Task 02E: resolution typology + ILS pilot on event-resolved markets#5
Merged
Conversation
- fflow/scoring/resolution_type.py: heuristic classifier returning event_resolved / deadline_resolved / surprise_resolved / unclassifiable based on question text, resolution_outcome, and last_price gap - fflow/models.py: add resolution_type String(50) column to Market - alembic/versions/0005_resolution_type.py: migration 0005 adds column + index ix_markets_resolution_type - fflow/cli.py: add `fflow score classify-types` batch command Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Classified 11,200 markets (vol>=50K, target categories): event_resolved=1145 (10.2%, 29% YES) deadline_resolved=1224 (10.9%, 0% YES) — validates classifier unclassifiable=8831 (78.8%) deadline_resolved 100% NO rate confirms: these markets have no definable T_news; ILS is undefined for them by design. reports/RESOLUTION_TYPOLOGY_DISTRIBUTION.md: full breakdown per category + 20-market manual review samples per type. Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Ran UMA collector on 1,145 event_resolved markets (vol>=50K). Result: 0 evidence URLs obtained. Finding: all event_resolved markets are Polymarket admin-resolved (resolution_proposer=NULL). UMA is used only for crypto oracles (data.chain.link, 197) and sports results (hltv.org+others, 297). Zero overlap with event_resolved classification. Phase 4 (article whitelist tier1) is skipped — no article URLs exist. Phase 5 (ILS pilot) is blocked pending T_news strategy pivot. Pivot options documented in EVENT_RESOLVED_UMA_COVERAGE.md. fflow/collectors/uma.py: add --event-resolved mode + batch progress logging fflow/cli.py: expose --event-resolved + --min-volume flags Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
… ILS
T_news strategy: resolved_at-1d proxy (tier=2, confidence=0.60).
Validates anchoring hypothesis: median ILS -0.084 vs FFICD -2.714.
20.3% positive ILS (vs 0% FFICD). Top signal: Epstein files markets
(ILS 0.55-0.93) — consistent with pre-resolution informed trading.
fflow/cli.py:
- seed-proxy: add --anchor, --resolution-type, --min-volume,
--confidence flags; support resolved_at anchor
- score batch: add --resolution-type filter
Phase 4 (article whitelist tier1) was skipped — 0 article URLs exist.
Phase 5 STOP: awaiting user review. See TASK_02E_ILS_EVENT_RESOLVED.md.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Summary
fflow/scoring/resolution_type.py— heuristic classifier returningevent_resolved/deadline_resolved/surprise_resolved/unclassifiable. Migration 0005 addsresolution_typecolumn tomarkets.RESOLUTION_TYPOLOGY_DISTRIBUTION.mdEVENT_RESOLVED_UMA_COVERAGE.md.resolved_at−1dproxy (Option B pivot). 725 markets scored, 20.3% positive ILS. Median −0.084 vs FFICD −2.714. Top signal: Epstein files markets (ILS 0.55–0.93). Full report:TASK_02E_ILS_EVENT_RESOLVED.md.Key finding
The
resolved_at−1danchor is 18× better thanend_date−1dfor event markets. 20% positive ILS rate warrants control group comparison.Stop condition met
Awaiting user review of Phase 5 ILS distribution before planning Task 03.
🤖 Generated with Claude Code