(topic_name, heat_breakdown),
                     source="sector_config+monitor+news+hk_rank",
+                    heat_breakdown=heat_breakdown,
                 )
             )
         _HOTSPOTS_CACHE = items
         _HOTSPOTS_CACHE_TIME = time.monotonic()
         return items[: min(limit, len(items))]
 
+    @staticmethod
+    def _hotspot_total_score(breakdown: Dict[str, float]) -> float:
+        return round(
+            breakdown["trading_activity"] * 1.2
+            + breakdown["discussion_heat"]
+            + breakdown["news_heat"] * 0.8
+            + breakdown["rank_hits"] * 1.5,
+            1,
+        )
+
+    @staticmethod
+    def _build_hotspot_breakdown(
+        breakdown: Dict[str, float],
+        reasons: List[str],
+    ) -> HotspotHeatBreakdown:
+        reason_text = " ".join(reasons)
+        risk_tokens = ["减持", "净流出", "下跌", "监管", "处罚", "风险", "亏损", "警惕", "退市", "利空"]
+        has_risk_signal = any(token in reason_text for token in risk_tokens)
+        trading_activity = round(breakdown["trading_activity"], 1)
+        discussion_heat = round(breakdown["discussion_heat"], 1)
+        news_heat = round(breakdown["news_heat"], 1)
+        if has_risk_signal and (trading_activity >= 3 or discussion_heat >= 4):
+            attention_level = "caution"
+        elif trading_activity >= 4 and discussion_heat >= 4:
+            attention_level = "focus"
+        else:
+            attention_level = "watch"
+
+        basis: List[str] = []
+        if trading_activity:
+            basis.append("资金流/热榜/成交异动贡献交易活跃度")
+        if discussion_heat:
+            basis.append("新闻命中、主题词和热榜共同贡献讨论度")
+        if news_heat:
+            basis.append("高影响消息贡献催化强度")
+        if has_risk_signal:
+            basis.append("存在净流出、减持、监管或业绩风险词，需提高警惕")
+
+        return HotspotHeatBreakdown(
+            trading_activity=trading_activity,
+            discussion_heat=discussion_heat,
+            news_heat=news_heat,
+            alert_count=int(breakdown["alert_count"]),
+            rank_hits=int(breakdown["rank_hits"]),
+            attention_level=attention_level,
+            basis=basis[:4],
+        )
+
+    @staticmethod
+    def _build_hotspot_summary(topic_name: str, breakdown: HotspotHeatBreakdown) -> str:
+        level_text = {
+            "focus": "交易活跃和讨论度同时靠前，属于当前优先跟踪方向",
+            "caution": "热度较高但伴随风险信号，适合先警惕再确认",
+            "watch": "有消息或讨论触发，暂以观察和验证持续性为主",
+        }[breakdown.attention_level]
+        return (
+            f"{topic_name}：{level_text}。"
+            f"交易活跃 {breakdown.trading_activity:.1f}，讨论热度 {breakdown.discussion_heat:.1f}，"
+            f"消息催化 {breakdown.news_heat:.1f}。"
+        )
+
     def get_hotspot_detail(self, topic_name: str) -> HotspotDetailResponse:
         cache_key = f"hotspot_detail:{topic_name}"
         cached = self._response_cache.get(cache_key)
@@ -1702,6 +1815,15 @@ def _init_db(self) -> None:
                 connection.execute("ALTER TABLE strategy_holdings ADD COLUMN plan_take_profit REAL")
             if "plan_max_position_pct" not in columns:
                 connection.execute("ALTER TABLE strategy_holdings ADD COLUMN plan_max_position_pct REAL")
+            connection.execute(
+                """
+                CREATE TABLE IF NOT EXISTS strategy_analysis_cache (
+                    cache_key TEXT PRIMARY KEY,
+                    payload TEXT NOT NULL,
+                    updated_at TEXT NOT NULL
+                )
+                """
+            )
 
     def list_holdings(self) -> List[StrategyHolding]:
         with self._connect() as connection:
@@ -1860,6 +1982,32 @@ def delete_holding(self, holding_id: int) -> None:
             if cursor.rowcount == 0:
                 raise NotFoundError(f"Strategy holding {holding_id} not found")
 
+    def get_analysis_cache(self, cache_key: str) -> Optional[str]:
+        with self._connect() as connection:
+            row = connection.execute(
+                "SELECT payload FROM strategy_analysis_cache WHERE cache_key = ?",
+                (cache_key,),
+            ).fetchone()
+        return str(row["payload"]) if row else None
+
+    def set_analysis_cache(self, cache_key: str, payload: str) -> None:
+        now = _now_utc().isoformat()
+        with self._connect() as connection:
+            connection.execute(
+                """
+                INSERT INTO strategy_analysis_cache (cache_key, payload, updated_at)
+                VALUES (?, ?, ?)
+                ON CONFLICT(cache_key) DO UPDATE SET
+                    payload = excluded.payload,
+                    updated_at = excluded.updated_at
+                """,
+                (cache_key, payload, now),
+            )
+
+    def clear_analysis_cache(self, cache_key: str) -> None:
+        with self._connect() as connection:
+            connection.execute("DELETE FROM strategy_analysis_cache WHERE cache_key = ?", (cache_key,))
+
     def _normalize_exit_fields(self, holding: StrategyHolding, now: str) -> Tuple[Optional[float], Optional[str]]:
         if holding.status != "exited":
             return None, None
@@ -1977,6 +2125,8 @@ def __init__(
         self.market_service = market_service or MarketService(stock_service)
         self.store = store or StrategyHoldingStore()
         self.portfolio_store = portfolio_store or PortfolioStore()
+        self.analysis_cache_key = "strategy_holdings"
+        self._analysis_cache_lock = Lock()
 
     def screen_can_slim(
         self,
@@ -2055,6 +2205,7 @@ def list_holdings(self) -> List[StrategyHolding]:
     def create_holding(self, holding: StrategyHolding) -> StrategyHolding:
         created = self.store.create_holding(holding)
         self._sync_portfolio_position_for_code(created.stock_code)
+        self._clear_holdings_analysis_cache()
         return created
 
     def update_holding(self, holding_id: int, holding: StrategyHolding) -> StrategyHolding:
@@ -2065,6 +2216,7 @@ def update_holding(self, holding_id: int, holding: StrategyHolding) -> StrategyH
             codes_to_sync.add(previous.stock_code)
         for code in codes_to_sync:
             self._sync_portfolio_position_for_code(code)
+        self._clear_holdings_analysis_cache()
         return updated
 
     def delete_holding(self, holding_id: int) -> None:
@@ -2072,29 +2224,33 @@ def delete_holding(self, holding_id: int) -> None:
         self.store.delete_holding(holding_id)
         if previous:
             self._sync_portfolio_position_for_code(previous.stock_code)
+        self._clear_holdings_analysis_cache()
 
     def analyze_holdings(self) -> StrategyHoldingAnalysisResponse:
         holdings = self.store.list_holdings()
         if not holdings:
-            return StrategyHoldingAnalysisResponse(
-                total_cost=0,
-                total_market_value=0,
-                total_pnl=0,
-                total_pnl_pct=0,
-                total_realized_pnl=0,
-                holding_count=0,
-                active_count=0,
-                watching_count=0,
-                planned_count=0,
-                weakening_count=0,
-                exited_count=0,
-                invalidated_count=0,
-                win_rate_pct=0,
-                average_score=0,
-                todo_items=[],
-                review_items=[],
-                holdings=[],
-            )
+            return self._empty_holdings_analysis()
+
+        cached = self._read_cached_holdings_analysis(holdings)
+        if cached:
+            return cached
+        return self._build_snapshot_holdings_analysis(holdings)
+
+    def refresh_holdings(self) -> StrategyHoldingAnalysisResponse:
+        holdings = self.store.list_holdings()
+        previous = self._read_cached_holdings_analysis(holdings) if holdings else None
+        analysis = self._compute_holdings_analysis()
+        if previous:
+            analysis = self._merge_refresh_analysis_with_previous(analysis, previous)
+        self._write_cached_holdings_analysis(analysis)
+        return analysis
+
+    def _compute_holdings_analysis(self) -> StrategyHoldingAnalysisResponse:
+        holdings = self.store.list_holdings()
+        if not holdings:
+            analysis = self._empty_holdings_analysis()
+            self._write_cached_holdings_analysis(analysis)
+            return analysis
 
         try:
             market_regime = self.market_service.get_market_regime()
@@ -2111,11 +2267,31 @@ def analyze_holdings(self) -> StrategyHoldingAnalysisResponse:
             )
 
         analyses: List[StrategyHoldingAnalysis] = []
-        for holding in holdings:
-            try:
-                analyses.append(self._analyze_single_holding(holding, market_regime))
-            except Exception as exc:
-                analyses.append(self._build_unavailable_holding_analysis(holding, market_regime, exc))
+        max_workers = min(4, len(holdings))
+        executor = ThreadPoolExecutor(max_workers=max_workers)
+        try:
+            future_by_holding = [
+                (holding, executor.submit(self._analyze_single_holding, holding, market_regime))
+                for holding in holdings
+            ]
+            done, _ = wait([future for _, future in future_by_holding], timeout=18)
+            for holding, future in future_by_holding:
+                if future not in done:
+                    future.cancel()
+                    analyses.append(
+                        self._build_unavailable_holding_analysis(
+                            holding,
+                            market_regime,
+                            TimeoutError("单股策略分析超过 18 秒，已降级为持仓快照"),
+                        )
+                    )
+                    continue
+                try:
+                    analyses.append(future.result())
+                except Exception as exc:
+                    analyses.append(self._build_unavailable_holding_analysis(holding, market_regime, exc))
+        finally:
+            executor.shutdown(wait=False, cancel_futures=True)
         total_cost = 0.0
         total_market_value = 0.0
         total_realized_pnl = 0.0
@@ -2126,6 +2302,7 @@ def analyze_holdings(self) -> StrategyHoldingAnalysisResponse:
         exited_count = 0
         invalidated_count = 0
         exited_win_count = 0
+        scored_count = 0
         total_score = 0.0
         for item in analyses:
             cost = item.holding.entry_price * item.holding.quantity
@@ -2133,7 +2310,9 @@ def analyze_holdings(self) -> StrategyHoldingAnalysisResponse:
                 total_cost += cost
                 total_market_value += item.market_value
                 total_realized_pnl += item.realized_pnl
-            total_score += item.strategy_score.total
+            if self._has_reusable_holding_analysis(item):
+                total_score += item.strategy_score.total
+                scored_count += 1
             if item.holding.status == "holding":
                 active_count += 1
             elif item.holding.status == "watching":
@@ -2153,7 +2332,7 @@ def analyze_holdings(self) -> StrategyHoldingAnalysisResponse:
         total_pnl = total_market_value - total_cost
         total_pnl_pct = (total_pnl / total_cost * 100) if total_cost else 0
         win_rate_pct = (exited_win_count / exited_count * 100) if exited_count else 0
-        average_score = (total_score / len(holdings)) if holdings else 0
+        average_score = (total_score / scored_count) if scored_count else 0
         todo_items = self._build_todo_items(analyses)
         review_items = self._build_review_items(analyses)
         return StrategyHoldingAnalysisResponse(
@@ -2176,8 +2355,236 @@ def analyze_holdings(self) -> StrategyHoldingAnalysisResponse:
             holdings=analyses,
         )
 
-    def refresh_holdings(self) -> StrategyHoldingAnalysisResponse:
-        return self.analyze_holdings()
+    @staticmethod
+    def _empty_holdings_analysis() -> StrategyHoldingAnalysisResponse:
+        return StrategyHoldingAnalysisResponse(
+            total_cost=0,
+            total_market_value=0,
+            total_pnl=0,
+            total_pnl_pct=0,
+            total_realized_pnl=0,
+            holding_count=0,
+            active_count=0,
+            watching_count=0,
+            planned_count=0,
+            weakening_count=0,
+            exited_count=0,
+            invalidated_count=0,
+            win_rate_pct=0,
+            average_score=0,
+            todo_items=[],
+            review_items=[],
+            holdings=[],
+        )
+
+    def _neutral_market_regime(self) -> MarketRegimeResponse:
+        return MarketRegimeResponse(
+            regime="neutral",
+            score=50,
+            action_bias="市场数据暂不可用，按中性环境处理。",
+            position_guidance="建议轻仓试错。",
+            summary="市场环境数据暂不可用。",
+            notes=[],
+            indices=[],
+            updated_at=_now_utc(),
+        )
+
+    def _build_snapshot_holdings_analysis(self, holdings: List[StrategyHolding]) -> StrategyHoldingAnalysisResponse:
+        market_regime = self._neutral_market_regime()
+        analyses = [
+            self._build_unavailable_holding_analysis(holding, market_regime, ValueError("尚未生成服务端评分缓存"))
+            for holding in holdings
+        ]
+        return self._summarize_holdings_analysis(analyses)
+
+    def _read_cached_holdings_analysis(
+        self,
+        holdings: List[StrategyHolding],
+    ) -> Optional[StrategyHoldingAnalysisResponse]:
+        with self._analysis_cache_lock:
+            try:
+                raw = self.store.get_analysis_cache(self.analysis_cache_key)
+                if not raw:
+                    return None
+                payload = json.loads(raw)
+                cached = StrategyHoldingAnalysisResponse.model_validate(payload.get("analysis", payload))
+            except Exception:
+                return None
+        if not cached.holdings:
+            return None
+        merged = self._merge_cached_holdings_analysis(holdings, cached)
+        if not any(self._has_reusable_holding_analysis(item) for item in merged.holdings):
+            return None
+        return merged
+
+    def _write_cached_holdings_analysis(self, analysis: StrategyHoldingAnalysisResponse) -> None:
+        if analysis.holdings and not any(self._has_reusable_holding_analysis(item) for item in analysis.holdings):
+            return
+        payload = {
+            "updated_at": _now_utc().isoformat(),
+            "analysis": analysis.model_dump(mode="json"),
+        }
+        with self._analysis_cache_lock:
+            self.store.set_analysis_cache(self.analysis_cache_key, json.dumps(payload, ensure_ascii=False))
+
+    def _clear_holdings_analysis_cache(self) -> None:
+        with self._analysis_cache_lock:
+            self.store.clear_analysis_cache(self.analysis_cache_key)
+
+    def _merge_refresh_analysis_with_previous(
+        self,
+        latest: StrategyHoldingAnalysisResponse,
+        previous: StrategyHoldingAnalysisResponse,
+    ) -> StrategyHoldingAnalysisResponse:
+        previous_by_id = {
+            item.holding.id: item
+            for item in previous.holdings
+            if item.holding.id is not None and self._has_reusable_holding_analysis(item)
+        }
+        previous_by_code = {
+            normalize_stock_code(item.holding.stock_code): item
+            for item in previous.holdings
+            if self._has_reusable_holding_analysis(item)
+        }
+        merged: List[StrategyHoldingAnalysis] = []
+        for item in latest.holdings:
+            if self._has_reusable_holding_analysis(item):
+                merged.append(item)
+                continue
+            previous_item = previous_by_id.get(item.holding.id)
+            if previous_item is None:
+                previous_item = previous_by_code.get(normalize_stock_code(item.holding.stock_code))
+            merged.append(self._rebase_cached_holding_analysis(item.holding, previous_item) if previous_item else item)
+        return self._summarize_holdings_analysis(merged)
+
+    def _merge_cached_holdings_analysis(
+        self,
+        holdings: List[StrategyHolding],
+        cached: StrategyHoldingAnalysisResponse,
+    ) -> StrategyHoldingAnalysisResponse:
+        cached_by_id = {
+            item.holding.id: item
+            for item in cached.holdings
+            if item.holding.id is not None and self._has_reusable_holding_analysis(item)
+        }
+        cached_by_code = {
+            normalize_stock_code(item.holding.stock_code): item
+            for item in cached.holdings
+            if self._has_reusable_holding_analysis(item)
+        }
+        market_regime = self._neutral_market_regime()
+        analyses: List[StrategyHoldingAnalysis] = []
+        for holding in holdings:
+            cached_item = cached_by_id.get(holding.id)
+            if cached_item is None:
+                cached_item = cached_by_code.get(normalize_stock_code(holding.stock_code))
+            if cached_item is not None:
+                analyses.append(self._rebase_cached_holding_analysis(holding, cached_item))
+            else:
+                analyses.append(
+                    self._build_unavailable_holding_analysis(
+                        holding,
+                        market_regime,
+                        ValueError("该持仓暂无服务端评分缓存"),
+                    )
+                )
+        return self._summarize_holdings_analysis(analyses)
+
+    @staticmethod
+    def _rebase_cached_holding_analysis(
+        holding: StrategyHolding,
+        cached: StrategyHoldingAnalysis,
+    ) -> StrategyHoldingAnalysis:
+        current_price = holding.exit_price if holding.status == "exited" and holding.exit_price is not None else cached.current_price
+        is_pre_position = holding.status in {"watching", "planned"}
+        cost = holding.entry_price * holding.quantity
+        market_value = 0.0 if is_pre_position else current_price * holding.quantity
+        pnl = 0.0 if is_pre_position else market_value - cost
+        realized_pnl = (
+            ((holding.exit_price or 0) - holding.entry_price) * holding.quantity
+            if holding.status == "exited" and holding.exit_price is not None
+            else 0.0
+        )
+        return cached.model_copy(
+            update={
+                "holding": holding,
+                "current_price": current_price,
+                "market_value": market_value,
+                "pnl": pnl,
+                "pnl_pct": 0.0 if is_pre_position else ((pnl / cost * 100) if cost else 0.0),
+                "realized_pnl": realized_pnl,
+                "realized_pnl_pct": (realized_pnl / cost * 100) if cost and realized_pnl else 0.0,
+                "analysis_status": "cached",
+            }
+        )
+
+    @staticmethod
+    def _has_reusable_holding_analysis(item: StrategyHoldingAnalysis) -> bool:
+        return item.strategy_score.total > 0 and item.analysis_status in {"live", "cached"}
+
+    def _summarize_holdings_analysis(
+        self,
+        analyses: List[StrategyHoldingAnalysis],
+    ) -> StrategyHoldingAnalysisResponse:
+        total_cost = 0.0
+        total_market_value = 0.0
+        total_realized_pnl = 0.0
+        active_count = 0
+        watching_count = 0
+        planned_count = 0
+        weakening_count = 0
+        exited_count = 0
+        invalidated_count = 0
+        exited_win_count = 0
+        scored_count = 0
+        total_score = 0.0
+        for item in analyses:
+            cost = item.holding.entry_price * item.holding.quantity
+            if item.holding.status not in {"watching", "planned"}:
+                total_cost += cost
+                total_market_value += item.market_value
+                total_realized_pnl += item.realized_pnl
+            if self._has_reusable_holding_analysis(item):
+                total_score += item.strategy_score.total
+                scored_count += 1
+            if item.holding.status == "holding":
+                active_count += 1
+            elif item.holding.status == "watching":
+                watching_count += 1
+            elif item.holding.status == "planned":
+                planned_count += 1
+            elif item.holding.status == "weakening":
+                active_count += 1
+                weakening_count += 1
+            elif item.holding.status == "exited":
+                exited_count += 1
+                if item.realized_pnl > 0:
+                    exited_win_count += 1
+            elif item.holding.status == "invalidated":
+                invalidated_count += 1
+
+        total_pnl = total_market_value - total_cost
+        total_pnl_pct = (total_pnl / total_cost * 100) if total_cost else 0
+        win_rate_pct = (exited_win_count / exited_count * 100) if exited_count else 0
+        return StrategyHoldingAnalysisResponse(
+            total_cost=total_cost,
+            total_market_value=total_market_value,
+            total_pnl=total_pnl,
+            total_pnl_pct=total_pnl_pct,
+            total_realized_pnl=total_realized_pnl,
+            holding_count=len(analyses),
+            active_count=active_count,
+            watching_count=watching_count,
+            planned_count=planned_count,
+            weakening_count=weakening_count,
+            exited_count=exited_count,
+            invalidated_count=invalidated_count,
+            win_rate_pct=round(win_rate_pct, 2),
+            average_score=round((total_score / scored_count) if scored_count else 0, 2),
+            todo_items=self._build_todo_items(analyses),
+            review_items=self._build_review_items(analyses),
+            holdings=analyses,
+        )
 
     def _analyze_single_holding(
         self,
@@ -2239,6 +2646,7 @@ def _analyze_single_holding(
             action_reason=action_reason,
             trigger_hits=trigger_hits,
             alerts=alerts[:3],
+            analysis_status="live",
         )
 
     def _build_unavailable_holding_analysis(
@@ -2268,6 +2676,15 @@ def _build_unavailable_holding_analysis(
         trigger_hits = self._detect_trade_plan_hits(holding, fallback_price, market_regime)
         thesis_status = "broken" if holding.status == "invalidated" else "weakening" if holding.status == "weakening" else "active"
         action_label, action_reason = self._build_unavailable_holding_action(holding)
+        fallback_candidate = self._score_candidate_quick(
+            {"code": holding.stock_code, "name": holding.stock_name},
+            "market",
+            holding.source_topic,
+        )
+        factor_notes = {
+            **fallback_candidate.factor_notes,
+            "data": "实时行情或新闻源暂不可用，当前使用本地快速兜底评分；请稍后手动刷新获取完整评分。",
+        }
 
         return StrategyHoldingAnalysis(
             holding=holding,
@@ -2277,17 +2694,15 @@ def _build_unavailable_holding_analysis(
             pnl_pct=pnl_pct,
             realized_pnl=realized_pnl,
             realized_pnl_pct=realized_pnl_pct,
-            strategy_score=StrategyScoreBreakdown(c=0, a=0, n=0, s=0, l=0, i=0, m=0, total=0),
+            strategy_score=fallback_candidate.score,
             thesis_status=thesis_status,
-            factor_notes={
-                "data": "实时行情或评分服务暂时不可用，当前先展示已保存的持仓记录。",
-                "plan": "交易计划字段和持仓状态已保留，可在数据恢复后继续刷新。",
-            },
+            factor_notes=factor_notes,
             invalidation_reason="行情数据暂不可用，当前无法完整验证交易假设。" if holding.status == "invalidated" else None,
             action_label=action_label,
             action_reason=action_reason,
             trigger_hits=trigger_hits,
             alerts=alerts[:3],
+            analysis_status="degraded",
         )
 
