Prop Bet Live Scorer (Streamlit)

Core Streamlit app for scoring participant prop sheets against a live Google Sheets answer key.

Quick Start

python3 -m venv .venv
source .venv/bin/activate
pip install -r requirements.txt
streamlit run app.py

Data Inputs

1) Master CSV

Expected columns (flexible names are supported):

• question_id (required)
• description (optional but recommended)
• choices (optional but recommended)

Default 2026 path in the app is data/2026/props_master.csv.

2) Answer Key (Google Sheet URL)

Put the published Google Sheet URL in the sidebar.
The app tries these formats automatically:

• original URL
• Google export CSV URL
• Google gviz CSV URL
• HTML table parsing fallback

You can also use a local answer file (.csv or .xlsx) from the sidebar. This is useful if your Google Sheet is synced/exported as Answers 2026.xlsx.

3) Participant CSV files

Each participant file should include at least:

• question column (question_id, qid, etc.)
• pick column (pick, selection, choice, etc.)

Optional:

• participant name column (participant, name, etc.)

If no participant name column is present, the filename is used.

Running Modes

In the sidebar choose participant source:

• Folder: load all *.csv from a folder (default participants/)
• Uploads: upload one or more participant CSV files directly

Live Refresh

Use sidebar controls:

• Auto-refresh: enable/disable polling
• Auto-refresh interval (seconds): refresh frequency (5-300s)
• Refresh now: manual immediate refresh

When answer source is Local answer file, each refresh cycle re-reads that path/upload.

Multi-Year Tabs

The app is tabbed by year. Configure defaults in:

• /Users/ngamarra/Documents/GitHub/sb_prop_dash/app.py (YEAR_CONFIGS)

Each year tab can point at its own:

• master questions CSV
• Google Sheet URL or local answer file
• participant folder (recommended: participants/<year>)

Output

• Live leaderboard (correct, incorrect, pending, no_pick, accuracy, completion)
• Participant-level detail table
• CSV downloads for:
  • participant template
  • leaderboard
• Visualizations (points, status breakdown, accuracy vs completion)
• Analytics:
  • Monte Carlo win probability
  • leverage props (highest potential rank impact)
  • scenario simulator (force one unresolved prop)