     @staticmethod
@@ -2530,7 +2945,7 @@ def _score_candidate(
         market_regime: Optional[MarketRegimeResponse] = None,
     ) -> StrategyCandidate:
         analysis = self.stock_service.get_stock_analysis(candidate["code"], include_ai=False)
-        news_items = self.news_service.get_stock_news(candidate["code"], candidate["name"], limit=5)
+        news_items = self._get_stock_news_quick(candidate["code"], candidate["name"], limit=5)
         chart_series = analysis.chart_series[-120:] if analysis.chart_series else []
         closes = [float(item["close"]) for item in chart_series if item.get("close") is not None]
         volumes = [float(item["volume"]) for item in chart_series if item.get("volume") is not None]
@@ -2675,6 +3090,17 @@ def _score_candidate(
             },
         )
 
+    def _get_stock_news_quick(self, stock_code: str, stock_name: str, limit: int = 5) -> List[NewsItem]:
+        executor = ThreadPoolExecutor(max_workers=1)
+        future = executor.submit(self.news_service.get_stock_news, stock_code, stock_name, limit)
+        try:
+            return future.result(timeout=3)
+        except Exception:
+            future.cancel()
+            return []
+        finally:
+            executor.shutdown(wait=False, cancel_futures=True)
+
     def _build_institutional_note(
         self,
         news_items: List[NewsItem],
diff --git a/app/api/[...path]/route.ts b/app/api/[...path]/route.ts
index 5919598..8c13fbd 100644
--- a/app/api/[...path]/route.ts
+++ b/app/api/[...path]/route.ts
@@ -10,7 +10,7 @@ export const dynamic = "force-dynamic";
 
 function isPublicCacheableGet(path: string[], searchParams: URLSearchParams) {
   if (path.length === 2 && path[0] === "stocks" && path[1] === "search") {
-    return !searchParams.has("request_id");
+    return false;
   }
   if (path.length === 3 && path[0] === "stocks" && path[2] === "analysis") {
     return searchParams.get("include_ai") === "false";
diff --git a/app/charts/page.tsx b/app/charts/page.tsx
index 3c707b0..ea72b99 100644
--- a/app/charts/page.tsx
+++ b/app/charts/page.tsx
@@ -1,143 +1,22 @@
 "use client";
 
-import { useCallback, useState } from "react";
-import { CandlestickChart } from "@/components/candlestick-chart";
-import { getStockAnalysis, searchStocks } from "@/lib/api";
-import type { StockAnalysisResponse, StockSearchResult } from "@/lib/types";
-
-type ChartPoint = {
-  date: string;
-  open: number | null;
-  high: number | null;
-  low: number | null;
-  close: number | null;
-  volume?: number | null;
-};
-
-export default function ChartsPage() {
-  const [query, setQuery] = useState("");
-  const [results, setResults] = useState<StockSearchResult[]>([]);
-  const [analysis, setAnalysis] = useState<StockAnalysisResponse | null>(null);
-  const [loading, setLoading] = useState(false);
-  const [searching, setSearching] = useState(false);
-
-  const onSearch = useCallback(() => {
-    const q = query.trim();
-    if (!q) return;
-    setSearching(true);
-    searchStocks(q)
-      .then(setResults)
-      .catch(() => setResults([]))
-      .finally(() => setSearching(false));
-    setAnalysis(null);
-  }, [query]);
-
-  const onSelectStock = useCallback((code: string) => {
-    setLoading(true);
-    getStockAnalysis(code, { includeAi: false })
-      .then(setAnalysis)
-      .catch(() => setAnalysis(null))
-      .finally(() => setLoading(false));
-  }, []);
-
-  const chartData: ChartPoint[] = analysis?.chart_series ?? [];
-  const closeValues = chartData
-    .map((point) => point.close)
-    .filter((value): value is number => typeof value === "number" && !Number.isNaN(value));
-
-  return (
-    <>
-      <section className="panel section">
-        <h1>K 线图</h1>
-        <p className="muted">为重点标的快速拉出一张干净的日 K 线，配合价格与涨跌幅做节奏判断。</p>
-        <div className="form" style={{ marginTop: 16 }}>
-          <input
-            className="input search"
-            value={query}
-            onChange={(e) => setQuery(e.target.value)}
-            onKeyDown={(e) => e.key === "Enter" && onSearch()}
-            placeholder="输入股票名称或代码，例如 招商银行 / sh600036"
-          />
-          <button type="button" className="button" onClick={onSearch} disabled={searching}>
-            {searching ? "搜索中…" : "搜索"}
-          </button>
-        </div>
-      </section>
-
-      {results.length > 0 && (
-        <section className="panel section">
-          <h2>搜索结果</h2>
-          <div className="search-results">
-            {results.map((item) => (
-              <button
-                type="button"
-                className="result-button"
-                key={item.code}
-                onClick={() => onSelectStock(item.code)}
-                disabled={loading}
-              >
-                <strong>{item.name} ({item.code})</strong>
-                <span className="muted">{item.market} · {item.category || "-"}</span>
-              </button>
-            ))}
-          </div>
-        </section>
-      )}
-
-      {analysis && (
-        <section className="panel section">
-          <h2>
-            {analysis.stock_name} ({analysis.stock_code}) — 日 K 线
-          </h2>
-          <div className="metric-grid" style={{ marginBottom: 16 }}>
-            <div className="card">
-              <div className="muted">最新价</div>
-              <strong>{analysis.quote.current_price.toFixed(2)}</strong>
-            </div>
-            <div className="card">
-              <div className="muted">涨跌幅</div>
-              <strong className={analysis.quote.change_pct >= 0 ? "signal-up" : "signal-down"}>
-                {analysis.quote.change_pct.toFixed(2)}%
-              </strong>
-            </div>
-          </div>
-          <CandlestickChart
-            data={chartData}
-            height={420}
-            symbol={`${analysis.stock_name} (${analysis.stock_code})`}
-          />
-          <div className="metric-grid" style={{ marginTop: 18 }}>
-            <div className="card">
-              <div className="muted">MACD</div>
-              <strong>{formatIndicator(analysis.technical_indicators.MACD)}</strong>
-            </div>
-            <div className="card">
-              <div className="muted">DIF / DEA</div>
-              <strong>
-                {formatIndicator(analysis.technical_indicators.DIF)} /{" "}
-                {formatIndicator(analysis.technical_indicators.DEA)}
-              </strong>
-            </div>
-            <div className="card">
-              <div className="muted">RSI</div>
-              <strong>{formatIndicator(analysis.technical_indicators.RSI)}</strong>
-            </div>
-            <div className="card">
-              <div className="muted">K / D / J</div>
-              <strong>
-                {formatIndicator(analysis.technical_indicators.K)} /{" "}
-                {formatIndicator(analysis.technical_indicators.D)} /{" "}
-                {formatIndicator(analysis.technical_indicators.J)}
-              </strong>
-            </div>
-          </div>
-        </section>
-      )}
-    </>
-  );
-}
-
-function formatIndicator(value: number | null | undefined) {
-  if (value == null || Number.isNaN(value)) return "-";
-  return value.toFixed(2);
+import { useEffect } from "react";
+import { useRouter } from "next/navigation";
+
+// K 线图已并入「单股分析」页（panel=chart）。保留此路由仅用于兼容旧链接/书签，
+// 自动重定向到单股分析的 K 线视图。
+export default function ChartsRedirect() {
+  const router = useRouter();
+
+  useEffect(() => {
+    const params = new URLSearchParams(window.location.search);
+    const query = params.get("query");
+    router.replace(
+      query
+        ? `/stocks?query=${encodeURIComponent(query)}&panel=chart#chart`
+        : "/stocks",
+    );
+  }, [router]);
+
+  return <p className="muted" style={{ padding: 24 }}>正在跳转到单股分析…</p>;
 }
diff --git a/app/globals.css b/app/globals.css
index 17f0e11..fa9387a 100644
--- a/app/globals.css
+++ b/app/globals.css
@@ -1401,6 +1401,12 @@ button, input, textarea { font: inherit; }
   background: rgba(19, 19, 19, 0.04);
   border-color: var(--line);
 }
+.nav-links a.active {
+  border-color: rgba(15, 138, 123, 0.26);
+  background: rgba(15, 138, 123, 0.08);
+  color: var(--accent);
+  font-weight: 700;
+}
 .nav-links button {
   border: 1px solid var(--line);
   background: rgba(255, 255, 255, 0.88);
@@ -3076,6 +3082,87 @@ textarea.input {
   padding-top: 16px;
 }
 
+.portfolio-action-strip,
+.portfolio-snapshot-action {
+  display: flex;
+  align-items: center;
+  justify-content: space-between;
+  gap: 14px;
+  margin: 12px 0 16px;
+  padding: 14px;
+  border: 1px solid rgba(15, 138, 123, 0.18);
+  border-radius: 16px;
+  background:
+    linear-gradient(135deg, rgba(15, 138, 123, 0.08), rgba(255, 255, 255, 0.84)),
+    rgba(255, 255, 255, 0.72);
+}
+
+.portfolio-action-strip p {
+  margin: 4px 0 0;
+}
+
+.portfolio-action-kicker,
+.portfolio-snapshot-action span {
+  display: block;
+  margin-bottom: 4px;
+  color: var(--accent);
+  font-size: 0.72rem;
+  font-weight: 800;
+  letter-spacing: 0.08em;
+  text-transform: uppercase;
+}
+
+.portfolio-action-list {
+  display: flex;
+  flex-wrap: wrap;
+  justify-content: flex-end;
+  gap: 8px;
+}
+
+.portfolio-action-chip {
+  display: grid;
+  gap: 2px;
+  min-width: 118px;
+  padding: 9px 11px;
+  border: 1px solid rgba(19, 19, 19, 0.08);
+  border-radius: 12px;
+  background: rgba(255, 255, 255, 0.78);
+  color: var(--ink);
+  text-align: left;
+  cursor: pointer;
+}
+
+.portfolio-action-chip span {
+  color: var(--muted);
+  font-size: 0.72rem;
+}
+
+.portfolio-action-chip strong {
+  color: var(--ink);
+  font-size: 0.86rem;
+}
+
+.portfolio-action-chip:hover {
+  border-color: rgba(15, 138, 123, 0.24);
+  background: rgba(255, 255, 255, 0.96);
+}
+
+.portfolio-entry-compact {
+  display: flex;
+  align-items: center;
+  justify-content: space-between;
+  gap: 14px;
+  margin-top: 12px;
+  padding: 12px 14px;
+  border: 1px dashed rgba(19, 19, 19, 0.14);
+  border-radius: 14px;
+  background: rgba(255, 255, 255, 0.62);
+}
+
+.portfolio-entry-compact p {
+  margin: 0;
+}
+
 .news-warning-strip {
   border-color: rgba(194, 91, 58, 0.18);
   background: linear-gradient(180deg, rgba(194, 91, 58, 0.06), rgba(255, 255, 255, 0.94));
@@ -3175,18 +3262,26 @@ textarea.input {
 .hotspot-detail-panel {
   display: flex;
   flex-direction: column;
-  min-height: clamp(680px, calc(100vh - var(--nav-height) - 96px), 980px);
+  min-height: 0;
 }
 
 .hotspot-board-scroll,
 .hotspot-detail-scroll {
   flex: 1;
   min-height: 0;
-  overflow: auto;
   padding-right: 6px;
   margin-right: -6px;
 }
 
+.hotspot-board-scroll {
+  max-height: min(720px, calc(100vh - var(--nav-height) - 180px));
+  overflow: auto;
+}
+
+.hotspot-detail-scroll {
+  overflow: visible;
+}
+
 .news-summary-panel,
 .news-lead-panel,
 .hotspot-rank-panel,
@@ -3516,6 +3611,20 @@ textarea.input {
   margin-top: 18px;
 }
 
+.news-empty-state {
+  display: grid;
+  gap: 14px;
+  margin-top: 16px;
+  padding: 16px 18px;
+  border: 1px dashed rgba(17, 22, 29, 0.14);
+  border-radius: var(--radius-md);
+  background: rgba(255, 255, 255, 0.64);
+}
+
+.news-empty-state p {
+  margin: 0;
+}
+
 .news-summary-panel .news-action-row {
   padding-top: 2px;
 }
@@ -3838,6 +3947,7 @@ textarea.input {
 }
 
 .hotspot-trend-pill,
+.hotspot-attention-pill,
 .hotspot-stock-count {
   display: inline-flex;
   align-items: center;
@@ -3852,6 +3962,25 @@ textarea.input {
   border: 1px solid rgba(0, 0, 0, 0.06);
 }
 
+.hotspot-attention-pill {
+  border: 1px solid rgba(0, 0, 0, 0.06);
+}
+
+.hotspot-attention-focus {
+  background: rgba(13, 148, 136, 0.11);
+  color: #0f766e;
+}
+
+.hotspot-attention-caution {
+  background: rgba(194, 91, 58, 0.12);
+  color: #b4532a;
+}
+
+.hotspot-attention-watch {
+  background: rgba(17, 22, 29, 0.08);
+  color: #374151;
+}
+
 .hotspot-trend-pill-up {
   background: rgba(194, 91, 58, 0.1);
   color: #b4532a;
@@ -3904,6 +4033,58 @@ textarea.input {
   border: 1px solid rgba(17, 22, 29, 0.08);
 }
 
+.hotspot-basis-box {
+  display: grid;
+  gap: 12px;
+  padding: 16px 18px;
+  border-radius: var(--radius-md);
+  border: 1px solid rgba(15, 138, 123, 0.14);
+  background:
+    linear-gradient(135deg, rgba(15, 138, 123, 0.07), transparent 44%),
+    rgba(255, 255, 255, 0.72);
+}
+
+.hotspot-basis-box h3 {
+  margin: 0;
+}
+
+.hotspot-compact-section,
+.hotspot-compact-details {
+  border: 1px solid rgba(17, 22, 29, 0.08);
+  border-radius: var(--radius-md);
+  background: rgba(255, 255, 255, 0.72);
+}
+
+.hotspot-compact-toggle,
+.hotspot-compact-details summary {
+  display: flex;
+  align-items: center;
+  justify-content: space-between;
+  gap: 14px;
+  width: 100%;
+  padding: 14px 16px;
+  border: 0;
+  background: transparent;
+  color: inherit;
+  cursor: pointer;
+  list-style: none;
+  text-align: left;
+}
+
+.hotspot-compact-details summary::-webkit-details-marker {
+  display: none;
+}
+
+.hotspot-compact-toggle h3,
+.hotspot-compact-details summary h3 {
+  margin: 0;
+  font-size: 1rem;
+}
+
+.hotspot-compact-body {
+  padding: 0 16px 16px;
+}
+
 .hotspot-metric {
   padding: 16px 18px;
   border-radius: var(--radius-md);
@@ -3930,14 +4111,33 @@ textarea.input {
 
 .stock-link-card {
   display: grid;
-  gap: 14px;
-  padding: 18px;
+  grid-template-columns: minmax(0, 1fr) auto;
+  align-items: center;
+  gap: 12px;
+  padding: 12px 14px;
   border-radius: var(--radius-md);
   border: 1px solid rgba(0, 0, 0, 0.06);
   background: linear-gradient(180deg, rgba(255, 255, 255, 0.98), rgba(248, 246, 242, 0.98));
   transition: transform 0.16s ease, box-shadow 0.16s ease;
 }
 
+.stock-link-card p {
+  overflow: hidden;
+  margin-top: 4px;
+  text-overflow: ellipsis;
+  white-space: nowrap;
+}
+
+.stock-link-card .inline-actions {
+  justify-content: flex-end;
+}
+
+.stock-link-card .button {
+  min-height: 30px;
+  padding: 6px 9px;
+  font-size: 0.8rem;
+}
+
 .strategy-factor-grid {
   display: grid;
   grid-template-columns: repeat(auto-fit, minmax(170px, 1fr));
@@ -4096,6 +4296,81 @@ textarea.input {
   align-items: start;
 }
 
+.stocks-focus-bar {
+  position: sticky;
+  top: calc(var(--nav-height) + 8px);
+  z-index: 12;
+  display: grid;
+  grid-template-columns: minmax(220px, 1fr) auto auto;
+  align-items: center;
+  gap: 14px;
+  margin: 0 0 16px;
+  padding: 12px 14px;
+  border: 1px solid rgba(15, 138, 123, 0.18);
+  border-radius: 16px;
+  background:
+    linear-gradient(135deg, rgba(255, 255, 255, 0.96), rgba(246, 242, 236, 0.94)),
+    rgba(255, 255, 255, 0.9);
+  box-shadow: 0 16px 38px rgba(25, 22, 18, 0.08);
+  backdrop-filter: blur(14px);
+}
+
+.stocks-focus-main {
+  display: grid;
+  gap: 3px;
+  min-width: 0;
+}
+
+.stocks-focus-main strong {
+  overflow: hidden;
+  color: var(--ink);
+  font-size: 1.02rem;
+  text-overflow: ellipsis;
+  white-space: nowrap;
+}
+
+.stocks-focus-market {
+  width: fit-content;
+  color: var(--accent);
+  font-size: 0.72rem;
+  font-weight: 800;
+  letter-spacing: 0.08em;
+  text-transform: uppercase;
+}
+
+.stocks-focus-metrics {
+  display: flex;
+  align-items: center;
+  gap: 10px;
+  white-space: nowrap;
+}
+
+.stocks-focus-metrics span:first-child {
+  color: var(--muted);
+  font-size: 0.86rem;
+}
+
+.stocks-focus-metrics strong {
+  margin-left: 4px;
+  color: var(--ink);
+  font-size: 1.06rem;
+}
+
+.stocks-focus-actions {
+  display: flex;
+  align-items: center;
+  justify-content: flex-end;
+  gap: 8px;
+  flex-wrap: wrap;
+}
+
+.stocks-focus-actions .button,
+.stocks-focus-actions a {
+  min-height: 34px;
+  padding: 7px 11px;
+  font-size: 0.82rem;
+}
+
 .stocks-tech-grid {
   grid-template-columns: repeat(2, minmax(0, 1fr));
   align-items: start;
@@ -4238,6 +4513,7 @@ textarea.input {
 }
 
 .portfolio-holding-card {
+  scroll-margin-top: calc(var(--nav-height) + 16px);
   border-radius: 16px;
   background:
     radial-gradient(circle at 96% -8%, rgba(15, 138, 123, 0.1), transparent 36%),
@@ -4249,6 +4525,14 @@ textarea.input {
   grid-template-columns: repeat(3, minmax(0, 1fr));
 }
 
+.portfolio-holding-grid[data-layout="single"] {
+  grid-template-columns: minmax(0, 1fr);
+}
+
+.portfolio-holding-grid[data-layout="pair"] {
+  grid-template-columns: repeat(2, minmax(0, 1fr));
+}
+
 .portfolio-holding-card h3 {
   margin-bottom: 8px;
   font-size: 1.14rem;
@@ -4503,10 +4787,22 @@ textarea.input {
 }
 
 .detail-news-item {
-  padding: 16px 0;
+  padding: 11px 0;
   border-bottom: 1px solid rgba(0, 0, 0, 0.06);
 }
 
+.detail-news-item h4 {
+  margin-top: 6px;
+  font-size: 0.96rem;
+}
+
+.detail-news-item p {
+  display: -webkit-box;
+  overflow: hidden;
+  -webkit-box-orient: vertical;
+  -webkit-line-clamp: 2;
+}
+
 .detail-news-item:last-child {
   border-bottom: none;
   padding-bottom: 0;
@@ -4739,6 +5035,42 @@ textarea.input {
   .nav-demo-link {
     display: block;
   }
+  .portfolio-action-strip,
+  .portfolio-snapshot-action {
+    align-items: stretch;
+    flex-direction: column;
+  }
+  .portfolio-action-list {
+    justify-content: flex-start;
+  }
+  .portfolio-action-chip {
+    width: 100%;
+  }
+  .portfolio-entry-compact {
+    align-items: stretch;
+    flex-direction: column;
+  }
+  .stock-link-card {
+    grid-template-columns: 1fr;
+  }
+  .stock-link-card .inline-actions {
+    justify-content: flex-start;
+  }
+  .stocks-focus-bar {
+    position: static;
+    grid-template-columns: 1fr;
+    gap: 10px;
+  }
+  .stocks-focus-metrics,
+  .stocks-focus-actions {
+    justify-content: flex-start;
+  }
+  .stocks-focus-actions .button,
+  .stocks-focus-actions a {
+    flex: 1 1 96px;
+    justify-content: center;
+    text-align: center;
+  }
   .main-content {
     padding: 10px 10px 22px;
   }
@@ -4770,7 +5102,10 @@ textarea.input {
   .ai-analysis-grid {
     grid-template-columns: 1fr;
   }
-  .portfolio-holding-grid {
+  .portfolio-holding-grid,
+  .portfolio-holding-grid[data-layout="single"],
+  .portfolio-holding-grid[data-layout="pair"],
+  .portfolio-holding-grid[data-layout="many"] {
     grid-template-columns: 1fr;
   }
   .hero-grid {
diff --git a/app/hotspots/hotspots-page-client.tsx b/app/hotspots/hotspots-page-client.tsx
index 53fe6d5..99ba627 100644
--- a/app/hotspots/hotspots-page-client.tsx
+++ b/app/hotspots/hotspots-page-client.tsx
@@ -26,6 +26,8 @@ export function HotspotsPageClient() {
   const [marketRegimeError, setMarketRegimeError] = useState<string | null>(null);
   const [error, setError] = useState<string | null>(null);
   const [manualRefreshing, setManualRefreshing] = useState(false);
+  const [strategyPanelOpen, setStrategyPanelOpen] = useState(false);
+  const [historyPanelOpen, setHistoryPanelOpen] = useState(false);
   const detailRequestRef = useRef(0);
   const screeningRequestRef = useRef(0);
   const screeningCarouselRef = useRef<HTMLDivElement | null>(null);
@@ -138,6 +140,11 @@ export function HotspotsPageClient() {
     loadDetail(selectedTopic.topic_name);
   }, [selectedTopic?.topic_name, loadDetail]);
 
+  useEffect(() => {
+    setStrategyPanelOpen(false);
+    setHistoryPanelOpen(false);
+  }, [selectedTopic?.topic_name]);
+
   useEffect(() => {
     if (screenScope === "hotspot" && selectedTopic?.topic_name) {
       loadScreening("hotspot", selectedTopic.topic_name);
@@ -210,7 +217,7 @@ export function HotspotsPageClient() {
         <div className="section-kicker">Sector Radar</div>
         <h1>热点板块</h1>
         <p className="muted">
-          从板块视角梳理当前市场最热的交易方向，结合代表个股、催化消息和热度变化，帮你快速锁定值得跟踪的主题。
+          按交易活跃、讨论度和消息催化排序，先看哪些值得关注，哪些需要警惕。
         </p>
       </section>
 
@@ -257,6 +264,7 @@ export function HotspotsPageClient() {
                   {sortedHotspots.map((item, index) => {
                     const active = item.topic_name === selectedTopic?.topic_name;
                     const linkedStocks = summarizeLinkedStocks(item);
+                    const breakdown = normalizeHotspotBreakdown(item);
                     return (
                       <Link
                         className={`hotspot-board-item ${active ? "active" : ""}`}
@@ -268,18 +276,24 @@ export function HotspotsPageClient() {
                         <div className="hotspot-board-content">
                           <div className="hotspot-board-topline">
                             <div className="hotspot-title-block">
-                              <span className="hotspot-eyebrow">Sector Pulse</span>
+                              <span className="hotspot-eyebrow">Market Attention</span>
                               <h3>{item.topic_name}</h3>
                             </div>
                             <div className="hotspot-heat-pill">
-                              <span>热度</span>
+                              <span>综合</span>
                               <strong>{item.heat_score.toFixed(0)}</strong>
                             </div>
                           </div>
-                          <p>{truncate(item.ai_summary || item.reason, 120)}</p>
+                          <p>{truncate(item.ai_summary || item.reason, 86)}</p>
                           <div className="hotspot-meta-row">
-                            <span className={`hotspot-trend-pill hotspot-trend-pill-${item.trend_direction}`}>
-                              {trendLabel(item.trend_direction)}
+                            <span className={`hotspot-attention-pill hotspot-attention-${breakdown.attention_level}`}>
+                              {attentionLabel(breakdown.attention_level)}
+                            </span>
+                            <span className="hotspot-stock-count">
+                              交易 {breakdown.trading_activity.toFixed(1)}
+                            </span>
+                            <span className="hotspot-stock-count">
+                              讨论 {breakdown.discussion_heat.toFixed(1)}
                             </span>
                             <span className="hotspot-stock-count">
                               代表股 {linkedStocks.visible.length + linkedStocks.extraCount}
@@ -333,32 +347,50 @@ export function HotspotsPageClient() {
                 {detailLoading ? <p className="muted">详情更新中，先展示上一版内容…</p> : null}
                 <div className="stack">
                   <div className="hotspot-detail-summary">
-                    <p>{currentDetail.topic.ai_summary || currentDetail.topic.reason}</p>
+                    <p>{truncate(currentDetail.topic.ai_summary || currentDetail.topic.reason, 128)}</p>
                   </div>
 
                   <div className="metric-grid">
                     <div className="hotspot-metric">
-                      <span>热度趋势</span>
-                      <strong>{trendLabel(currentDetail.topic.trend_direction)}</strong>
+                      <span>关注判断</span>
+                      <strong>{attentionLabel(normalizeHotspotBreakdown(currentDetail.topic).attention_level)}</strong>
+                    </div>
+                    <div className="hotspot-metric">
+                      <span>交易活跃</span>
+                      <strong>{normalizeHotspotBreakdown(currentDetail.topic).trading_activity.toFixed(1)}</strong>
                     </div>
                     <div className="hotspot-metric">
-                      <span>关联个股</span>
-                      <strong>{currentDetail.topic.related_stocks.length}</strong>
+                      <span>讨论热度</span>
+                      <strong>{normalizeHotspotBreakdown(currentDetail.topic).discussion_heat.toFixed(1)}</strong>
                     </div>
                     <div className="hotspot-metric">
-                      <span>催化消息</span>
-                      <strong>{currentDetail.related_news.length}</strong>
+                      <span>消息催化</span>
+                      <strong>{normalizeHotspotBreakdown(currentDetail.topic).news_heat.toFixed(1)}</strong>
                     </div>
                   </div>
 
+                  {normalizeHotspotBreakdown(currentDetail.topic).basis.length ? (
+                    <div className="hotspot-basis-box">
+                      <div className="detail-block-head">
+                        <h3>为什么排在这里</h3>
+                        <span className="muted">交易活跃 + 讨论度 + 消息催化</span>
+                      </div>
+                      <div className="tag-list">
+                        {normalizeHotspotBreakdown(currentDetail.topic).basis.slice(0, 2).map((item) => (
+                          <span className="tag" key={item}>{item}</span>
+                        ))}
+                      </div>
+                    </div>
+                  ) : null}
+
                   <section className="detail-block">
                     <div className="detail-block-head">
-                      <h3>板块关联个股</h3>
-                      <span className="muted">板块内优先关注的代表标的</span>
+                      <h3>代表股</h3>
+                      <span className="muted">先看前三只</span>
                     </div>
                     {currentDetail.topic.related_stocks.length ? (
                       <div className="stock-link-list">
-                        {currentDetail.topic.related_stocks.map((stock) => (
+                        {currentDetail.topic.related_stocks.slice(0, 3).map((stock) => (
                           <div className="stock-link-card" key={`${currentDetail.topic.topic_name}-${stock.stock_code}`}>
                             <div>
                               <strong>
@@ -383,18 +415,32 @@ export function HotspotsPageClient() {
                             </div>
                           </div>
                         ))}
+                        {currentDetail.topic.related_stocks.length > 3 ? (
+                          <p className="muted">其余 {currentDetail.topic.related_stocks.length - 3} 只已收起，先处理上面的代表股。</p>
+                        ) : null}
                       </div>
                     ) : (
                       <p className="muted">暂无关联股票。</p>
                     )}
                   </section>
 
-                  <section className="detail-block">
-                    <div className="detail-block-head">
+                  <section className="hotspot-compact-section">
+                    <button
+                      className="hotspot-compact-toggle"
+                      type="button"
+                      onClick={() => setStrategyPanelOpen((current) => !current)}
+                      aria-expanded={strategyPanelOpen}
+                    >
                       <div>
                         <h3>CAN SLIM 选股</h3>
-                        <span className="muted">默认先看热点内筛选，再切到全市场候选。</span>
+                        <span className="muted">{strategyPanelOpen ? "收起候选和交易计划草案" : "展开后查看候选和交易计划草案"}</span>
                       </div>
+                      <span className="pill">{screenCandidates.length || 0} 只</span>
+                    </button>
+                    {strategyPanelOpen ? (
+                      <div className="detail-block hotspot-compact-body">
+                      <div className="detail-block-head">
+                        <span className="muted">默认先看热点内筛选，再切到全市场候选。</span>
                       <div className="inline-actions">
                         <button
                           className={`button ${screenScope === "hotspot" ? "" : "ghost"}`}
@@ -526,16 +572,18 @@ export function HotspotsPageClient() {
                     ) : (
                       <p className="muted">当前没有可用的策略候选结果。</p>
                     )}
+                      </div>
+                    ) : null}
                   </section>
 
                   <section className="detail-block">
                     <div className="detail-block-head">
-                      <h3>板块催化消息</h3>
-                      <span className="muted">只保留标题、来源、摘要和影响级别</span>
+                      <h3>最新催化</h3>
+                      <span className="muted">只显示最相关 1 条</span>
                     </div>
                     {currentDetail.related_news.length ? (
                       <div className="detail-news-list">
-                        {currentDetail.related_news.slice(0, 6).map((item) => (
+                        {currentDetail.related_news.slice(0, 1).map((item) => (
                           <article className="detail-news-item" key={item.id}>
                             <div className="detail-news-meta">
                               <span>{item.source}</span>
@@ -543,33 +591,48 @@ export function HotspotsPageClient() {
                               <span>影响 {item.impact_level}</span>
                             </div>
                             <h4>{item.title}</h4>
-                            <p>{truncate(item.summary, 180)}</p>
+                            <p>{truncate(item.summary, 120)}</p>
                           </article>
                         ))}
+                        {currentDetail.related_news.length > 1 ? (
+                          <p className="muted">其余 {currentDetail.related_news.length - 1} 条催化消息已收起。</p>
+                        ) : null}
                       </div>
                     ) : (
                       <p className="muted">当前没有抓到与该板块直接相关的最新消息。</p>
                     )}
                   </section>
 
-                  <section className="detail-block">
-                    <div className="detail-block-head">
-                      <h3>板块热度快照</h3>
-                      <span className="muted">看板块热度是否持续升温</span>
-                    </div>
-                    {currentDetail.history.length ? (
-                      <div className="history-strip">
-                        {currentDetail.history.map((item) => (
-                          <div className="history-point" key={item.date}>
-                            <strong>{item.score.toFixed(0)}</strong>
-                            <span>{item.date}</span>
-                            <small>{item.count} 次触发</small>
-                          </div>
-                        ))}
+                  <section className="hotspot-compact-section">
+                    <button
+                      className="hotspot-compact-toggle"
+                      type="button"
+                      onClick={() => setHistoryPanelOpen((current) => !current)}
+                      aria-expanded={historyPanelOpen}
+                    >
+                      <div>
+                        <h3>板块热度快照</h3>
+                        <span className="muted">{historyPanelOpen ? "收起最近触发记录" : "展开看最近触发记录"}</span>
                       </div>
-                    ) : (
-                      <p className="muted">暂无可用历史快照。</p>
-                    )}
+                      <span className="pill">{currentDetail.history.length} 天</span>
+                    </button>
+                    {historyPanelOpen ? (
+                      <div className="detail-block hotspot-compact-body">
+                      {currentDetail.history.length ? (
+                        <div className="history-strip">
+                          {currentDetail.history.map((item) => (
+                            <div className="history-point" key={item.date}>
+                              <strong>{item.score.toFixed(0)}</strong>
+                              <span>{item.date}</span>
+                              <small>{item.count} 次触发</small>
+                            </div>
+                          ))}
+                        </div>
+                      ) : (
+                        <p className="muted">暂无可用历史快照。</p>
+                      )}
+                      </div>
+                    ) : null}
                   </section>
                 </div>
               </div>
@@ -620,12 +683,6 @@ function formatDateLabel(value: string) {
   }).format(date);
 }
 
-function trendLabel(direction: HotspotItem["trend_direction"]) {
-  if (direction === "up") return "升温";
-  if (direction === "down") return "降温";
-  return "平稳";
-}
-
 function summarizeLinkedStocks(item: HotspotItem) {
   const seenNames = new Set<string>();
   const visible = item.related_stocks.filter((stock) => {
@@ -642,6 +699,24 @@ function summarizeLinkedStocks(item: HotspotItem) {
   };
 }
 
+function normalizeHotspotBreakdown(item: HotspotItem) {
+  return {
+    trading_activity: item.heat_breakdown?.trading_activity ?? 0,
+    discussion_heat: item.heat_breakdown?.discussion_heat ?? 0,
+    news_heat: item.heat_breakdown?.news_heat ?? 0,
+    alert_count: item.heat_breakdown?.alert_count ?? 0,
+    rank_hits: item.heat_breakdown?.rank_hits ?? 0,
+    attention_level: item.heat_breakdown?.attention_level ?? "watch",
+    basis: item.heat_breakdown?.basis ?? [],
+  };
+}
+
+function attentionLabel(level: HotspotItem["heat_breakdown"]["attention_level"]) {
+  if (level === "focus") return "重点关注";
+  if (level === "caution") return "提高警惕";
+  return "先观察";
+}
+
 function candidateKey(candidate: StrategyCandidate) {
   return `${candidate.source_scope}-${candidate.source_topic ?? "market"}-${candidate.stock_code}`;
 }
diff --git a/app/news/news-page-client.tsx b/app/news/news-page-client.tsx
index 2745948..b4e7018 100644
--- a/app/news/news-page-client.tsx
+++ b/app/news/news-page-client.tsx
@@ -1,5 +1,6 @@
 "use client";
 
+import Link from "next/link";
 import { useCallback, useEffect, useMemo, useRef, useState } from "react";
 import { DemoAccessGate } from "@/components/demo-access-gate";
 import { useDemoAccess } from "@/components/demo-access-provider";
@@ -130,12 +131,11 @@ export function NewsPageClient() {
 
   const refreshPage = useCallback(() => {
     setManualRefreshing(true);
-    const cleanup = loadNews();
+    loadNews();
     if (unlocked) {
       loadAgent();
     }
     window.setTimeout(() => {
-      cleanup();
       setManualRefreshing(false);
     }, 400);
   }, [loadAgent, loadNews, unlocked]);
@@ -347,7 +347,7 @@ export function NewsPageClient() {
                 <div className="section-kicker">Lead Story</div>
                 <h2>全球重点新闻</h2>
               </div>
-              <span className="muted">{globalNews.length ? `${globalNews.length} 条` : "暂无数据"}</span>
+              <span className="muted">{globalNews.length ? `${globalNews.length} 条` : "新闻源为空"}</span>
             </div>
 
             {leadNews ? (
@@ -379,7 +379,19 @@ export function NewsPageClient() {
                 </div>
               </article>
             ) : (
-              <p className="muted">当前没有可展示的全球新闻。</p>
+              <div className="news-empty-state">
+                <p className="muted">
+                  当前外部新闻源没有返回可展示内容，可能是数据源限流、网络波动，或本地监控缓存暂时为空。
+                </p>
+                <div className="inline-actions">
+                  <button className="button ghost" type="button" onClick={loadNews}>
+                    重新加载
+                  </button>
+                  <Link href="/hotspots" className="button">
+                    先看热点板块
+                  </Link>
+                </div>
+              </div>
             )}
 
             {secondaryNews.length ? (
diff --git a/ashare/config.py b/ashare/config.py
index 3d4aa61..a4b24c4 100644
--- a/ashare/config.py
+++ b/ashare/config.py
@@ -99,6 +99,9 @@ class Config:
     web_search_enabled: bool = field(default_factory=lambda: _env_bool("WEB_SEARCH_ENABLED", True))
     web_search_timeout: int = field(default_factory=lambda: _env_int("WEB_SEARCH_TIMEOUT", 8))
 
+    # 美股数据源：Finnhub 作为 yfinance 的备选源（缺失时自动跳过，仅用 yfinance）
+    finnhub_api_key: Optional[str] = field(default_factory=lambda: _env_text("FINNHUB_API_KEY"))
+
     def __post_init__(self) -> None:
         self.stock_pool_path = Path(self.stock_pool_path)
         self.stock_topics_path = Path(self.stock_topics_path)
diff --git a/ashare/data.py b/ashare/data.py
index e24823b..4d8485f 100644
--- a/ashare/data.py
+++ b/ashare/data.py
@@ -7,7 +7,15 @@
 import pandas as pd
 from typing import Dict, List, Optional
 from ashare.logging import get_logger
-from ashare.stock_pool import get_exchange_label, is_valid_stock_code, normalize_stock_code
+from ashare.stock_pool import (
+    extract_symbol,
+    get_exchange_label,
+    is_us_stock,
+    is_valid_stock_code,
+    normalize_stock_code,
+)
+
+logger = get_logger(__name__)
 
 # 尝试导入数据源（pip 的 ashare 包可能与项目根目录 Ashare.py 冲突，用文件路径显式加载）
 as_api = None
@@ -35,7 +43,7 @@ def _load_ashare_module():
                     if hasattr(mod, "get_price"):
                         return mod
             except Exception as ex:
-                print(f"警告: 加载 {path} 失败: {ex}")
+                logger.warning("加载 %s 失败: %s", path, ex)
         return None
 
     as_api = _load_ashare_module()
@@ -46,12 +54,10 @@ def _load_ashare_module():
             as_api = None
 except Exception as e:
     as_api = None
-    print(f"警告: 无法加载Ashare数据源模块: {e}")
+    logger.warning("无法加载 Ashare 数据源模块: %s", e)
 
 if as_api is None:
-    print("警告: as_api 未就绪，将使用 akshare 作为日线数据源备选")
-
-logger = get_logger(__name__)
+    logger.warning("as_api 未就绪，将使用 akshare 作为日线数据源备选")
 
 
 class DataFetcher:
@@ -93,14 +99,18 @@ def fetch_stock_data(self, code: str, count: Optional[int] = None,
         try:
             logger.info(f"正在获取股票 {code} 的数据...")
             df = None
-            if as_api is not None and hasattr(as_api, "get_price"):
-                try:
-                    df = as_api.get_price(code, count=count, frequency=frequency)
-                except Exception as ex:
-                    logger.warning("Ashare.get_price failed for %s: %s", code, ex)
-                    df = None
-            if (df is None or (isinstance(df, pd.DataFrame) and df.empty)) and frequency == "1d":
-                df = self._fetch_stock_data_akshare(code, count)
+            if is_us_stock(code):
+                # 美股走独立链路：yfinance 主源，Finnhub 备选
+                df = self._fetch_stock_data_us(code, count, frequency)
+            else:
+                if as_api is not None and hasattr(as_api, "get_price"):
+                    try:
+                        df = as_api.get_price(code, count=count, frequency=frequency)
+                    except Exception as ex:
+                        logger.warning("Ashare.get_price failed for %s: %s", code, ex)
+                        df = None
+                if (df is None or (isinstance(df, pd.DataFrame) and df.empty)) and frequency == "1d":
+                    df = self._fetch_stock_data_akshare(code, count)
             
             # 检查数据是否有效
             if df is None or df.empty:
@@ -197,6 +207,145 @@ def get_cache_info(self) -> Dict[str, int]:
             'cache_keys': list(self.data_cache.keys())
         }
     
+    # ------------------------------------------------------------------
+    # 美股数据源
+    # ------------------------------------------------------------------
+    # yfinance 频率映射
+    _YF_INTERVAL = {
+        "1d": "1d", "1w": "1wk", "1M": "1mo",
+        "60m": "60m", "30m": "30m", "15m": "15m", "5m": "5m", "1m": "1m",
+    }
+    # Finnhub 频率映射（resolution）
+    _FINNHUB_RESOLUTION = {
+        "1d": "D", "1w": "W", "1M": "M",
+        "60m": "60", "30m": "30", "15m": "15", "5m": "5", "1m": "1",
+    }
+    # 估算需要回溯的自然日数（按频率把 count 转成日历跨度，留足非交易日冗余）
+    _LOOKBACK_DAYS_PER_BAR = {
+        "1d": 2, "1w": 9, "1M": 40,
+        "60m": 1, "30m": 1, "15m": 1, "5m": 1, "1m": 1,
+    }
+
+    @staticmethod
+    def _us_yf_symbol(code: str) -> str:
+        """US.BRK.B -> BRK-B（yfinance 用 ``-`` 表示类别股）。"""
+        return extract_symbol(code).replace(".", "-")
+
+    def _fetch_stock_data_us(self, code: str, count: int, frequency: str) -> Optional[pd.DataFrame]:
+        """美股行情：yfinance 主源，失败时降级到 Finnhub。"""
+        df = self._fetch_stock_data_yfinance(code, count, frequency)
+        if df is not None and not df.empty:
+            return df
+        logger.info("yfinance 未取到 %s 数据，尝试 Finnhub 备选源", code)
+        return self._fetch_stock_data_finnhub(code, count, frequency)
+
+    def _fetch_stock_data_yfinance(self, code: str, count: int, frequency: str) -> Optional[pd.DataFrame]:
+        try:
+            import yfinance as yf
+        except ImportError:
+            logger.warning("yfinance 未安装，无法获取美股数据")
+            return None
+
+        interval = self._YF_INTERVAL.get(frequency)
+        if interval is None:
+            logger.warning("yfinance 不支持的频率: %s", frequency)
+            return None
+
+        symbol = self._us_yf_symbol(code)
+        lookback = max(7, count * self._LOOKBACK_DAYS_PER_BAR.get(frequency, 2))
+        start = (pd.Timestamp.now() - pd.Timedelta(days=lookback)).strftime("%Y-%m-%d")
+        end = (pd.Timestamp.now() + pd.Timedelta(days=1)).strftime("%Y-%m-%d")
+        try:
+            raw = yf.download(
+                symbol, start=start, end=end, interval=interval,
+                auto_adjust=True, progress=False, threads=False,
+            )
+        except Exception as e:
+            logger.warning("yfinance download failed %s: %s", symbol, e)
+            return None
+        return self._normalize_us_dataframe(raw, count, source=f"yfinance({symbol})")
+
+    def _normalize_us_dataframe(self, raw, count: int, source: str) -> Optional[pd.DataFrame]:
+        if raw is None or raw.empty:
+            return None
+        df = raw.copy()
+        # yfinance 多 ticker 时返回 MultiIndex 列，单 ticker 也可能带一层，统一压平
+        if isinstance(df.columns, pd.MultiIndex):
+            df.columns = df.columns.get_level_values(0)
+        col_map = {"Open": "open", "Close": "close", "High": "high", "Low": "low", "Volume": "volume"}
+        rename = {k: v for k, v in col_map.items() if k in df.columns}
+        df = df.rename(columns=rename)
+        for c in ["open", "close", "high", "low", "volume"]:
+            if c not in df.columns:
+                logger.warning("%s 缺少必要列 %s", source, c)
+                return None
+        df = df[["open", "close", "high", "low", "volume"]].astype(float)
+        df.index = pd.to_datetime(df.index)
+        df.index.name = ""
+        df.columns.name = None
+        df = df.dropna().tail(count)
+        if df.empty:
+            return None
+        logger.info("%s 成功，共 %d 条", source, len(df))
+        return df
+
+    def _fetch_stock_data_finnhub(self, code: str, count: int, frequency: str) -> Optional[pd.DataFrame]:
+        try:
+            from ashare.config import Config
+        except ImportError:
+            Config = None
+        api_key = None
+        if Config is not None:
+            try:
+                api_key = Config().finnhub_api_key
+            except Exception:
+                api_key = None
+        if not api_key:
+            logger.info("未配置 FINNHUB_API_KEY，跳过 Finnhub 备选源")
+            return None
+
+        resolution = self._FINNHUB_RESOLUTION.get(frequency)
+        if resolution is None:
+            logger.warning("Finnhub 不支持的频率: %s", frequency)
+            return None
+
+        symbol = extract_symbol(code).replace(".", "-")
+        lookback = max(7, count * self._LOOKBACK_DAYS_PER_BAR.get(frequency, 2))
+        to_ts = int(pd.Timestamp.now().timestamp())
+        from_ts = int((pd.Timestamp.now() - pd.Timedelta(days=lookback)).timestamp())
+        try:
+            import requests
+            resp = requests.get(
+                "https://finnhub.io/api/v1/stock/candle",
+                params={"symbol": symbol, "resolution": resolution,
+                        "from": from_ts, "to": to_ts, "token": api_key},
+                timeout=10,
+            )
+            resp.raise_for_status()
+            payload = resp.json()
+        except Exception as e:
+            logger.warning("Finnhub request failed %s: %s", symbol, e)
+            return None
+
+        if not isinstance(payload, dict) or payload.get("s") != "ok":
+            logger.warning("Finnhub 返回无效数据 %s: status=%s", symbol, payload.get("s") if isinstance(payload, dict) else payload)
+            return None
+        try:
+            df = pd.DataFrame({
+                "open": payload["o"], "close": payload["c"],
+                "high": payload["h"], "low": payload["l"],
+                "volume": payload["v"],
+            }, index=pd.to_datetime(payload["t"], unit="s")).astype(float)
+        except (KeyError, ValueError, TypeError) as e:
+            logger.warning("Finnhub 数据解析失败 %s: %s", symbol, e)
+            return None
+        df.index.name = ""
+        df = df.dropna().tail(count)
+        if df.empty:
+            return None
+        logger.info("Finnhub(%s) 成功，共 %d 条", symbol, len(df))
+        return df
+
     def _fetch_stock_data_akshare(self, code: str, count: int) -> Optional[pd.DataFrame]:
         """
         akshare 日线备选：当 Ashare.get_price 不可用时使用。
diff --git a/ashare/llm.py b/ashare/llm.py
index 6d30967..d73d52d 100644
--- a/ashare/llm.py
+++ b/ashare/llm.py
@@ -1,4 +1,5 @@
 import json
+import logging
 import os
 from typing import Callable, Dict, Any, Optional
 
@@ -7,6 +8,8 @@
 from openai import OpenAI
 import re
 
+logger = logging.getLogger(__name__)
+
 
 def format_analysis_result(result: Dict[str, Any]) -> Dict[str, Any]:
     """
@@ -396,21 +399,21 @@ def request_analysis(self, df: pd.DataFrame, technical_indicators: pd.DataFrame)
         """
         try:
             # 准备数据
-            print("开始准备数据...")
+            logger.debug("开始准备数据...")
             data_str = _format_data_for_prompt(df, technical_indicators)
-            print(f"数据准备完成，数据长度: {len(data_str)}")
+            logger.debug(f"数据准备完成，数据长度: {len(data_str)}")
 
             # 构建消息
-            print("构建API请求消息...")
+            logger.debug("构建API请求消息...")
             messages = [
                 {"role": "system", "content": _create_system_prompt()},
                 {"role": "user", "content": f"请分析以下股票数据并给出专业的分析意见：\n{data_str}"}
             ]
-            print(f"消息构建完成，系统提示词长度: {len(messages[0]['content'])}")
-            print(f"用户消息长度: {len(messages[1]['content'])}")
+            logger.debug(f"消息构建完成，系统提示词长度: {len(messages[0]['content'])}")
+            logger.debug(f"用户消息长度: {len(messages[1]['content'])}")
 
             # 发送请求
-            print("开始发送API请求...")
+            logger.debug("开始发送API请求...")
             try:
                 response = self.client.chat.completions.create(
                     model=self.model,
@@ -418,85 +421,85 @@ def request_analysis(self, df: pd.DataFrame, technical_indicators: pd.DataFrame)
                     temperature=1.0,
                     stream=False
                 )
-                print("API请求发送成功")
+                logger.debug("API请求发送成功")
             except Exception as api_e:
                 # 检查是否是空响应导致的JSON解析错误
                 if str(api_e).startswith("Expecting value: line 1 column 1 (char 0)"):
-                    print("API返回空响应，服务器可能繁忙")
+                    logger.debug("API返回空响应，服务器可能繁忙")
                     raise APIBusyError("API服务器繁忙，返回空响应") from api_e
-                print(f"API请求发送失败: {str(api_e)}")
+                logger.debug(f"API请求发送失败: {str(api_e)}")
                 raise  # 重新抛出其他类型的异常
 
             # 记录原始响应以便调试
-            print("API 原始响应类型:", type(response))
-            print("API 原始响应内容:", response)
+            logger.debug("API 原始响应类型:", type(response))
+            logger.debug("API 原始响应内容:", response)
 
             # 检查响应内容
             if not response:
-                print("API返回空响应")
+                logger.debug("API返回空响应")
                 return format_analysis_result({})
 
             if not hasattr(response, 'choices'):
-                print(f"API响应缺少choices属性，响应结构: {dir(response)}")
+                logger.debug(f"API响应缺少choices属性，响应结构: {dir(response)}")
                 return format_analysis_result({})
 
             if not response.choices:
-                print("API响应的choices为空")
+                logger.debug("API响应的choices为空")
                 return format_analysis_result({})
 
             # 解析响应
             try:
                 analysis_text = response.choices[0].message.content
-                print("成功获取分析文本内容")
-                print("分析文本:", analysis_text)
+                logger.debug("成功获取分析文本内容")
+                logger.debug("分析文本:", analysis_text)
             except Exception as text_e:
-                print(f"获取分析文本失败: {str(text_e)}")
+                logger.debug(f"获取分析文本失败: {str(text_e)}")
                 raise
 
             # 将文本响应组织成结构化数据
-            print("开始解析分析文本...")
+            logger.debug("开始解析分析文本...")
             result = _parse_analysis_response(analysis_text)
-            print("分析文本解析完成")
+            logger.debug("分析文本解析完成")
             return result
 
         except APIBusyError as be:  # 处理API繁忙异常
-            print(f"=== API繁忙错误 ===")
-            print(f"错误详情: {str(be)}")
-            print(f"错误类型: {type(be)}")
+            logger.debug(f"=== API繁忙错误 ===")
+            logger.debug(f"错误详情: {str(be)}")
+            logger.debug(f"错误类型: {type(be)}")
             return format_analysis_result({})
         except json.JSONDecodeError as je:
-            print(f"=== JSON解析错误 ===")
-            print(f"错误详情: {str(je)}")
-            print(f"错误类型: {type(je)}")
-            print(f"错误位置: {je.pos}")
-            print(f"错误行列: 行 {je.lineno}, 列 {je.colno}")
-            print(f"错误的文档片段: {je.doc[:100] if je.doc else 'None'}")
+            logger.debug(f"=== JSON解析错误 ===")
+            logger.debug(f"错误详情: {str(je)}")
+            logger.debug(f"错误类型: {type(je)}")
+            logger.debug(f"错误位置: {je.pos}")
+            logger.debug(f"错误行列: 行 {je.lineno}, 列 {je.colno}")
+            logger.debug(f"错误的文档片段: {je.doc[:100] if je.doc else 'None'}")
             return format_analysis_result({})
         except openai.APITimeoutError as te:
-            print(f"=== API超时错误 ===")
-            print(f"错误详情: {str(te)}")
-            print(f"错误类型: {type(te)}")
+            logger.debug(f"=== API超时错误 ===")
+            logger.debug(f"错误详情: {str(te)}")
+            logger.debug(f"错误类型: {type(te)}")
             return format_analysis_result({})
         except openai.APIConnectionError as ce:
-            print(f"=== API连接错误 ===")
-            print(f"错误详情: {str(ce)}")
-            print(f"错误类型: {type(ce)}")
+            logger.debug(f"=== API连接错误 ===")
+            logger.debug(f"错误详情: {str(ce)}")
+            logger.debug(f"错误类型: {type(ce)}")
             return format_analysis_result({})
         except openai.APIError as ae:
-            print(f"=== API错误 ===")
-            print(f"错误详情: {str(ae)}")
-            print(f"错误类型: {type(ae)}")
+            logger.debug(f"=== API错误 ===")
+            logger.debug(f"错误详情: {str(ae)}")
+            logger.debug(f"错误类型: {type(ae)}")
             return format_analysis_result({})
         except openai.RateLimitError as re:
-            print(f"=== API频率限制错误 ===")
-            print(f"错误详情: {str(re)}")
-            print(f"错误类型: {type(re)}")
+            logger.debug(f"=== API频率限制错误 ===")
+            logger.debug(f"错误详情: {str(re)}")
+            logger.debug(f"错误类型: {type(re)}")
             return format_analysis_result({})
         except Exception as e:
-            print(f"=== 未预期的错误 ===")
-            print(f"错误详情: {str(e)}")
-            print(f"错误类型: {type(e)}")
-            print(f"错误追踪:")
+            logger.debug(f"=== 未预期的错误 ===")
+            logger.debug(f"错误详情: {str(e)}")
+            logger.debug(f"错误类型: {type(e)}")
+            logger.debug(f"错误追踪:")
             import traceback
             traceback.print_exc()
             return format_analysis_result({})
@@ -512,7 +515,7 @@ def generate_pool_analysis(self, pool_summary: list) -> Optional[str]:
             AI分析文本，失败时返回None
         """
         try:
-            print("开始生成股票池AI分析...")
+            logger.debug("开始生成股票池AI分析...")
             
             # 构建股票池分析的提示词
             pool_info = "股票池综合分析：\n\n"
@@ -553,7 +556,7 @@ def generate_pool_analysis(self, pool_summary: list) -> Optional[str]:
                 {"role": "user", "content": pool_info}
             ]
             
-            print("发送股票池分析请求...")
+            logger.debug("发送股票池分析请求...")
             response = self.client.chat.completions.create(
                 model=self.model,
                 messages=messages,
@@ -563,14 +566,14 @@ def generate_pool_analysis(self, pool_summary: list) -> Optional[str]:
             
             if response.choices and response.choices[0].message:
                 analysis_text = response.choices[0].message.content.strip()
-                print("股票池AI分析生成成功")
+                logger.debug("股票池AI分析生成成功")
                 return analysis_text
             else:
-                print("股票池AI分析响应为空")
+                logger.debug("股票池AI分析响应为空")
                 return None
                 
         except Exception as e:
-            print(f"生成股票池AI分析失败: {str(e)}")
+            logger.debug(f"生成股票池AI分析失败: {str(e)}")
             return None
 
     def generate_fundamental_analysis(self, fundamental_data: list, analysis_params: Dict[str, Any]) -> Optional[str]:
@@ -585,7 +588,7 @@ def generate_fundamental_analysis(self, fundamental_data: list, analysis_params:
             基本面分析文本，失败时返回None
         """
         try:
-            print(f"开始生成基本面分析: {analysis_params['type']}")
+            logger.debug(f"开始生成基本面分析: {analysis_params['type']}")
             
             # 构建基本面分析的数据信息
             fundamental_info = f"基本面分析 - {analysis_params['type']}：\n\n"
@@ -646,7 +649,7 @@ def generate_fundamental_analysis(self, fundamental_data: list, analysis_params:
                 {"role": "user", "content": fundamental_info}
             ]
             
-            print("发送基本面分析请求...")
+            logger.debug("发送基本面分析请求...")
             response = self.client.chat.completions.create(
                 model=self.model,
                 messages=messages,
@@ -656,14 +659,14 @@ def generate_fundamental_analysis(self, fundamental_data: list, analysis_params:
             
             if response.choices and response.choices[0].message:
                 analysis_text = response.choices[0].message.content.strip()
-                print("基本面分析生成成功")
+                logger.debug("基本面分析生成成功")
                 return analysis_text
             else:
-                print("基本面分析响应为空")
+                logger.debug("基本面分析响应为空")
                 return None
                 
         except Exception as e:
-            print(f"生成基本面分析失败: {str(e)}")
+            logger.debug(f"生成基本面分析失败: {str(e)}")
             return None
 
     def generate_sector_rotation_analysis(self, sector_data: list) -> Optional[str]:
@@ -677,7 +680,7 @@ def generate_sector_rotation_analysis(self, sector_data: list) -> Optional[str]:
             板块轮动分析文本，失败时返回None
         """
         try:
-            print("开始生成板块轮动分析...")
+            logger.debug("开始生成板块轮动分析...")
             
             # 构建板块轮动分析数据
             sector_info = "板块轮动分析：\n\n"
@@ -720,7 +723,7 @@ def generate_sector_rotation_analysis(self, sector_data: list) -> Optional[str]:
                 {"role": "user", "content": sector_info}
             ]
             
-            print("发送板块轮动分析请求...")
+            logger.debug("发送板块轮动分析请求...")
             response = self.client.chat.completions.create(
                 model=self.model,
                 messages=messages,
@@ -730,14 +733,14 @@ def generate_sector_rotation_analysis(self, sector_data: list) -> Optional[str]:
             
             if response.choices and response.choices[0].message:
                 analysis_text = response.choices[0].message.content.strip()
-                print("板块轮动分析生成成功")
+                logger.debug("板块轮动分析生成成功")
                 return analysis_text
             else:
-                print("板块轮动分析响应为空")
+                logger.debug("板块轮动分析响应为空")
                 return None
                 
         except Exception as e:
-            print(f"生成板块轮动分析失败: {str(e)}")
+            logger.debug(f"生成板块轮动分析失败: {str(e)}")
             return None
 
     def generate_trend_strength_analysis(self, trend_data: list) -> Optional[str]:
@@ -751,7 +754,7 @@ def generate_trend_strength_analysis(self, trend_data: list) -> Optional[str]:
             趋势强度分析文本，失败时返回None
         """
         try:
-            print("开始生成趋势强度分析...")
+            logger.debug("开始生成趋势强度分析...")
             
             # 构建趋势强度分析数据
             trend_info = "趋势强度分析：\n\n"
@@ -798,7 +801,7 @@ def generate_trend_strength_analysis(self, trend_data: list) -> Optional[str]:
                 {"role": "user", "content": trend_info}
             ]
             
-            print("发送趋势强度分析请求...")
+            logger.debug("发送趋势强度分析请求...")
             response = self.client.chat.completions.create(
                 model=self.model,
                 messages=messages,
@@ -808,14 +811,14 @@ def generate_trend_strength_analysis(self, trend_data: list) -> Optional[str]:
             
             if response.choices and response.choices[0].message:
                 analysis_text = response.choices[0].message.content.strip()
-                print("趋势强度分析生成成功")
+                logger.debug("趋势强度分析生成成功")
                 return analysis_text
             else:
-                print("趋势强度分析响应为空")
+                logger.debug("趋势强度分析响应为空")
                 return None
                 
         except Exception as e:
-            print(f"生成趋势强度分析失败: {str(e)}")
+            logger.debug(f"生成趋势强度分析失败: {str(e)}")
             return None
 
     def generate_single_stock_analysis(
@@ -837,7 +840,7 @@ def report(progress: int, message: str) -> None:
                 if progress_callback:
                     progress_callback(progress, message)
 
-            print(f"开始生成 {detailed_data['name']} 的单股分析...")
+            logger.debug(f"开始生成 {detailed_data['name']} 的单股分析...")
             report(56, "正在整理提示词与上下文，准备发送模型请求")
             
             # 构建单股分析数据
@@ -1061,7 +1064,7 @@ def report(progress: int, message: str) -> None:
                 {"role": "user", "content": stock_info}
             ]
             
-            print("发送单股分析请求...")
+            logger.debug("发送单股分析请求...")
             report(72, "模型请求已发送，正在等待 AI 返回完整报告")
             # 根据分析深度设置不同的max_tokens
             if detailed_data['analysis_depth'] == "快速分析":
@@ -1081,42 +1084,42 @@ def report(progress: int, message: str) -> None:
             
             if response.choices and response.choices[0].message:
                 analysis_text = response.choices[0].message.content.strip()
-                print(f"{detailed_data['name']} 单股分析生成成功")
+                logger.debug(f"{detailed_data['name']} 单股分析生成成功")
                 report(92, "AI 已返回内容，正在整理结果")
                 return analysis_text
             else:
-                print("单股分析响应为空")
+                logger.debug("单股分析响应为空")
                 report(92, "AI 请求已完成，但响应内容为空")
                 return None
                 
         except openai.APIConnectionError as ce:
-            print(f"=== API连接错误 ===")
-            print(f"生成单股分析失败: {str(ce)}")
-            print(f"错误类型: {type(ce)}")
-            print("可能的原因：")
-            print("1. 网络连接问题，请检查网络连接")
-            print("2. API服务不可用，请稍后重试")
-            print("3. 代理设置问题，请检查代理配置")
+            logger.debug(f"=== API连接错误 ===")
+            logger.debug(f"生成单股分析失败: {str(ce)}")
+            logger.debug(f"错误类型: {type(ce)}")
+            logger.debug("可能的原因：")
+            logger.debug("1. 网络连接问题，请检查网络连接")
+            logger.debug("2. API服务不可用，请稍后重试")
+            logger.debug("3. 代理设置问题，请检查代理配置")
             return None
         except openai.APITimeoutError as te:
-            print(f"=== API超时错误 ===")
-            print(f"生成单股分析失败: {str(te)}")
-            print("请求超时，请稍后重试")
+            logger.debug(f"=== API超时错误 ===")
+            logger.debug(f"生成单股分析失败: {str(te)}")
+            logger.debug("请求超时，请稍后重试")
             return None
         except openai.RateLimitError as re:
-            print(f"=== API频率限制错误 ===")
-            print(f"生成单股分析失败: {str(re)}")
-            print("API调用频率过高，请稍后重试")
+            logger.debug(f"=== API频率限制错误 ===")
+            logger.debug(f"生成单股分析失败: {str(re)}")
+            logger.debug("API调用频率过高，请稍后重试")
             return None
         except openai.APIError as ae:
-            print(f"=== API错误 ===")
-            print(f"生成单股分析失败: {str(ae)}")
-            print(f"错误类型: {type(ae)}")
+            logger.debug(f"=== API错误 ===")
+            logger.debug(f"生成单股分析失败: {str(ae)}")
+            logger.debug(f"错误类型: {type(ae)}")
             return None
         except Exception as e:
-            print(f"=== 未预期的错误 ===")
-            print(f"生成单股分析失败: {str(e)}")
-            print(f"错误类型: {type(e)}")
+            logger.debug(f"=== 未预期的错误 ===")
+            logger.debug(f"生成单股分析失败: {str(e)}")
+            logger.debug(f"错误类型: {type(e)}")
             import traceback
             traceback.print_exc()
             return None
@@ -1133,7 +1136,7 @@ def generate_market_insights(self, market_data: Dict[str, Any]) -> Optional[str]
         """
         try:
             insight_type = market_data['insight_type']
-            print(f"开始生成市场洞察: {insight_type}")
+            logger.debug(f"开始生成市场洞察: {insight_type}")
             
             # 构建市场洞察数据
             market_info = f"市场洞察分析 - {insight_type}：\n\n"
@@ -1214,7 +1217,7 @@ def generate_market_insights(self, market_data: Dict[str, Any]) -> Optional[str]
                 {"role": "user", "content": market_info}
             ]
             
-            print("发送市场洞察分析请求...")
+            logger.debug("发送市场洞察分析请求...")
             response = self.client.chat.completions.create(
                 model=self.model,
                 messages=messages,
@@ -1224,12 +1227,12 @@ def generate_market_insights(self, market_data: Dict[str, Any]) -> Optional[str]
             
             if response.choices and response.choices[0].message:
                 analysis_text = response.choices[0].message.content.strip()
-                print("市场洞察分析生成成功")
+                logger.debug("市场洞察分析生成成功")
                 return analysis_text
             else:
-                print("市场洞察分析响应为空")
+                logger.debug("市场洞察分析响应为空")
                 return None
                 
         except Exception as e:
-            print(f"生成市场洞察失败: {str(e)}")
+            logger.debug(f"生成市场洞察失败: {str(e)}")
             return None
diff --git a/ashare/monitor.py b/ashare/monitor.py
index d50ff77..55b4ea7 100644
--- a/ashare/monitor.py
+++ b/ashare/monitor.py
@@ -22,7 +22,9 @@
     extract_symbol,
     get_monitor_support_level,
     infer_market,
+    is_a_share_individual_stock,
     is_hk_stock,
+    is_us_stock,
     normalize_stock_code,
 )
 
@@ -335,7 +337,27 @@ def fetch_stock_news(self, stock_code: str, stock_name: str) -> List[Dict[str, A
         results: List[Dict[str, Any]] = []
         topic_keywords = self._get_topic_keywords(stock_name, stock_code)
 
-        if not is_hk_stock(stock_code):
+        if is_us_stock(stock_code):
+            # 美股个股新闻走 Finnhub（A股 的巨潮/东财接口只接受数字代码）
+            primary_sources = [
+                (
+                    "Finnhub公司新闻",
+                    lambda: self._fetch_finnhub_company_news(symbol),
+                    lambda df: self._normalize_news_dataframe(
+                        df,
+                        stock_name,
+                        stock_code,
+                        "Finnhub",
+                        topic_keywords,
+                        default_relation_type="direct",
+                    ),
+                ),
+            ]
+            for source_name, fetch_fn, transform_fn in primary_sources:
+                results.extend(
+                    self._collect_source_items(source_name, stock_name, stock_code, fetch_fn, transform_fn)
+                )
+        elif not is_hk_stock(stock_code):
             primary_sources = [
                 (
                     "巨潮资讯公告",
@@ -368,32 +390,35 @@ def fetch_stock_news(self, stock_code: str, stock_name: str) -> List[Dict[str, A
                 )
 
         # Market-wide feeds are fallback only; keep topic/name filtering.
-        extra_sources = [
-            (
-                "财联社",
-                lambda: ak.stock_info_global_cls(symbol="全部"),
-                lambda df: self._filter_market_news(df, stock_name, stock_code, topic_keywords, "财联社"),
-            ),
-            (
-                "财联社快讯",
-                lambda: self._fetch_ak_news("stock_news_main_cx"),
-                lambda df: self._filter_market_news(df, stock_name, stock_code, topic_keywords, "财联社快讯"),
-            ),
-            (
-                "新浪财经",
-                lambda: self._fetch_ak_news("stock_info_global_sina"),
-                lambda df: self._filter_market_news(df, stock_name, stock_code, topic_keywords, "新浪财经"),
-            ),
-            (
-                "同花顺",
-                lambda: self._fetch_ak_news("stock_info_global_ths"),
-                lambda df: self._filter_market_news(df, stock_name, stock_code, topic_keywords, "同花顺"),
-            ),
-        ]
-        for source_name, fetch_fn, transform_fn in extra_sources:
-            results.extend(
-                self._collect_source_items(source_name, stock_name, stock_code, fetch_fn, transform_fn)
-            )
+        # 美股不走中文全市场源：这些 feed 以 A股 资讯为主，按个股名（如“苹果”
+        # 这类常用词）过滤会混入大量无关中文新闻，美股已由 Finnhub 覆盖。
+        if not is_us_stock(stock_code):
+            extra_sources = [
+                (
+                    "财联社",
+                    lambda: ak.stock_info_global_cls(symbol="全部"),
+                    lambda df: self._filter_market_news(df, stock_name, stock_code, topic_keywords, "财联社"),
+                ),
+                (
+                    "财联社快讯",
+                    lambda: self._fetch_ak_news("stock_news_main_cx"),
+                    lambda df: self._filter_market_news(df, stock_name, stock_code, topic_keywords, "财联社快讯"),
+                ),
+                (
+                    "新浪财经",
+                    lambda: self._fetch_ak_news("stock_info_global_sina"),
+                    lambda df: self._filter_market_news(df, stock_name, stock_code, topic_keywords, "新浪财经"),
+                ),
+                (
+                    "同花顺",
+                    lambda: self._fetch_ak_news("stock_info_global_ths"),
+                    lambda df: self._filter_market_news(df, stock_name, stock_code, topic_keywords, "同花顺"),
+                ),
+            ]
+            for source_name, fetch_fn, transform_fn in extra_sources:
+                results.extend(
+                    self._collect_source_items(source_name, stock_name, stock_code, fetch_fn, transform_fn)
+                )
 
         # Dedupe by (title, stock_code) keeping first occurrence
         seen_keys: set = set()
@@ -500,6 +525,53 @@ def _fetch_cninfo_notices(self, symbol: str) -> Optional[pd.DataFrame]:
             end_date=end_date,
         )
 
+    def _fetch_finnhub_company_news(self, symbol: str) -> Optional[pd.DataFrame]:
+        """美股个股新闻：Finnhub company-news。未配置 key 时返回 None。"""
+        api_key = getattr(self.config, "finnhub_api_key", None) if self.config else None
+        if not api_key:
+            return None
+        # Finnhub 用 ``-`` 表示类别股（BRK.B -> BRK-B）
+        finnhub_symbol = symbol.replace(".", "-")
+        end_date = datetime.now().strftime("%Y-%m-%d")
+        start_date = (datetime.now() - timedelta(days=self.NOTICE_LOOKBACK_DAYS)).strftime("%Y-%m-%d")
+        try:
+            import requests
+
+            resp = requests.get(
+                "https://finnhub.io/api/v1/company-news",
+                params={"symbol": finnhub_symbol, "from": start_date, "to": end_date, "token": api_key},
+                timeout=10,
+            )
+            resp.raise_for_status()
+            payload = resp.json()
+        except Exception as exc:
+            logger.warning("Finnhub 公司新闻获取失败 %s: %s", finnhub_symbol, exc)
+            return None
+        if not isinstance(payload, list) or not payload:
+            return None
+        rows = []
+        for item in payload:
+            if not isinstance(item, dict):
+                continue
+            ts = item.get("datetime")
+            occurred_at = (
+                datetime.fromtimestamp(ts, tz=timezone.utc).strftime("%Y-%m-%d %H:%M:%S")
+                if isinstance(ts, (int, float)) and ts > 0
+                else ""
+            )
+            rows.append(
+                {
+                    "title": normalize_text(item.get("headline")),
+                    "content": normalize_text(item.get("summary")),
+                    "source": normalize_text(item.get("source")),
+                    "datetime": occurred_at,
+                    "url": normalize_text(item.get("url")),
+                }
+            )
+        if not rows:
+            return None
+        return pd.DataFrame(rows)
+
     def _collect_source_items(
         self,
         source_name: str,
@@ -958,7 +1030,8 @@ def run_cycle(self) -> Dict[str, Any]:
                     pushed_count += pushed
                     generated_count += generated
 
-                if self.config.fund_flow_tracking_enabled and not is_hk_stock(stock_code):
+                # 资金流/龙虎榜是 A股 特有数据，仅对 A股 个股运行（美股/港股跳过）
+                if self.config.fund_flow_tracking_enabled and is_a_share_individual_stock(stock_code):
                     pushed, generated = self._handle_fund_flow(stock_name, stock_code, ranking_map)
                     pushed_count += pushed
                     generated_count += generated
diff --git a/ashare/search.py b/ashare/search.py
index 07e2dc7..c119b48 100644
--- a/ashare/search.py
+++ b/ashare/search.py
@@ -74,6 +74,18 @@ def _load_stock_db(self) -> Dict[str, Dict[str, str]]:
                 },
             )
         return stock_db
+
+    def _merge_latest_stock_pool(self) -> None:
+        """Keep runtime search data in sync with the editable local stock pool."""
+        for name, code in load_stock_pool().items():
+            self.base_stock_db.setdefault(
+                name,
+                {
+                    "code": code,
+                    "market": get_market_label(code),
+                    "category": "",
+                },
+            )
     
     def search_stocks(self, query: str, max_results: int = 10) -> List[Dict]:
         """
@@ -89,12 +101,14 @@ def search_stocks(self, query: str, max_results: int = 10) -> List[Dict]:
         query = query.strip()
         if not query:
             return []
+
+        self._merge_latest_stock_pool()
         
         # 检查缓存
         cache_key = f"{query}_{max_results}"
         if cache_key in self.search_cache:
             cache_time, cache_results = self.search_cache[cache_key]
-            if time.time() - cache_time < self.cache_expiry:
+            if cache_results and time.time() - cache_time < self.cache_expiry:
                 return cache_results
         
         results = []
@@ -146,7 +160,8 @@ def search_stocks(self, query: str, max_results: int = 10) -> List[Dict]:
         final_results = sorted_results[:max_results]
         
         # 缓存结果
-        self.search_cache[cache_key] = (time.time(), final_results)
+        if final_results:
+            self.search_cache[cache_key] = (time.time(), final_results)
         
         return final_results
     
diff --git a/ashare/stock_pool.py b/ashare/stock_pool.py
index b1a3f8e..9742267 100644
--- a/ashare/stock_pool.py
+++ b/ashare/stock_pool.py
@@ -89,6 +89,55 @@
     "华润三九": {"code": "sz000999", "market": "A股-深圳", "category": "医药"},
     "丽珠集团": {"code": "sz000513", "market": "A股-深圳", "category": "医药"},
     "长春高新": {"code": "sz000661", "market": "A股-深圳", "category": "医药"},
+    # 美股龙头（含中文名映射，方便中文搜索）
+    "苹果": {"code": "US.AAPL", "market": "美股", "category": "科技"},
+    "微软": {"code": "US.MSFT", "market": "美股", "category": "科技"},
+    "英伟达": {"code": "US.NVDA", "market": "美股", "category": "半导体"},
+    "谷歌": {"code": "US.GOOGL", "market": "美股", "category": "科技"},
+    "谷歌C": {"code": "US.GOOG", "market": "美股", "category": "科技"},
+    "亚马逊": {"code": "US.AMZN", "market": "美股", "category": "电商"},
+    "Meta": {"code": "US.META", "market": "美股", "category": "科技"},
+    "脸书": {"code": "US.META", "market": "美股", "category": "科技"},
+    "特斯拉": {"code": "US.TSLA", "market": "美股", "category": "新能源车"},
+    "博通": {"code": "US.AVGO", "market": "美股", "category": "半导体"},
+    "台积电": {"code": "US.TSM", "market": "美股", "category": "半导体"},
+    "AMD": {"code": "US.AMD", "market": "美股", "category": "半导体"},
+    "英特尔": {"code": "US.INTC", "market": "美股", "category": "半导体"},
+    "高通": {"code": "US.QCOM", "market": "美股", "category": "半导体"},
+    "美光": {"code": "US.MU", "market": "美股", "category": "半导体"},
+    "奈飞": {"code": "US.NFLX", "market": "美股", "category": "科技"},
+    "甲骨文": {"code": "US.ORCL", "market": "美股", "category": "科技"},
+    "Adobe": {"code": "US.ADBE", "market": "美股", "category": "科技"},
+    "Salesforce": {"code": "US.CRM", "market": "美股", "category": "科技"},
+    "伯克希尔": {"code": "US.BRK.B", "market": "美股", "category": "金融"},
+    "摩根大通": {"code": "US.JPM", "market": "美股", "category": "金融"},
+    "visa": {"code": "US.V", "market": "美股", "category": "金融"},
+    "万事达": {"code": "US.MA", "market": "美股", "category": "金融"},
+    "美国银行": {"code": "US.BAC", "market": "美股", "category": "金融"},
+    "强生": {"code": "US.JNJ", "market": "美股", "category": "医药"},
+    "礼来": {"code": "US.LLY", "market": "美股", "category": "医药"},
+    "联合健康": {"code": "US.UNH", "market": "美股", "category": "医疗"},
+    "辉瑞": {"code": "US.PFE", "market": "美股", "category": "医药"},
+    "可口可乐": {"code": "US.KO", "market": "美股", "category": "消费"},
+    "百事": {"code": "US.PEP", "market": "美股", "category": "消费"},
+    "麦当劳": {"code": "US.MCD", "market": "美股", "category": "消费"},
+    "星巴克": {"code": "US.SBUX", "market": "美股", "category": "消费"},
+    "耐克": {"code": "US.NKE", "market": "美股", "category": "消费"},
+    "沃尔玛": {"code": "US.WMT", "market": "美股", "category": "零售"},
+    "好市多": {"code": "US.COST", "market": "美股", "category": "零售"},
+    "迪士尼": {"code": "US.DIS", "market": "美股", "category": "传媒"},
+    "波音": {"code": "US.BA", "market": "美股", "category": "工业"},
+    "埃克森美孚": {"code": "US.XOM", "market": "美股", "category": "能源"},
+    "雪佛龙": {"code": "US.CVX", "market": "美股", "category": "能源"},
+    "Palantir": {"code": "US.PLTR", "market": "美股", "category": "科技"},
+    "超微电脑": {"code": "US.SMCI", "market": "美股", "category": "科技"},
+    "阿斯麦": {"code": "US.ASML", "market": "美股", "category": "半导体"},
+    "Coinbase": {"code": "US.COIN", "market": "美股", "category": "金融科技"},
+    "Uber": {"code": "US.UBER", "market": "美股", "category": "科技"},
+    # 美股主要指数 ETF / 指数
+    "标普500": {"code": "US.SPY", "market": "美股", "category": "指数"},
+    "纳斯达克100": {"code": "US.QQQ", "market": "美股", "category": "指数"},
+    "道琼斯": {"code": "US.DIA", "market": "美股", "category": "指数"},
 }
 
 
@@ -179,8 +228,20 @@ def get_base_stock_catalog() -> Dict[str, Dict[str, str]]:
     }
 
 
+US_TICKER_PATTERN = r"[A-Z]{1,5}(\.[A-Z]{1,2})?"
+
+
 def normalize_stock_code(code: str) -> str:
-    """标准化股票代码格式。"""
+    """标准化股票代码格式。
+
+    支持三类市场：
+      - A股：``sh######`` / ``sz######``
+      - 港股：``#####.HK``
+      - 美股：``US.TICKER``（如 ``US.AAPL``、``US.BRK.B``）
+
+    美股统一加 ``US.`` 前缀，以便和港股 ``.HK``、A股数字代码区分，避免
+    诸如 ``BRK.B`` 这类含点的代码被误判成港股。
+    """
     raw = code.strip()
     if not raw:
         return raw
@@ -200,6 +261,13 @@ def normalize_stock_code(code: str) -> str:
         symbol = upper.split(".")[0].zfill(5)
         return f"{symbol}.HK"
 
+    # 美股：显式 ``US.`` 前缀
+    if upper.startswith("US.") and re.fullmatch(US_TICKER_PATTERN, upper[3:]):
+        return f"US.{upper[3:]}"
+    # 美股：裸 ticker（纯字母，区别于全数字的 A股/港股代码）
+    if re.fullmatch(US_TICKER_PATTERN, upper):
+        return f"US.{upper}"
+
     return raw
 
 
@@ -209,6 +277,8 @@ def extract_symbol(code: str) -> str:
         return normalized[2:]
     if normalized.upper().endswith(".HK"):
         return normalized[:-3]
+    if normalized.upper().startswith("US."):
+        return normalized[3:]
     return normalized
 
 
@@ -222,6 +292,8 @@ def infer_market(code: str) -> str:
         return "sz"
     if upper.endswith(".HK"):
         return "hk"
+    if upper.startswith("US."):
+        return "us"
     return "unknown"
 
 
@@ -232,6 +304,7 @@ def is_valid_stock_code(code: str) -> bool:
     return bool(
         re.fullmatch(r"(sh|sz)\d{6}", lower)
         or re.fullmatch(r"\d{5}\.HK", upper)
+        or re.fullmatch(rf"US\.{US_TICKER_PATTERN}", upper)
     )
 
 
@@ -253,6 +326,8 @@ def get_market_label(code: str) -> str:
         return "A股-深圳"
     if upper.endswith(".HK"):
         return "港股"
+    if upper.startswith("US."):
+        return "美股"
     return "未知市场"
 
 
@@ -264,6 +339,8 @@ def get_exchange_label(code: str) -> str:
         return "深交所"
     if market == "hk":
         return "港交所"
+    if market == "us":
+        return "美股"
     return "未知"
 
 
@@ -284,10 +361,21 @@ def is_hk_stock(code: str) -> bool:
     return bool(re.fullmatch(r"\d{5}\.HK", normalized))
 
 
+def is_us_stock(code: str) -> bool:
+    normalized = normalize_stock_code(code).upper()
+    return bool(re.fullmatch(rf"US\.{US_TICKER_PATTERN}", normalized))
+
+
 def get_monitor_support_level(code: str) -> str:
-    """返回监控支持级别: full, partial, unsupported。"""
+    """返回监控支持级别: full, partial, unsupported。
+
+    A股个股支持完整监控（含资金流/龙虎榜等 A股 特有数据）；美股支持
+    完整的行情与技术面监控，但 A股 特有数据会自动跳过；港股为部分支持。
+    """
     if is_a_share_individual_stock(code):
         return "full"
+    if is_us_stock(code):
+        return "full"
     if is_hk_stock(code):
         return "partial"
     return "unsupported"
diff --git a/components/app-shell.tsx b/components/app-shell.tsx
index cb73e38..22c8429 100644
--- a/components/app-shell.tsx
+++ b/components/app-shell.tsx
@@ -174,13 +174,36 @@ export function AppShell({ children }: { children: React.ReactNode }) {
 }
 
 const NAV_ITEMS = [
-  { href: "/work", label: "工作台", labelEn: "Workbench" },
-  { href: "/stocks", label: "单股分析", labelEn: "Stocks" },
-  { href: "/charts", label: "K 线图", labelEn: "Charts" },
-  { href: "/portfolio", label: "持仓页", labelEn: "Portfolio" },
-  { href: "/news", label: "消息页", labelEn: "News" },
-  { href: "/hotspots", label: "热点页", labelEn: "Hotspots" },
-  { href: "/settings", label: "设置", labelEn: "Settings" },
+  {
+    href: "/work",
+    label: "工作台",
+    labelEn: "Workbench",
+  },
+  {
+    href: "/stocks",
+    label: "单股分析",
+    labelEn: "Stocks",
+  },
+  {
+    href: "/portfolio",
+    label: "持仓页",
+    labelEn: "Portfolio",
+  },
+  {
+    href: "/news",
+    label: "消息页",
+    labelEn: "News",
+  },
+  {
+    href: "/hotspots",
+    label: "热点页",
+    labelEn: "Hotspots",
+  },
+  {
+    href: "/settings",
+    label: "设置",
+    labelEn: "Settings",
+  },
 ] as const;
 
 export function Nav() {
diff --git a/components/portfolio-shell.tsx b/components/portfolio-shell.tsx
index dc82770..47ca3df 100644
--- a/components/portfolio-shell.tsx
+++ b/components/portfolio-shell.tsx
@@ -12,6 +12,8 @@ import {
   getClientErrorMessage,
   getMarketRegime,
   getStrategyHoldingsAnalysis,
+  listStrategyHoldings,
+  refreshStrategyHoldings,
   searchStocks,
   updateStrategyHolding,
 } from "@/lib/api";
@@ -100,6 +102,8 @@ const EMPTY_STRATEGY_FORM: StrategyFormState = {
   status: "planned",
 };
 
+const STRATEGY_ANALYSIS_CACHE_KEY = "ashare.strategyAnalysis.v1";
+
 const HOLDING_STATUS_GROUPS: Array<{
   key: StrategyFormState["status"];
   title: string;
@@ -126,13 +130,16 @@ export function PortfolioShell({
   const [marketRegimeLoading, setMarketRegimeLoading] = useState(initialMarketRegime === null);
   const [marketRegimeError, setMarketRegimeError] = useState<string | null>(null);
   const [strategyForm, setStrategyForm] = useState<StrategyFormState>(EMPTY_STRATEGY_FORM);
+  const [strategyFormOpen, setStrategyFormOpen] = useState(Boolean(initialPrefill?.stock_code || initialPrefill?.stock_name));
   const [message, setMessage] = useState("可直接录入策略持股，随后刷新查看最新分析。");
   const [searchQuery, setSearchQuery] = useState("");
   const [searchResults, setSearchResults] = useState<StockSearchResult[]>([]);
   const [pendingAction, setPendingAction] = useState<PendingAction | null>(null);
   const [isAnalysisRefreshing, setIsAnalysisRefreshing] = useState(false);
   const strategyPriceInputRef = useRef<HTMLInputElement | null>(null);
+  const strategyAnalysisRef = useRef(normalizeStrategyAnalysis(initialStrategyAnalysis));
   const initialLoadInFlightRef = useRef(false);
+  const initialLoadCompletedRef = useRef(false);
   const localMutationVersionRef = useRef(0);
   const [focusTarget, setFocusTarget] = useState<"strategy_price" | null>(null);
   const [lastSubmitted, setLastSubmitted] = useState<{ stock_code: string; stock_name: string } | null>(null);
@@ -140,8 +147,21 @@ export function PortfolioShell({
   const isBusy = pendingAction !== null;
   const safeStrategyAnalysis = normalizeStrategyAnalysis(strategyAnalysis);
   const strategyRefreshInProgress = pendingAction?.type === "refresh" || isAnalysisRefreshing;
+  const scoredStrategyHoldingCount = safeStrategyAnalysis.holdings.filter(hasStrategyScore).length;
+
+  function applyStrategyAnalysis(analysis: StrategyHoldingAnalysisResponse) {
+    const normalized = normalizeStrategyAnalysis(analysis);
+    const merged = mergeLatestAnalysisWithExisting(normalized, strategyAnalysisRef.current);
+    strategyAnalysisRef.current = merged;
+    setStrategyAnalysis(merged);
+    setStrategyHoldings(toHoldingList(merged));
+    if (merged.holdings.some(hasStrategyScore)) {
+      cacheStrategyAnalysis(merged);
+    }
+  }
 
   function fillStrategyForm(holding: StrategyHolding) {
+    setStrategyFormOpen(true);
     setStrategyForm({
       id: holding.id,
       strategy_key: holding.strategy_key,
@@ -166,6 +186,7 @@ export function PortfolioShell({
 
   function resetStrategyForm() {
     setStrategyForm(EMPTY_STRATEGY_FORM);
+    setStrategyFormOpen(false);
   }
 
   function markLocalMutation() {
@@ -180,11 +201,25 @@ export function PortfolioShell({
     }
   }, [focusTarget]);
 
+  useEffect(() => {
+    strategyAnalysisRef.current = safeStrategyAnalysis;
+  }, [safeStrategyAnalysis]);
+
+  useEffect(() => {
+    const cachedAnalysis = readCachedStrategyAnalysis();
+    if (!cachedAnalysis?.holdings.length) {
+      return;
+    }
+    setStrategyAnalysis((prev) => (normalizeStrategyAnalysis(prev).holdings.length ? prev : cachedAnalysis));
+    setStrategyHoldings((prev) => (prev.length ? prev : toHoldingList(cachedAnalysis)));
+  }, []);
+
   useEffect(() => {
     if (!initialPrefill?.stock_code && !initialPrefill?.stock_name) {
       return;
     }
     setSearchQuery(prefillLabel);
+    setStrategyFormOpen(true);
     setStrategyForm((prev) => ({
       ...prev,
       id: undefined,
@@ -207,7 +242,7 @@ export function PortfolioShell({
   }, [initialPrefill, prefillLabel]);
 
   useEffect(() => {
-    if (!loaded || !unlocked || initialLoadInFlightRef.current) {
+    if (!loaded || !unlocked || initialLoadInFlightRef.current || initialLoadCompletedRef.current) {
       return;
     }
     if (safeStrategyAnalysis.holdings.length && marketRegime) {
@@ -218,6 +253,20 @@ export function PortfolioShell({
     setIsAnalysisRefreshing(true);
     setMarketRegimeLoading(true);
     setMarketRegimeError(null);
+    void listStrategyHoldings()
+      .then((holdings) => {
+        if (localMutationVersionRef.current !== requestMutationVersion) {
+          return;
+        }
+        setStrategyHoldings(holdings);
+        setStrategyAnalysis((prev) => {
+          const normalizedPrev = normalizeStrategyAnalysis(prev);
+          return mergeHoldingsWithExistingAnalysis(holdings, normalizedPrev);
+        });
+      })
+      .catch(() => {
+        // Full analysis below still has a chance to hydrate the page.
+      });
     void Promise.allSettled([
       getStrategyHoldingsAnalysis(),
       getMarketRegime(),
@@ -226,8 +275,7 @@ export function PortfolioShell({
         if (analysisResult.status === "fulfilled") {
           if (localMutationVersionRef.current === requestMutationVersion) {
             const analysis = normalizeStrategyAnalysis(analysisResult.value);
-            setStrategyAnalysis(analysis);
-            setStrategyHoldings(toHoldingList(analysis));
+            applyStrategyAnalysis(analysis);
           }
         }
         if (marketRegimeResult.status === "fulfilled") {
@@ -242,6 +290,7 @@ export function PortfolioShell({
         }
       })
       .finally(() => {
+        initialLoadCompletedRef.current = true;
         initialLoadInFlightRef.current = false;
         setIsAnalysisRefreshing(false);
         setMarketRegimeLoading(false);
@@ -251,6 +300,7 @@ export function PortfolioShell({
   function applySearchResult(item: StockSearchResult) {
     setSearchQuery(item.name);
     setSearchResults([]);
+    setStrategyFormOpen(true);
     setStrategyForm((prev) => ({
       ...prev,
       id: undefined,
@@ -265,6 +315,23 @@ export function PortfolioShell({
     return normalizeStrategyAnalysis(analysis).holdings.map((item) => item.holding);
   }
 
+  function mergeHoldingsWithExistingAnalysis(
+    holdings: StrategyHolding[],
+    existingAnalysis: StrategyHoldingAnalysisResponse,
+  ) {
+    const existingById = new Map(
+      existingAnalysis.holdings
+        .filter((item) => item.holding.id != null)
+        .map((item) => [item.holding.id, item]),
+    );
+    const existingByCode = new Map(
+      existingAnalysis.holdings.map((item) => [item.holding.stock_code, item]),
+    );
+    return recalculateAnalysisSummary(
+      holdings.map((holding) => buildLocalHoldingAnalysis(holding, existingById.get(holding.id) ?? existingByCode.get(holding.stock_code))),
+    );
+  }
+
   function recalculateAnalysisSummary(holdings: StrategyHoldingAnalysisResponse["holdings"]): StrategyHoldingAnalysisResponse {
     const investedHoldings = holdings.filter((item) => item.holding.status !== "watching" && item.holding.status !== "planned");
     const total_cost = investedHoldings.reduce((sum, item) => sum + item.holding.entry_price * item.holding.quantity, 0);
@@ -305,22 +372,6 @@ export function PortfolioShell({
     };
   }
 
-  async function refreshStrategyAnalysis(options?: { successMessage?: string; errorPrefix?: string }) {
-    setIsAnalysisRefreshing(true);
-    try {
-      const latest = await getStrategyHoldingsAnalysis();
-      setStrategyAnalysis(normalizeStrategyAnalysis(latest));
-      setStrategyHoldings(toHoldingList(latest));
-      if (options?.successMessage) {
-        setMessage(options.successMessage);
-      }
-    } catch (error) {
-      setMessage(`${options?.errorPrefix ?? "刷新失败"}: ${getClientErrorMessage(error)}`);
-    } finally {
-      setIsAnalysisRefreshing(false);
-    }
-  }
-
   function syncHoldingInAnalysis(updatedHolding: StrategyHolding) {
     setStrategyAnalysis((prev) => {
       const existing = prev.holdings.find((item) => item.holding.id === updatedHolding.id);
@@ -354,7 +405,14 @@ export function PortfolioShell({
       try {
         const results = await searchStocks(searchQuery.trim());
         setSearchResults(results);
-        setMessage(results.length ? `找到 ${results.length} 条匹配结果。` : "没有找到匹配股票。");
+        const exactResult = results.find((item) => item.match_type === "exact_name" || item.match_type === "exact_code");
+        const resultToApply = results.length === 1 ? results[0] : exactResult;
+        if (resultToApply) {
+          applySearchResult(resultToApply);
+          setMessage(`已找到并带入 ${resultToApply.name} (${resultToApply.code})。`);
+        } else {
+          setMessage(results.length ? `找到 ${results.length} 条匹配结果，请选择一只填入表单。` : "没有找到匹配股票。");
+        }
       } catch (error) {
         setMessage(`搜索失败: ${getClientErrorMessage(error)}`);
       } finally {
@@ -460,13 +518,11 @@ export function PortfolioShell({
     void (async () => {
       try {
         const [latestResult, marketRegimeResult] = await Promise.allSettled([
-          getStrategyHoldingsAnalysis(),
+          refreshStrategyHoldings(),
           getMarketRegime(),
         ]);
         if (latestResult.status === "fulfilled") {
-          const latest = normalizeStrategyAnalysis(latestResult.value);
-          setStrategyAnalysis(latest);
-          setStrategyHoldings(toHoldingList(latest));
+          applyStrategyAnalysis(latestResult.value);
         } else {
           throw latestResult.reason;
         }
@@ -647,6 +703,44 @@ export function PortfolioShell({
       <section className="content-grid">
         <div className="panel section">
           <h2>策略总览</h2>
+          <div className="portfolio-action-strip">
+            <div>
+              <span className="portfolio-action-kicker">今日优先</span>
+              <strong>
+                {safeStrategyAnalysis.todo_items.length
+                  ? `${safeStrategyAnalysis.todo_items[0].stock_name} · ${safeStrategyAnalysis.todo_items[0].action_label}`
+                  : "暂无必须处理动作"}
+              </strong>
+              <p className="muted">
+                {safeStrategyAnalysis.todo_items.length
+                  ? safeStrategyAnalysis.todo_items[0].action_reason
+                  : "可以先保持观察，或刷新策略持股获取最新判断。"}
+              </p>
+            </div>
+            <div className="portfolio-action-list">
+              {safeStrategyAnalysis.todo_items.slice(0, 2).map((item) => (
+                <Link
+                  href={`#holding-${item.stock_code}`}
+                  className="portfolio-action-chip"
+                  key={`${item.holding_id}-${item.stock_code}-${item.action_label}`}
+                >
+                  <span>{item.stock_name}</span>
+                  <strong>{item.action_label}</strong>
+                </Link>
+              ))}
+              {!safeStrategyAnalysis.todo_items.length ? (
+                <button
+                  className="portfolio-action-chip"
+                  type="button"
+                  onClick={onRefreshStrategyAnalysis}
+                  disabled={isBusy || isAnalysisRefreshing}
+                >
+                  <span>策略模型</span>
+                  <strong>{strategyRefreshInProgress ? "刷新中" : "刷新一次"}</strong>
+                </button>
+              ) : null}
+            </div>
+          </div>
           <div className="metric-grid">
             <div className="card">
               <div className="muted">策略持股总成本</div>
@@ -694,6 +788,23 @@ export function PortfolioShell({
         </div>
 
         <div className="panel section">
+          <div className="news-section-head">
+            <div>
+              <div className="section-kicker">Strategy Entry</div>
+              <h2>{strategyForm.id ? "编辑策略持股" : "新增策略持股"}</h2>
+            </div>
+            <button
+              className="button ghost"
+              type="button"
+              onClick={() => setStrategyFormOpen((current) => !current)}
+              aria-expanded={strategyFormOpen}
+            >
+              {strategyFormOpen ? "收起" : "展开录入"}
+            </button>
+          </div>
+
+          {strategyFormOpen ? (
+            <>
           <form className="stack" onSubmit={onSearch}>
             <label className="label">
               搜索股票（可选，带入下方表单）
@@ -732,7 +843,6 @@ export function PortfolioShell({
 
           <form className="stack" onSubmit={onSubmitStrategy} style={{ marginTop: 18 }}>
             <div>
-              <h2 style={{ marginBottom: 6 }}>{strategyForm.id ? "编辑策略持股" : "新增策略持股"}</h2>
               <p className="muted" style={{ fontSize: 13, margin: 0 }}>
                 {strategyForm.id
                   ? "改完点保存；计划细节在下方折叠里。"
@@ -969,6 +1079,15 @@ export function PortfolioShell({
               </button>
             </div>
           </form>
+            </>
+          ) : (
+            <div className="portfolio-entry-compact">
+              <p className="muted">日常复盘先看下方持仓动作；需要新增、编辑或从搜索带入股票时再展开。</p>
+              <button className="button" type="button" onClick={() => setStrategyFormOpen(true)}>
+                添加/记录新标的
+              </button>
+            </div>
+          )}
         </div>
       </section>
 
@@ -1057,7 +1176,7 @@ export function PortfolioShell({
               </div>
               <div className="card">
                 <div className="muted">平均策略分</div>
-                <strong>{safeStrategyAnalysis.average_score.toFixed(2)}</strong>
+                <strong>{scoredStrategyHoldingCount ? safeStrategyAnalysis.average_score.toFixed(2) : "待刷新"}</strong>
               </div>
             </div>
             <div className="stack" style={{ gap: 18 }}>
@@ -1072,9 +1191,13 @@ export function PortfolioShell({
                       <h3>{group.title}</h3>
                       <p className="muted">{group.description}</p>
                     </div>
-                    <div className="position-grid portfolio-holding-grid">
+                    <div className="position-grid portfolio-holding-grid" data-layout={holdingGridLayout(items.length)}>
                       {items.map((item) => (
-                        <div className="card portfolio-holding-card" key={`${item.holding.id}-${item.holding.stock_code}`}>
+                        <div
+                          className="card portfolio-holding-card"
+                          id={`holding-${item.holding.stock_code}`}
+                          key={`${item.holding.id}-${item.holding.stock_code}`}
+                        >
                           <h3>{item.holding.stock_name} ({item.holding.stock_code})</h3>
                           <p className="muted">
                             {item.holding.strategy_key} · 计划价/成本 {item.holding.entry_price.toFixed(2)} · 数量 {item.holding.quantity}
@@ -1090,7 +1213,7 @@ export function PortfolioShell({
                             </div>
                             <div className="card portfolio-key-metric">
                               <div className="muted portfolio-key-metric-label">总分</div>
-                              <strong>{item.strategy_score.total.toFixed(1)}</strong>
+                              <strong>{hasStrategyScore(item) ? item.strategy_score.total.toFixed(1) : "待刷新"}</strong>
                             </div>
                             <div className="card portfolio-key-metric">
                               <div className="muted portfolio-key-metric-label">状态</div>
@@ -1125,12 +1248,12 @@ export function PortfolioShell({
                                 <div key={`${item.holding.id}-${label}`} className="portfolio-canslim-item">
                                   <div className="portfolio-canslim-item-head">
                                     <span>{label}</span>
-                                    <span>{score.toFixed(0)}</span>
+                                    <span>{hasStrategyScore(item) ? score.toFixed(0) : "-"}</span>
                                   </div>
                                   <div className="portfolio-canslim-track">
                                     <div
                                       style={{
-                                        width: `${Math.max(0, Math.min(100, score))}%`,
+                                        width: hasStrategyScore(item) ? `${Math.max(0, Math.min(100, score))}%` : "0%",
                                         height: "100%",
                                         background: scoreTone(score),
                                       }}
@@ -1284,6 +1407,12 @@ function statusLabel(status: StrategyHolding["status"]) {
   return "已失效";
 }
 
+function holdingGridLayout(count: number) {
+  if (count <= 1) return "single";
+  if (count === 2) return "pair";
+  return "many";
+}
+
 function buildTodoItems(holdings: StrategyHoldingAnalysisResponse["holdings"]) {
   return holdings
     .map((item) => ({
@@ -1342,6 +1471,7 @@ function normalizeStrategyAnalysis(
           action_label: item?.action_label ?? "继续持有",
           action_reason: item?.action_reason ?? "",
           invalidation_reason: item?.invalidation_reason ?? null,
+          analysis_status: normalizeAnalysisStatus(item),
         }))
       : [],
     todo_items: Array.isArray(analysis.todo_items) ? analysis.todo_items : [],
@@ -1349,6 +1479,101 @@ function normalizeStrategyAnalysis(
   };
 }
 
+function normalizeAnalysisStatus(item: Partial<StrategyHoldingAnalysis> | undefined): StrategyHoldingAnalysis["analysis_status"] {
+  const status = item?.analysis_status;
+  if (status === "live" || status === "cached" || status === "degraded" || status === "local") {
+    return status;
+  }
+  return (item?.strategy_score?.total ?? 0) > 0 ? "cached" : "local";
+}
+
+function readCachedStrategyAnalysis(): StrategyHoldingAnalysisResponse | null {
+  if (typeof window === "undefined") {
+    return null;
+  }
+  try {
+    const raw = window.localStorage.getItem(STRATEGY_ANALYSIS_CACHE_KEY);
+    if (!raw) {
+      return null;
+    }
+    return normalizeStrategyAnalysis(JSON.parse(raw) as StrategyHoldingAnalysisResponse);
+  } catch {
+    return null;
+  }
+}
+
+function cacheStrategyAnalysis(analysis: StrategyHoldingAnalysisResponse) {
+  if (typeof window === "undefined") {
+    return;
+  }
+  try {
+    window.localStorage.setItem(STRATEGY_ANALYSIS_CACHE_KEY, JSON.stringify(normalizeStrategyAnalysis(analysis)));
+  } catch {
+    // Ignore storage quota or private mode failures; the live analysis is still displayed.
+  }
+}
+
+function mergeLatestAnalysisWithExisting(
+  latest: StrategyHoldingAnalysisResponse,
+  existing: StrategyHoldingAnalysisResponse,
+) {
+  const existingById = new Map(
+    existing.holdings
+      .filter((item) => item.holding.id != null)
+      .map((item) => [item.holding.id, item]),
+  );
+  const existingByCode = new Map(existing.holdings.map((item) => [item.holding.stock_code, item]));
+  const mergedHoldings = latest.holdings.map((item) => {
+    if (hasStrategyScore(item)) {
+      return item;
+    }
+    const previous = existingById.get(item.holding.id) ?? existingByCode.get(item.holding.stock_code);
+    return previous && hasStrategyScore(previous) ? buildLocalHoldingAnalysis(item.holding, previous) : item;
+  });
+  return recalculateStrategyAnalysis(latest, mergedHoldings);
+}
+
+function recalculateStrategyAnalysis(
+  base: StrategyHoldingAnalysisResponse,
+  holdings: StrategyHoldingAnalysisResponse["holdings"],
+): StrategyHoldingAnalysisResponse {
+  const investedHoldings = holdings.filter((item) => item.holding.status !== "watching" && item.holding.status !== "planned");
+  const total_cost = investedHoldings.reduce((sum, item) => sum + item.holding.entry_price * item.holding.quantity, 0);
+  const total_market_value = investedHoldings.reduce((sum, item) => sum + item.market_value, 0);
+  const total_realized_pnl = investedHoldings.reduce((sum, item) => sum + item.realized_pnl, 0);
+  const active_count = holdings.filter((item) => item.holding.status === "holding" || item.holding.status === "weakening").length;
+  const watching_count = holdings.filter((item) => item.holding.status === "watching").length;
+  const planned_count = holdings.filter((item) => item.holding.status === "planned").length;
+  const weakening_count = holdings.filter((item) => item.holding.status === "weakening").length;
+  const exited_count = holdings.filter((item) => item.holding.status === "exited").length;
+  const invalidated_count = holdings.filter((item) => item.holding.status === "invalidated").length;
+  const scoredHoldings = holdings.filter(hasStrategyScore);
+  const total_pnl = total_market_value - total_cost;
+  const exited_win_count = holdings.filter((item) => item.holding.status === "exited" && item.realized_pnl > 0).length;
+  return {
+    ...base,
+    total_cost,
+    total_market_value,
+    total_pnl,
+    total_pnl_pct: total_cost ? (total_pnl / total_cost) * 100 : 0,
+    total_realized_pnl,
+    holding_count: holdings.length,
+    active_count,
+    watching_count,
+    planned_count,
+    weakening_count,
+    exited_count,
+    invalidated_count,
+    win_rate_pct: exited_count ? (exited_win_count / exited_count) * 100 : 0,
+    average_score: scoredHoldings.length
+      ? scoredHoldings.reduce((sum, item) => sum + item.strategy_score.total, 0) / scoredHoldings.length
+      : 0,
+    todo_items: buildTodoItems(holdings),
+    review_items: buildReviewItems(holdings),
+    holdings,
+  };
+}
+
 function buildLocalHoldingAnalysis(
   holding: StrategyHolding,
   existing?: StrategyHoldingAnalysis,
@@ -1391,6 +1616,7 @@ function buildLocalHoldingAnalysis(
     trigger_hits: existing?.trigger_hits ?? [],
     alerts:
       existing?.alerts ?? ["当前展示的是本地记录快照，最新评分与盈亏请手动刷新。"],
+    analysis_status: existing && hasStrategyScore(existing) ? "cached" : "local",
   };
 }
 
@@ -1409,8 +1635,15 @@ function scoreTone(score: number) {
   return "linear-gradient(90deg, #ef4444, #f87171)";
 }
 
+function hasStrategyScore(item: StrategyHoldingAnalysis) {
+  return item.strategy_score.total > 0 && (item.analysis_status === "live" || item.analysis_status === "cached");
+}
+
 function buildRiskNotes(item: StrategyHoldingAnalysis) {
   const notes: string[] = [];
+  if (!hasStrategyScore(item)) {
+    return (item.alerts.length ? item.alerts : [item.action_reason || "已显示本地记录，等待最新评分刷新。"]).slice(0, 2);
+  }
   const lowFactors = ([
     ["C", item.strategy_score.c],
     ["A", item.strategy_score.a],
diff --git a/components/portfolio-snapshot-panel.tsx b/components/portfolio-snapshot-panel.tsx
index 17aab4c..f9883da 100644
--- a/components/portfolio-snapshot-panel.tsx
+++ b/components/portfolio-snapshot-panel.tsx
@@ -1,6 +1,7 @@
 "use client";
 
 import { useEffect, useState } from "react";
+import Link from "next/link";
 
 import { getPortfolioAnalysis } from "@/lib/api";
 import type { PortfolioAnalysisResponse } from "@/lib/types";
@@ -16,6 +17,9 @@ const PORTFOLIO_COPY = {
     returnRate: "收益率",
     technicalRisk: "技术风险",
     unavailable: "后端不可达时，这里会显示组合快照。",
+    nextAction: "下一步",
+    openPortfolio: "打开持仓页",
+    calm: "组合暂无明确调仓提示。",
   },
   en: {
     title: "Portfolio Snapshot",
@@ -25,6 +29,9 @@ const PORTFOLIO_COPY = {
     returnRate: "Return",
     technicalRisk: "Technical Risk",
     unavailable: "Portfolio metrics will appear here after the backend is available.",
+    nextAction: "Next action",
+    openPortfolio: "Open portfolio",
+    calm: "No urgent portfolio action right now.",
   },
 } as const;
 
@@ -64,24 +71,35 @@ export function PortfolioSnapshotPanel({ locale = "zh" }: { locale?: PortfolioSn
       {loading ? (
         <p className="muted">{copy.loading}</p>
       ) : portfolio ? (
-        <div className="metric-grid">
-          <div className="card">
-            <div className="muted">{copy.totalMarketValue}</div>
-            <strong>{portfolio.total_market_value.toFixed(2)}</strong>
+        <div className="stack">
+          <div className="portfolio-snapshot-action">
+            <div>
+              <span>{copy.nextAction}</span>
+              <strong>{portfolio.rebalance_suggestions[0] ?? copy.calm}</strong>
+            </div>
+            <Link href="/portfolio" className="button ghost">
+              {copy.openPortfolio}
+            </Link>
           </div>
-          <div className="card">
-            <div className="muted">{copy.totalPnl}</div>
-            <strong className={portfolio.total_pnl >= 0 ? "signal-up" : "signal-down"}>
-              {portfolio.total_pnl.toFixed(2)}
-            </strong>
-          </div>
-          <div className="card">
-            <div className="muted">{copy.returnRate}</div>
-            <strong>{portfolio.total_pnl_pct.toFixed(2)}%</strong>
-          </div>
-          <div className="card">
-            <div className="muted">{copy.technicalRisk}</div>
-            <strong>{portfolio.technical_risk}</strong>
+          <div className="metric-grid">
+            <div className="card">
+              <div className="muted">{copy.totalMarketValue}</div>
+              <strong>{portfolio.total_market_value.toFixed(2)}</strong>
+            </div>
+            <div className="card">
+              <div className="muted">{copy.totalPnl}</div>
+              <strong className={portfolio.total_pnl >= 0 ? "signal-up" : "signal-down"}>
+                {portfolio.total_pnl.toFixed(2)}
+              </strong>
+            </div>
+            <div className="card">
+              <div className="muted">{copy.returnRate}</div>
+              <strong>{portfolio.total_pnl_pct.toFixed(2)}%</strong>
+            </div>
+            <div className="card">
+              <div className="muted">{copy.technicalRisk}</div>
+              <strong>{portfolio.technical_risk}</strong>
+            </div>
           </div>
         </div>
       ) : (
diff --git a/components/stocks-page-client.tsx b/components/stocks-page-client.tsx
index 70c5b56..a6a75a6 100644
--- a/components/stocks-page-client.tsx
+++ b/components/stocks-page-client.tsx
@@ -41,6 +41,7 @@ export function StocksPageClient({
 
   const [analysis, setAnalysis] = useState<StockAnalysisResponse | null>(null);
   const [analysisLoading, setAnalysisLoading] = useState(false);
+  const [aiLoading, setAiLoading] = useState(false);
   const [analysisError, setAnalysisError] = useState("");
   const [analysisIncludesAi, setAnalysisIncludesAi] = useState(false);
 
@@ -114,6 +115,7 @@ export function StocksPageClient({
       setAnalysis(null);
       setAnalysisError("");
       setAnalysisLoading(false);
+      setAiLoading(false);
       return () => {
         cancelled = true;
       };
@@ -125,33 +127,66 @@ export function StocksPageClient({
     if (canReuseAnalysis) {
       setAnalysisError("");
       setAnalysisLoading(false);
+      setAiLoading(false);
       return () => {
         cancelled = true;
       };
     }
 
-    setAnalysisLoading(true);
     setAnalysisError("");
 
-    getStockAnalysis(selected.code, { includeAi: shouldLoadAi, requestId: initialRequestId })
-      .then((result) => {
+    // 两段式加载：先用 include_ai=false 快速拿到行情/指标/K 线渲染出来，
+    // 再在后台补充 AI 文本，整个过程图表不会被清空或被进度条接管。
+    if (!sameStockLoaded) {
+      setAnalysisLoading(true);
+      getStockAnalysis(selected.code, { includeAi: false, requestId: initialRequestId })
+        .then((base) => {
+          if (cancelled) {
+            return;
+          }
+          setAnalysis(base);
+          setAnalysisIncludesAi(false);
+          // shouldLoadAi 为真时，下一次 effect 重跑会进入下面的 AI 补充分支。
+        })
+        .catch((error) => {
+          if (cancelled) {
+            return;
+          }
+          setAnalysis(null);
+          setAnalysisIncludesAi(false);
+          setAnalysisError(error instanceof Error ? error.message : "单股分析生成失败");
+        })
+        .finally(() => {
+          if (!cancelled) {
+            setAnalysisLoading(false);
+          }
+        });
+
+      return () => {
+        cancelled = true;
+      };
+    }
+
+    // 已经有该股票的基础分析，仅缺 AI：后台增量加载，保留现有图表可见。
+    setAiLoading(true);
+    getStockAnalysis(selected.code, { includeAi: true, requestId: initialRequestId })
+      .then((full) => {
         if (cancelled) {
           return;
         }
-        setAnalysis(result);
-        setAnalysisIncludesAi(shouldLoadAi);
+        setAnalysis(full);
+        setAnalysisIncludesAi(true);
       })
       .catch((error) => {
         if (cancelled) {
           return;
         }
-        setAnalysis(null);
-        setAnalysisIncludesAi(false);
-        setAnalysisError(error instanceof Error ? error.message : "单股分析生成失败");
+        // AI 补充失败时保留已显示的基础分析，仅提示 AI 部分出错。
+        setAnalysisError(error instanceof Error ? error.message : "AI 分析生成失败");
       })
       .finally(() => {
         if (!cancelled) {
-          setAnalysisLoading(false);
+          setAiLoading(false);
         }
       });
 
@@ -285,34 +320,53 @@ export function StocksPageClient({
 
       {analysis ? (
         <>
+          <section className="stocks-focus-bar" aria-label="当前股票快捷概览">
+            <div className="stocks-focus-main">
+              <span className="stocks-focus-market">{analysis.market || selected.market}</span>
+              <strong>{analysis.stock_name} ({analysis.stock_code})</strong>
+              <span className="muted">信号 {analysis.signal_summary.overall_signal} · 评分 {analysis.signal_summary.overall_score}/5</span>
+            </div>
+            <div className="stocks-focus-metrics">
+              <span>
+                现价 <strong>{analysis.quote.current_price.toFixed(2)}</strong>
+              </span>
+              <span className={analysis.quote.change_pct >= 0 ? "signal-up" : "signal-down"}>
+                {analysis.quote.change_pct.toFixed(2)}%
+              </span>
+            </div>
+            <div className="stocks-focus-actions">
+              <Link href={`/stocks?query=${encodeURIComponent(selected.code)}&panel=chart#chart`} prefetch={false}>K线</Link>
+              <StockPanelLink stockCode={selected.code} panel="ai">AI</StockPanelLink>
+              <StockPanelLink stockCode={selected.code} panel="news">新闻</StockPanelLink>
+              <Link href={`/portfolio?stock_code=${encodeURIComponent(analysis.stock_code)}&stock_name=${encodeURIComponent(analysis.stock_name)}&cost_price=${encodeURIComponent(String(analysis.quote.current_price))}&status=planned&return_to=${encodeURIComponent(`/stocks?query=${analysis.stock_code}`)}&return_label=${encodeURIComponent("单股分析")}`} className="button ghost">
+                加计划
+              </Link>
+            </div>
+          </section>
+
           <section className={`panel section ${activePanel === "overview" ? "focus-panel" : ""}`} id="overview">
-            <div className="metric-grid">
-              <div className="card">
-                <div className="muted">最新价</div>
-                <strong>{analysis.quote.current_price.toFixed(2)}</strong>
-              </div>
-              <div className="card">
-                <div className="muted">涨跌幅</div>
-                <strong className={analysis.quote.change_pct >= 0 ? "signal-up" : "signal-down"}>
-                  {analysis.quote.change_pct.toFixed(2)}%
-                </strong>
-              </div>
-              <div className="card">
-                <div className="muted">综合信号</div>
-                <strong>{analysis.signal_summary.overall_signal}</strong>
-              </div>
-              <div className="card">
-                <div className="muted">评分</div>
-                <strong>{analysis.signal_summary.overall_score}/5</strong>
+            <div className="news-section-head">
+              <div>
+                <div className="section-kicker">Technical Context</div>
+                <h2>技术环境</h2>
               </div>
+              <span className="muted">减少重复价格信息，优先看关键指标。</span>
+            </div>
+            <div className="metric-grid">
+              {buildOverviewIndicators(analysis.technical_indicators).map(([key, value]) => (
+                <div className="card" key={key}>
+                  <div className="muted">{formatIndicatorLabel(key)}</div>
+                  <strong>{value === null ? "-" : value.toFixed(2)}</strong>
+                </div>
+              ))}
             </div>
           </section>
 
           <section className="content-grid">
-            <div className="panel section" id="chart">
-              <h2>K 线缩略</h2>
-              <CandlestickChart data={analysis.chart_series} height={280} symbol={`${analysis.stock_name} ${analysis.stock_code}`} />
-              <p className="muted">使用轻量级蜡烛图组件展示最近行情，支持自适应宽度。</p>
+            <div className={`panel section ${activePanel === "chart" ? "focus-panel" : ""}`} id="chart">
+              <h2>K 线图</h2>
+              <CandlestickChart data={analysis.chart_series} height={activePanel === "chart" ? 420 : 280} symbol={`${analysis.stock_name} ${analysis.stock_code}`} />
+              <p className="muted">展示最近行情的日 K 线，支持自适应宽度；配合上方价格与涨跌幅做节奏判断。</p>
               <div style={{ marginTop: 18 }}>
                 <h3 style={{ marginBottom: 12 }}>技术分析建议</h3>
                 <div className="news-grid stocks-tech-grid">
@@ -345,6 +399,8 @@ export function StocksPageClient({
                   title="AI 分析已锁定"
                   description="解锁后可以查看更长的 AI 观点、结论和操作建议。"
                 />
+              ) : shouldLoadAi && aiLoading && !analysisIncludesAi ? (
+                <p className="muted">AI 正在生成分析报告，行情与指标已就绪，请稍候…</p>
               ) : shouldLoadAi && analysis.ai_insight.enabled ? (
                 <AICarousel
                   content={analysis.ai_insight.content || analysis.ai_insight.error || ""}
@@ -396,24 +452,40 @@ export function StocksPageClient({
 }
 
 function buildDirectCodeFallback(query: string): StockSearchResult[] {
-  if (!/^[a-z]{2}\d{6}$/i.test(query)) {
-    return [];
+  const trimmed = query.trim();
+  // A股 sh/sz 数字代码
+  if (/^[a-z]{2}\d{6}$/i.test(trimmed)) {
+    const code = trimmed.toLowerCase();
+    return [
+      {
+        name: `代码 ${code}`,
+        code,
+        market: code.startsWith("sh") ? "A股-上海" : "A股-深圳",
+        category: "",
+        score: 50,
+        match_type: "direct_code",
+      },
+    ];
   }
-  const code = query.toLowerCase();
-  return [
-    {
-      name: `代码 ${code}`,
-      code,
-      market: code.startsWith("sh") ? "A股-上海" : "A股-深圳",
-      category: "",
-      score: 50,
-      match_type: "direct_code",
-    },
-  ];
+  // 美股 ticker（纯字母，可带类别后缀，如 AAPL、BRK.B）
+  if (/^[a-z]{1,5}(\.[a-z]{1,2})?$/i.test(trimmed)) {
+    const code = `US.${trimmed.toUpperCase()}`;
+    return [
+      {
+        name: `代码 ${code}`,
+        code,
+        market: "美股",
+        category: "",
+        score: 50,
+        match_type: "direct_code",
+      },
+    ];
+  }
+  return [];
 }
 
 function normalizePanel(value?: string | null) {
-  if (value === "ai" || value === "news") {
+  if (value === "ai" || value === "news" || value === "chart") {
     return value;
   }
   return "overview";
@@ -432,3 +504,23 @@ function formatSentiment(value: NewsItem["sentiment"]) {
   if (value === "bearish") return "偏利空";
   return "中性";
 }
+
+function buildOverviewIndicators(indicators: StockAnalysisResponse["technical_indicators"]) {
+  const preferredKeys = ["MA5", "MA20", "MA60", "RSI", "MACD"];
+  const preferred = preferredKeys
+    .filter((key) => Object.prototype.hasOwnProperty.call(indicators, key))
+    .map((key) => [key, indicators[key]] as [string, number | null]);
+  const fallback = Object.entries(indicators).filter(([key]) => !preferredKeys.includes(key));
+  return [...preferred, ...fallback].slice(0, 4);
+}
+
+function formatIndicatorLabel(key: string) {
+  const labels: Record<string, string> = {
+    MA5: "MA5",
+    MA20: "MA20",
+    MA60: "MA60",
+    RSI: "RSI",
+    MACD: "MACD",
+  };
+  return labels[key] ?? key;
+}
diff --git a/data/stock_pool.json b/data/stock_pool.json
index 3ac77ae..d5875c7 100644
--- a/data/stock_pool.json
+++ b/data/stock_pool.json
@@ -5,11 +5,21 @@
   "中国石油股份": "00857.HK",
   "中国船舶": "sh600150",
   "云南锗业": "sz002428",
+  "埃斯顿": "sz002747",
+  "埃斯顿机器人": "sz002747",
   "优必选": "09880.HK",
   "兆易创新": "sh603986",
+  "华勤技术": "sh603296",
   "富瀚微": "sz300613",
   "寒武纪": "sh688256",
+  "沃尔核材": "sz002130",
   "永鼎股份": "sh600105",
   "海光信息": "sh688041",
+  "澜起科技": "sh688008",
+  "瑞芯微": "sh603893",
+  "申菱环境": "sz301018",
+  "长电": "sh600584",
+  "长电科技": "sh600584",
+  "麦格米特": "sz002851",
   "阿里巴巴": "09988.HK"
-}
\ No newline at end of file
+}
diff --git a/eslint.config.mjs b/eslint.config.mjs
new file mode 100644
index 0000000..b998323
--- /dev/null
+++ b/eslint.config.mjs
@@ -0,0 +1,22 @@
+import nextVitals from "eslint-config-next/core-web-vitals";
+import nextTypeScript from "eslint-config-next/typescript";
+
+export default [
+  {
+    ignores: [
+      ".next/**",
+      ".next-dev/**",
+      ".next.bak-*/**",
+      "node_modules/**",
+    ],
+  },
+  ...nextVitals,
+  ...nextTypeScript,
+  {
+    rules: {
+      "react-hooks/immutability": "off",
+      "react-hooks/preserve-manual-memoization": "off",
+      "react-hooks/set-state-in-effect": "off",
+    },
+  },
+];
diff --git a/lib/api.ts b/lib/api.ts
index efd1f64..bee1599 100644
--- a/lib/api.ts
+++ b/lib/api.ts
@@ -257,7 +257,7 @@ export function getHotspots(): Promise<HotspotItem[]> {
 }
 
 export function getGlobalNews(): Promise<GlobalNewsItem[]> {
-  return requestWithTimeout("/api/news/global", 8000);
+  return requestWithTimeout("/api/news/global", 8000, { cache: "no-store" });
 }
 
 export function webSearch(query: string, limit = 8): Promise<WebSearchResult[]> {
@@ -371,7 +371,7 @@ export async function deleteStrategyHolding(id: number): Promise<void> {
 export function getStrategyHoldingsAnalysis(options?: {
   requestInit?: RequestInit;
 }): Promise<StrategyHoldingAnalysisResponse> {
-  return requestWithTimeout("/api/strategy-holdings/analysis", 8000, {
+  return requestWithTimeout("/api/strategy-holdings/analysis", 30000, {
     ...(options?.requestInit ?? {}),
     cache: "no-store",
   });
diff --git a/lib/types.ts b/lib/types.ts
index 29c67ba..25aaf6e 100644
--- a/lib/types.ts
+++ b/lib/types.ts
@@ -108,6 +108,15 @@ export type HotspotItem = {
   trend_direction: "up" | "down" | "flat";
   ai_summary?: string | null;
   source: string;
+  heat_breakdown: {
+    trading_activity: number;
+    discussion_heat: number;
+    news_heat: number;
+    alert_count: number;
+    rank_hits: number;
+    attention_level: "focus" | "caution" | "watch";
+    basis: string[];
+  };
   related_stocks: Array<{
     stock_name: string;
     stock_code: string;
@@ -252,6 +261,7 @@ export type StrategyHoldingAnalysis = {
   action_reason: string;
   trigger_hits: string[];
   alerts: string[];
+  analysis_status: "live" | "cached" | "degraded" | "local";
 };
 
 export type StrategyTodoItem = {
diff --git a/package.json b/package.json
index b4f2447..dd3e68e 100644
--- a/package.json
+++ b/package.json
@@ -7,7 +7,7 @@
     "pm2:start": "pm2 start ecosystem.config.cjs",
     "pm2:stop": "pm2 delete openashare-api openashare-web",
     "pm2:flush": "pm2 flush",
-    "lint": "next lint"
+    "lint": "eslint app components lib next.config.ts"
   },
   "engines": {
     "node": ">=20 <23"
diff --git a/reports/us_tech_h2_2026_sector_analysis.html b/reports/us_tech_h2_2026_sector_analysis.html
new file mode 100644
index 0000000..f6a68a5
--- /dev/null
+++ b/reports/us_tech_h2_2026_sector_analysis.html
@@ -0,0 +1,364 @@
+<!doctype html>
+<html lang="zh-CN">
+<head>
+  <meta charset="utf-8">
+  <meta name="viewport" content="width=device-width, initial-scale=1">
+  <title>美股科技四大主题：2026 下半年理性投资分析</title>
+  <style>
+    :root {
+      color-scheme: light;
+      --ink: #172033;
+      --muted: #5a667a;
+      --line: #d8dee8;
+      --bg: #f6f8fb;
+      --panel: #ffffff;
+      --blue: #1f6feb;
+      --green: #1b7f4b;
+      --amber: #a56200;
+      --red: #b42318;
+    }
+    * { box-sizing: border-box; }
+    body {
+      margin: 0;
+      font-family: -apple-system, BlinkMacSystemFont, "Segoe UI", Arial, "PingFang SC", "Microsoft YaHei", sans-serif;
+      color: var(--ink);
+      background: var(--bg);
+      line-height: 1.55;
+    }
+    main {
+      max-width: 1180px;
+      margin: 0 auto;
+      padding: 36px 22px 56px;
+    }
+    header {
+      padding: 28px 0 18px;
+      border-bottom: 2px solid var(--ink);
+    }
+    h1 {
+      margin: 0 0 12px;
+      font-size: clamp(28px, 4vw, 44px);
+      line-height: 1.12;
+      letter-spacing: 0;
+    }
+    h2 {
+      margin: 38px 0 14px;
+      font-size: 25px;
+      letter-spacing: 0;
+    }
+    h3 {
+      margin: 24px 0 10px;
+      font-size: 19px;
+      letter-spacing: 0;
+    }
+    p { margin: 9px 0; }
+    a { color: var(--blue); text-decoration: none; }
+    a:hover { text-decoration: underline; }
+    .subtitle {
+      max-width: 900px;
+      color: var(--muted);
+      font-size: 16px;
+    }
+    .stamp {
+      display: inline-block;
+      margin-top: 10px;
+      color: var(--muted);
+      font-size: 13px;
+    }
+    .grid {
+      display: grid;
+      grid-template-columns: repeat(4, minmax(0, 1fr));
+      gap: 12px;
+      margin-top: 18px;
+    }
+    .tile {
+      background: var(--panel);
+      border: 1px solid var(--line);
+      border-radius: 8px;
+      padding: 15px;
+      min-height: 136px;
+    }
+    .tile strong {
+      display: block;
+      margin-bottom: 6px;
+      font-size: 15px;
+    }
+    .tile .score {
+      font-size: 22px;
+      font-weight: 700;
+      margin-bottom: 4px;
+    }
+    .note {
+      background: #fff8e6;
+      border: 1px solid #f1d28a;
+      border-radius: 8px;
+      padding: 12px 14px;
+      color: #503a00;
+      margin: 16px 0;
+    }
+    table {
+      width: 100%;
+      border-collapse: collapse;
+      background: var(--panel);
+      border: 1px solid var(--line);
+      border-radius: 8px;
+      overflow: hidden;
+      margin: 12px 0 18px;
+      font-size: 14px;
+    }
+    th, td {
+      padding: 11px 10px;
+      border-bottom: 1px solid var(--line);
+      vertical-align: top;
+      text-align: left;
+    }
+    th {
+      background: #eef3f9;
+      font-size: 13px;
+      color: #33415c;
+    }
+    tr:last-child td { border-bottom: 0; }
+    .tag {
+      display: inline-block;
+      padding: 2px 8px;
+      border-radius: 999px;
+      font-size: 12px;
+      font-weight: 600;
+      white-space: nowrap;
+    }
+    .advance { background: #e7f6ed; color: var(--green); }
+    .watch { background: #fff4da; color: var(--amber); }
+    .spec { background: #fce8e6; color: var(--red); }
+    .neutral { background: #eaf0ff; color: #2451a6; }
+    .two-col {
+      display: grid;
+      grid-template-columns: 1fr 1fr;
+      gap: 16px;
+    }
+    .section {
+      background: var(--panel);
+      border: 1px solid var(--line);
+      border-radius: 8px;
+      padding: 18px;
+      margin-top: 14px;
+    }
+    ul { padding-left: 20px; }
+    li { margin: 7px 0; }
+    .small { color: var(--muted); font-size: 13px; }
+    .source-list li { margin: 8px 0; }
+    @media (max-width: 900px) {
+      .grid, .two-col { grid-template-columns: 1fr; }
+      main { padding: 24px 14px 42px; }
+      table { display: block; overflow-x: auto; white-space: nowrap; }
+      th, td { white-space: normal; min-width: 150px; }
+    }
+  </style>
+</head>
+<body>
+<main>
+  <header>
+    <h1>美股科技四大主题：2026 下半年理性投资分析</h1>
+    <p class="subtitle">覆盖机器人/物理 AI、CPU 与 AI 硬件、存储、量子计算。本文是研究优先级和估值温度框架，不是买卖建议；结论基于截至 2026-06-12 美国市场附近可见的公开信息与新闻源。</p>
+    <span class="stamp">投资视角：中期 6-12 个月，偏基本面 + 催化剂 + 估值纪律。</span>
+  </header>
+
+  <section>
+    <h2>一页结论</h2>
+    <div class="grid">
+      <div class="tile">
+        <strong>最强基本面顺风</strong>
+        <div class="score">存储</div>
+        <p>HBM、企业 SSD、近线 HDD 同时受 AI 数据中心拉动，但涨幅很大，追高需要用合约、价格和供给扩张验证。</p>
+      </div>
+      <div class="tile">
+        <strong>最大利润池</strong>
+        <div class="score">AI 硬件</div>
+        <p>NVDA、AVGO、TSM、ASML、AMD 仍是核心链条；下半年不怕需求差，怕预期太满和指引不够激进。</p>
+      </div>
+      <div class="tile">
+        <strong>分化最明显</strong>
+        <div class="score">机器人</div>
+        <p>真实收入在手术、仓储、自动化测试；人形机器人和物理 AI 更像远期期权，验证点在量产和单位经济。</p>
+      </div>
+      <div class="tile">
+        <strong>最像期权</strong>
+        <div class="score">量子</div>
+        <p>政策和技术里程碑会带来交易性行情，但多数纯量子公司仍是高倍销售额、亏损和融资风险。</p>
+      </div>
+    </div>
+    <div class="note">我的排序：下半年风险调整后优先级为 存储 > AI 硬件 > 机器人/物理 AI > 量子计算。若只看长期技术天花板，量子和人形机器人很诱人；若看 2026 年下半年兑现度，存储和 AI 硬件更硬。</div>
+  </section>
+
+  <section>
+    <h2>板块走势与估值温度</h2>
+    <table>
+      <thead>
+        <tr>
+          <th>板块</th>
+          <th>下半年基础情景</th>
+          <th>估值温度</th>
+          <th>核心验证点</th>
+          <th>最大风险</th>
+        </tr>
+      </thead>
+      <tbody>
+        <tr>
+          <td>机器人和物理 AI</td>
+          <td>震荡上行但高度分化。真实订单/装机/程序量支持的公司更稳；人形机器人主题更多靠 Tesla 等事件驱动。</td>
+          <td><span class="tag watch">偏贵但分层</span> ISRG 是高质量溢价，TSLA 是可选性溢价，SYM/TER 更看订单兑现。</td>
+          <td>Optimus/robotaxi 进展、仓储自动化 backlog 转收入、da Vinci 程序量、工业自动化订单底部。</td>
+          <td>演示热度不能转化为收入；制造业资本开支下行；监管和安全事故。</td>
+        </tr>
+        <tr>
+          <td>CPU 与 AI 硬件</td>
+          <td>需求仍强，但近期 AVGO 指引引发市场对“beat-and-raise”预期的再定价。下半年更像高波动的基本面牛市。</td>
+          <td><span class="tag neutral">核心龙头不一定最贵</span> NVDA 已被部分资料描述为约 20x forward P/E；AMD/ARM/AVGO 是预期敏感型。</td>
+          <td>Blackwell/Rubin 供给、AMD GPU 第二来源份额、AVGO ASIC 客户集中度、TSM/ASML 产能与价格。</td>
+          <td>云厂商 capex 降速、自研 ASIC 替代、出口限制、利率上行导致长久期压缩。</td>
+        </tr>
+        <tr>
+          <td>存储</td>
+          <td>基本面最顺。DRAM/HBM、NAND、HDD 均从 AI 推理、训练、日志、数据湖受益；但交易已经拥挤。</td>
+          <td><span class="tag advance">盈利倍数仍低但周期性强</span> MU 被报道约 10x forward P/E；WDC/STX/SNDK 涨幅后需防周期顶部估值错觉。</td>
+          <td>DRAM/NAND 合约价、HBM 份额、HDD LTAs、企业 SSD 价格、库存重建。</td>
+          <td>供给扩张、客户双重采购后去库存、AI 资本开支降温、价格涨幅挤压下游需求。</td>
+        </tr>
+        <tr>
+          <td>量子计算</td>
+          <td>技术和政策催化多，股价弹性大；但 2026 年下半年仍不宜按成熟盈利模型估值。</td>
+          <td><span class="tag spec">极高期权价值</span> IONQ/RGTI/QBTS 等更多看 2027-2030 收入折现，纯 PE 不适用。</td>
+          <td>IonQ 256-qubit 里程碑、DARPA/政府项目、D-Wave bookings、Rigetti fidelity/资本开支、IBM/Google 路线图。</td>
+          <td>收入体量太小、亏损扩大、融资稀释、技术路线不确定、新闻驱动后的回撤。</td>
+        </tr>
+      </tbody>
+    </table>
+  </section>
+
+  <section>
+    <h2>机器人和物理 AI</h2>
+    <div class="section">
+      <h3>走势判断</h3>
+      <p>下半年主线不是“所有机器人都涨”，而是“能证明落地收入的机器人涨得更稳”。医疗机器人和仓储自动化已经有收入与装机基础；人形机器人和具身智能的天花板更高，但短期估值主要靠里程碑。Tesla 的 Optimus、robotaxi 和 AI 芯片叙事会继续影响板块风险偏好，但目前仍需要把演示、试生产和可规模化利润分开看。</p>
+      <h3>候选标的</h3>
+      <table>
+        <thead><tr><th>标的</th><th>暴露路径</th><th>估值/状态</th><th>下半年看点</th><th>研究优先级</th></tr></thead>
+        <tbody>
+          <tr><td>NVDA</td><td>机器人训练/仿真/边缘计算、Jetson、Isaac、AI 工厂基础设施</td><td>AI 硬件龙头，forward P/E 已从狂热区回落，但仍受数据中心周期支配</td><td>物理 AI 软件栈是否从叙事转订单</td><td><span class="tag advance">A</span></td></tr>
+          <tr><td>TSLA</td><td>Optimus、robotaxi、FSD、边缘推理芯片</td><td>机器人期权大，汽车基本面承压时估值更依赖未来叙事</td><td>Optimus 量产线、Cybercab/robotaxi 城市扩张、监管事件</td><td><span class="tag watch">B</span></td></tr>
+          <tr><td>ISRG</td><td>da Vinci 和 Ion 手术机器人</td><td>高质量成长溢价；Q1 2026 程序量和收入增长支持估值</td><td>da Vinci 5 渗透、程序量增长、国际扩张</td><td><span class="tag advance">A</span></td></tr>
+          <tr><td>SYM</td><td>仓储自动化系统，Walmart 等客户</td><td>高增长但客户集中度和项目节奏风险较高</td><td>backlog 转收入、毛利率、客户扩展</td><td><span class="tag watch">B</span></td></tr>
+          <tr><td>TER</td><td>协作机器人 Universal Robots/MiR，另有半导体测试</td><td>机器人业务不是唯一驱动，近期更多由 AI 半导体测试支撑</td><td>AI 测试需求与工业机器人周期修复</td><td><span class="tag advance">A-</span></td></tr>
+          <tr><td>ROK</td><td>工业自动化、工厂控制、制造业数字化</td><td>更像工业周期复苏标的，不是纯 AI 机器人</td><td>制造业 PMI、订单、软件/服务占比</td><td><span class="tag watch">B</span></td></tr>
+          <tr><td>AMZN</td><td>仓储机器人、物流自动化、AWS AI 工作负载</td><td>机器人价值被 AWS 和零售主业务稀释</td><td>物流效率改善是否体现在 margin</td><td><span class="tag watch">B</span></td></tr>
+        </tbody>
+      </table>
+    </div>
+  </section>
+
+  <section>
+    <h2>CPU 与 AI 硬件</h2>
+    <div class="section">
+      <h3>走势判断</h3>
+      <p>AI 硬件仍是利润池最大的板块，但下半年会从“只要 AI 就重估”转向“谁的订单、产能、毛利和客户集中度能经得住高预期”。近期 Broadcom 即使公布强劲 AI 收入，仍因指引没有超过市场高预期而引发板块波动，这说明估值风险已经从需求证伪转为预期管理。</p>
+      <h3>候选标的</h3>
+      <table>
+        <thead><tr><th>标的</th><th>暴露路径</th><th>估值/状态</th><th>下半年看点</th><th>研究优先级</th></tr></thead>
+        <tbody>
+          <tr><td>NVDA</td><td>GPU、网络、Grace/Vera CPU、AI rack-scale 系统</td><td>资料显示 forward P/E 约 20x，若盈利增长兑现，龙头估值反而不极端</td><td>Blackwell Ultra、Rubin、网络收入、毛利率</td><td><span class="tag advance">A</span></td></tr>
+          <tr><td>AMD</td><td>EPYC CPU、Instinct GPU、AI 第二来源</td><td>股价已反映 GPU 份额提升预期，仍低于 NVDA 生态确定性</td><td>Meta/OpenAI 等大客户份额、MI 系列供给、CPU 份额</td><td><span class="tag advance">A-</span></td></tr>
+          <tr><td>AVGO</td><td>定制 ASIC、网络、VMware 基础设施软件</td><td>AI ASIC 高成长但客户集中，近期市值回撤显示预期很高</td><td>Google/OpenAI/Anthropic 等 ASIC 订单和指引</td><td><span class="tag advance">A-</span></td></tr>
+          <tr><td>TSM</td><td>先进制程代工、CoWoS、AI 芯片供应瓶颈</td><td>高质量供应链核心，估值取决于先进节点价格和资本开支回报</td><td>3nm/2nm 价格、CoWoS 扩产、客户 capex</td><td><span class="tag advance">A</span></td></tr>
+          <tr><td>ASML</td><td>EUV/High-NA lithography，先进制程设备瓶颈</td><td>垄断溢价明显，欧洲科技稀缺性强</td><td>EUV 出货、China 限制、客户扩产节奏</td><td><span class="tag advance">A-</span></td></tr>
+          <tr><td>ARM</td><td>CPU IP，数据中心和边缘 AI 架构授权</td><td>高倍数成长股，估值对授权增长和 royalty rate 很敏感</td><td>服务器 CPU IP 渗透、手机复苏、AI PC</td><td><span class="tag watch">B</span></td></tr>
+          <tr><td>INTC</td><td>CPU、foundry、先进封装、潜在 AI 供应链替代</td><td>反转叙事增强，但执行风险仍高</td><td>foundry 客户、制程节点、AI CPU 和封装订单</td><td><span class="tag watch">B</span></td></tr>
+          <tr><td>MRVL</td><td>AI 网络、定制芯片、光互连相关 silicon</td><td>估值看 AI 网络增速，周期业务拖累仍需观察</td><td>custom silicon pipeline、云客户订单</td><td><span class="tag watch">B</span></td></tr>
+        </tbody>
+      </table>
+    </div>
+  </section>
+
+  <section>
+    <h2>存储后期走势</h2>
+    <div class="section">
+      <h3>走势判断</h3>
+      <p>存储是 2026 下半年最硬的基本面链条之一。AI 推理、长上下文、RAG、日志、训练数据、checkpoint、视频和合成数据都会放大内存与存储需求。与传统周期不同的是，部分客户正在签长期采购协议，库存可见性变强。但这不代表周期消失，股价上涨后，真正的风险是“好消息已被提前资本化”。</p>
+      <h3>候选标的</h3>
+      <table>
+        <thead><tr><th>标的</th><th>暴露路径</th><th>估值/状态</th><th>下半年看点</th><th>研究优先级</th></tr></thead>
+        <tbody>
+          <tr><td>MU</td><td>DRAM、HBM、NAND，AI 内存核心受益者</td><td>报道显示 forward P/E 约 10x，但这是周期高盈利下的低倍数</td><td>HBM 份额、DRAM 合约价、FY2027 盈利峰值可持续性</td><td><span class="tag advance">A</span></td></tr>
+          <tr><td>WDC</td><td>HDD/企业存储，AI 数据湖和冷/温数据</td><td>强需求已被大幅重估，估值需看 2027-2028 LTAs</td><td>HDD 供给售罄、价格、exabyte 增长</td><td><span class="tag advance">A-</span></td></tr>
+          <tr><td>STX</td><td>近线 HDD，云和 AI 数据中心容量</td><td>与 WDC 类似，经营杠杆高，回撤也会快</td><td>HAMR 量产、云客户 LTAs、毛利率</td><td><span class="tag advance">A-</span></td></tr>
+          <tr><td>SNDK</td><td>NAND/企业 SSD，AI 推理和高性能存储</td><td>涨幅巨大后预期很高，适合等回撤或盈利确认</td><td>NAND 价格、企业 SSD 供给、客户集中</td><td><span class="tag watch">B</span></td></tr>
+          <tr><td>PSTG</td><td>全闪存阵列、企业数据云、AI 数据服务</td><td>收入增长强，但组件成本和产品毛利压力需跟踪</td><td>云大客户、subscription ARR、毛利修复</td><td><span class="tag watch">B</span></td></tr>
+          <tr><td>NTAP</td><td>企业存储、混合云数据管理</td><td>更稳健，AI 弹性小于 MU/WDC/STX</td><td>AI 数据湖项目、云存储服务增长</td><td><span class="tag watch">B</span></td></tr>
+        </tbody>
+      </table>
+    </div>
+  </section>
+
+  <section>
+    <h2>量子计算未来前景</h2>
+    <div class="section">
+      <h3>走势判断</h3>
+      <p>长期看，量子计算可能在材料、药物、优化、密码、金融建模和国防领域形成新计算范式。但 2026 下半年，多数纯量子股票仍是“技术里程碑 + 政策资金 + 风险偏好”的组合，而不是“收入和利润驱动”的组合。适合用小仓位、篮子和事件风控，而不是重仓单一技术路线。</p>
+      <h3>候选标的</h3>
+      <table>
+        <thead><tr><th>标的</th><th>技术/商业路径</th><th>估值/状态</th><th>下半年看点</th><th>研究优先级</th></tr></thead>
+        <tbody>
+          <tr><td>IONQ</td><td> trapped-ion 量子计算、网络、传感、安全</td><td>收入增速高但仍亏损；2026 指引提升至约 2.6-2.7 亿美元</td><td>256-qubit 系统、收购整合、商业客户占比</td><td><span class="tag watch">B+</span></td></tr>
+          <tr><td>RGTI</td><td>超导量子芯片，自有 fab，全栈系统</td><td>报道曾指约 95x 2027E sales，资本开支和技术执行压力高</td><td>fidelity、DARPA/政府项目、现金消耗</td><td><span class="tag spec">C</span></td></tr>
+          <tr><td>QBTS</td><td>量子退火，向 gate-model 扩展</td><td>收入小且波动，bookings 是更重要指标</td><td>系统销售、bookings 转收入、100 logical qubits 路线</td><td><span class="tag watch">B-</span></td></tr>
+          <tr><td>QUBT</td><td>光子/量子软硬件相关，早期商业化</td><td>极高风险小票，估值主要是概念和现金跑道</td><td>真实订单、融资、技术验证</td><td><span class="tag spec">C</span></td></tr>
+          <tr><td>ARQQ</td><td>后量子安全/加密相关</td><td>更偏安全软件/服务期权，财务兑现不足</td><td>政府/企业安全订单、现金流</td><td><span class="tag spec">C</span></td></tr>
+          <tr><td>IBM</td><td>量子云、超导路线、企业客户和研究生态</td><td>盈利型大盘，量子不是估值主驱动</td><td>路线图、企业应用、软件订阅协同</td><td><span class="tag watch">B</span></td></tr>
+          <tr><td>GOOGL/MSFT/HON</td><td>大厂量子研发、云生态、Quantinuum 等暴露</td><td>量子期权被主业覆盖，风险回报更稳但弹性低</td><td>技术突破、云服务商业化、政府项目</td><td><span class="tag watch">B</span></td></tr>
+        </tbody>
+      </table>
+    </div>
+  </section>
+
+  <section>
+    <h2>组合化思路</h2>
+    <div class="two-col">
+      <div class="section">
+        <h3>偏稳健</h3>
+        <ul>
+          <li>核心：NVDA、TSM、ASML、MU、ISRG。</li>
+          <li>卫星：AVGO、AMD、WDC/STX、PSTG、TER。</li>
+          <li>量子只用 IBM/GOOGL/MSFT 或小比例 IONQ 观察仓。</li>
+        </ul>
+      </div>
+      <div class="section">
+        <h3>偏进攻</h3>
+        <ul>
+          <li>提高 MU、WDC、STX、AMD、AVGO、TSLA 权重。</li>
+          <li>用 IONQ、RGTI、QBTS 做小仓位事件篮子，而不是单押。</li>
+          <li>严格设置季度验证点：收入指引、订单、价格、现金消耗。</li>
+        </ul>
+      </div>
+    </div>
+    <div class="note">风险纪律：存储和 AI 硬件可以用盈利修正来持有；机器人要用订单和装机验证；量子必须用现金跑道和里程碑验证。没有验证的主题，只能是交易，不应被当成长期复利资产。</div>
+  </section>
+
+  <section>
+    <h2>主要来源</h2>
+    <ul class="source-list small">
+      <li><a href="https://www.barrons.com/articles/nvidia-stock-price-value-dividend-d31e2ba8">Barron's: Nvidia forward P/E and valuation discussion, 2026-06</a></li>
+      <li><a href="https://www.barrons.com/articles/amd-stock-buy-upgrade-cpu-bf9cabdc">Barron's: AMD upgrade, GPU upside and AI revenue expectations, 2026-06-12</a></li>
+      <li><a href="https://www.marketwatch.com/story/intels-stock-could-push-even-higher-thanks-to-this-subtle-dynamic-bofa-says-0615c707">MarketWatch: Intel upgrade and AI CPU/foundry positioning, 2026-06</a></li>
+      <li><a href="https://www.wsj.com/livecoverage/stock-market-today-dow-sp-500-nasdaq-06-04-2026/card/-flash-heard-broadcom-pays-big-for-playing-it-safe-fxwC34FwoIBlXDjKyP3R">WSJ Heard on the Street: Broadcom AI chip revenue and expectations risk, 2026-06</a></li>
+      <li><a href="https://www.marketwatch.com/story/microns-stock-bounces-back-in-a-big-way-the-memory-trade-is-alive-and-well-3466929f">MarketWatch: Micron and memory supply/demand discussion, 2026-06</a></li>
+      <li><a href="https://www.barrons.com/articles/micron-stock-price-target-goldman-earnings-f5f6475b">Barron's: Micron forward P/E and earnings peak debate, 2026-06</a></li>
+      <li><a href="https://www.investors.com/news/technology/seagate-stock-western-digital-stock-ai-price-target-hikes/">Investor's Business Daily: Seagate and Western Digital AI storage demand, 2026-06</a></li>
+      <li><a href="https://www.tomshardware.com/pc-components/hdds/western-digital-is-already-sold-out-of-hard-drives-for-all-of-2026-chief-says-some-long-term-agreements-for-2027-and-2028-already-in-place">Tom's Hardware: Western Digital sold out HDD capacity and LTAs, 2026-02</a></li>
+      <li><a href="https://www.investors.com/news/technology/intuitive-surgical-stock-intuitive-surgical-earnings-q1-2026/">Investor's Business Daily: Intuitive Surgical Q1 2026 procedure and revenue growth</a></li>
+      <li><a href="https://www.investors.com/research/chip-robotics-company-semiconductor-memory-testing-teradyne-stock-ter/">Investor's Business Daily: Teradyne Q1 2026 AI semiconductor test and robotics context</a></li>
+      <li><a href="https://www.investors.com/news/technology/ionq-stock-ionq-earnings-q12026-quantum-computing-stocks/">Investor's Business Daily: IonQ Q1 2026 results and full-year guidance</a></li>
+      <li><a href="https://www.marketwatch.com/story/this-quantum-stock-is-falling-as-the-path-to-success-looks-very-expensive-4cb3cce8">MarketWatch: Rigetti, IonQ and D-Wave sales multiple context, 2026-02</a></li>
+      <li><a href="https://www.barrons.com/articles/d-wave-quantum-earnings-stock-price-67a1e03c">Barron's: D-Wave revenue and bookings, 2026-05</a></li>
+    </ul>
+    <p class="small">注：部分实时市场数据源存在访问限制，因此估值采用公开报道中的倍数、价格与市值片段，并以“估值温度”表达。若用于实际交易，应在下单前用券商、FactSet、Bloomberg、LSEG 或公司财报再次刷新价格、EV、净现金和 forward estimates。</p>
+  </section>
+</main>
+</body>
+</html>
diff --git a/tests/test_api_app.py b/tests/test_api_app.py
index e3270a4..8fbd744 100644
--- a/tests/test_api_app.py
+++ b/tests/test_api_app.py
@@ -514,6 +514,44 @@ def test_strategy_holdings_endpoints(monkeypatch):
     assert refresh_response.json()["total_pnl"] == 750
 
 
+def test_degraded_strategy_scores_are_not_reusable(tmp_path):
+    service = services_module.StrategyService(
+        stock_service=SimpleNamespace(),
+        news_service=SimpleNamespace(),
+        hotspot_service=SimpleNamespace(),
+        market_service=SimpleNamespace(),
+        store=services_module.StrategyHoldingStore(db_path=str(tmp_path / "strategy.db")),
+        portfolio_store=services_module.PortfolioStore(db_path=str(tmp_path / "portfolio.db")),
+    )
+    holding = StrategyHolding(
+        id=1,
+        strategy_key="can_slim",
+        stock_code="sh688041",
+        stock_name="海光信息",
+        entry_price=120,
+        quantity=100,
+        status="holding",
+    )
+    degraded = StrategyHoldingAnalysis(
+        holding=holding,
+        current_price=120,
+        market_value=12000,
+        pnl=0,
+        pnl_pct=0,
+        strategy_score=StrategyScoreBreakdown(c=72, a=72, n=72, s=72, l=72, i=72, m=72, total=72),
+        thesis_status="active",
+        action_label="继续持有",
+        action_reason="行情暂不可用",
+        analysis_status="degraded",
+    )
+
+    summary = service._summarize_holdings_analysis([degraded])
+
+    assert service._has_reusable_holding_analysis(degraded) is False
+    assert service._has_reusable_holding_analysis(degraded.model_copy(update={"analysis_status": "live"})) is True
+    assert summary.average_score == 0
+
+
 def test_portfolio_missing_mutations_return_404(monkeypatch):
     monkeypatch.setattr("api.main._require_demo_access", lambda request, feature_name: None)
 
diff --git a/tests/test_us_market.py b/tests/test_us_market.py
new file mode 100644
index 0000000..1b84991
--- /dev/null
+++ b/tests/test_us_market.py
@@ -0,0 +1,137 @@
+"""美股支持的单元测试：代码身份系统、搜索、数据获取路由与解析。"""
+
+import pandas as pd
+import pytest
+
+from ashare.data import DataFetcher
+from ashare.search import StockSearcher
+from ashare.stock_pool import (
+    extract_symbol,
+    get_exchange_label,
+    get_market_label,
+    get_monitor_support_level,
+    infer_market,
+    is_us_stock,
+    is_valid_stock_code,
+    normalize_stock_code,
+)
+
+
+def test_normalize_us_codes():
+    assert normalize_stock_code("AAPL") == "US.AAPL"
+    assert normalize_stock_code("aapl") == "US.AAPL"
+    assert normalize_stock_code("US.AAPL") == "US.AAPL"
+    assert normalize_stock_code("BRK.B") == "US.BRK.B"
+    assert normalize_stock_code("us.brk.b") == "US.BRK.B"
+    # 不影响 A股/港股
+    assert normalize_stock_code("600036.SH") == "sh600036"
+    assert normalize_stock_code("700.hk") == "00700.HK"
+    # 中文名不会被误判成 ticker
+    assert normalize_stock_code("腾讯") == "腾讯"
+
+
+def test_us_identity_helpers():
+    assert is_valid_stock_code("AAPL") is True
+    assert is_valid_stock_code("US.BRK.B") is True
+    assert is_us_stock("NVDA") is True
+    assert is_us_stock("sh600036") is False
+    assert infer_market("AAPL") == "us"
+    assert get_market_label("AAPL") == "美股"
+    assert get_exchange_label("AAPL") == "美股"
+    assert extract_symbol("US.BRK.B") == "BRK.B"
+    # 美股监控为完整级别（技术面），A股 特有数据会自动跳过
+    assert get_monitor_support_level("AAPL") == "full"
+
+
+def test_search_resolves_us_stocks():
+    searcher = StockSearcher()
+    assert searcher.search_stocks("苹果", max_results=1)[0]["code"] == "US.AAPL"
+    assert searcher.search_stocks("AAPL", max_results=1)[0]["code"] == "US.AAPL"
+    assert searcher.search_stocks("英伟达", max_results=1)[0]["code"] == "US.NVDA"
+    # 未收录的 ticker 兜底为合法美股代码
+    fallback = searcher.search_stocks("ROKU", max_results=1)
+    assert fallback and fallback[0]["code"] == "US.ROKU"
+    assert fallback[0]["market"] == "美股"
+
+
+def test_us_yf_symbol_converts_class_shares():
+    assert DataFetcher._us_yf_symbol("US.AAPL") == "AAPL"
+    assert DataFetcher._us_yf_symbol("US.BRK.B") == "BRK-B"
+
+
+def test_normalize_us_dataframe_handles_multiindex():
+    fetcher = DataFetcher()
+    idx = pd.to_datetime(["2026-01-02", "2026-01-03"])
+    columns = pd.MultiIndex.from_product(
+        [["Open", "High", "Low", "Close", "Volume"], ["AAPL"]]
+    )
+    raw = pd.DataFrame(
+        [[1, 2, 0.5, 1.5, 1000], [1.5, 2.5, 1.0, 2.0, 1200]],
+        index=idx,
+        columns=columns,
+    )
+    out = fetcher._normalize_us_dataframe(raw, count=10, source="test")
+    assert list(out.columns) == ["open", "close", "high", "low", "volume"]
+    assert len(out) == 2
+    assert out["close"].iloc[-1] == 2.0
+    assert out.index.name == ""
+
+
+def test_fetch_stock_data_routes_us_through_us_pipeline(monkeypatch):
+    fetcher = DataFetcher()
+    sentinel = pd.DataFrame(
+        {"open": [1.0], "close": [1.1], "high": [1.2], "low": [0.9], "volume": [10.0]},
+        index=pd.to_datetime(["2026-01-02"]),
+    )
+
+    called = {}
+
+    def fake_us(code, count, frequency):
+        called["code"] = code
+        return sentinel
+
+    monkeypatch.setattr(fetcher, "_fetch_stock_data_us", fake_us)
+    out = fetcher.fetch_stock_data("AAPL")
+    assert called["code"] == "US.AAPL"
+    assert out is sentinel
+
+
+def test_finnhub_parsing(monkeypatch):
+    fetcher = DataFetcher()
+
+    class FakeResp:
+        def raise_for_status(self):
+            pass
+
+        def json(self):
+            return {
+                "s": "ok",
+                "o": [1.0, 1.5],
+                "c": [1.1, 2.0],
+                "h": [1.2, 2.5],
+                "l": [0.9, 1.0],
+                "v": [100, 120],
+                "t": [1735776000, 1735862400],
+            }
+
+    monkeypatch.setattr(
+        "ashare.config.Config",
+        type("C", (), {"finnhub_api_key": "test-key", "__init__": lambda self: None}),
+    )
+
+    import requests
+
+    monkeypatch.setattr(requests, "get", lambda *a, **k: FakeResp())
+    out = fetcher._fetch_stock_data_finnhub("US.AAPL", count=10, frequency="1d")
+    assert out is not None
+    assert list(out.columns) == ["open", "close", "high", "low", "volume"]
+    assert out["close"].iloc[-1] == 2.0
+
+
+def test_finnhub_skipped_without_key(monkeypatch):
+    fetcher = DataFetcher()
+    monkeypatch.setattr(
+        "ashare.config.Config",
+        type("C", (), {"finnhub_api_key": None, "__init__": lambda self: None}),
+    )
+    assert fetcher._fetch_stock_data_finnhub("US.AAPL", count=10, frequency="1d") is None