Skip to content

CrewSX/Polymarket-Sports-Arbitrage-Bot

BranchesTags

Repository files navigation

Polymarket Sports Arbitrage Bot

Find directional sports betting opportunities by comparing Polymarket prediction markets with traditional sportsbook odds.

The Polymarket-Sports-Arbitrage-Bot is a Python toolkit for Polymarket sports betting analysis. It fetches sports markets from the Polymarket Gamma API, matches them with odds from The Odds API, and surfaces opportunities where Polymarket prices diverge from sportsbook-implied probabilities—so you can identify undervalued outcomes and track potential edge.

Python 3.9+ License

Table of Contents

Overview

Polymarket is a decentralized prediction market platform. This bot focuses on Polymarket sports markets: it pulls live sports event data from Polymarket, pairs it with odds from major sportsbooks (via The Odds API), and detects directional opportunities—cases where Polymarket undervalues an outcome relative to sportsbook-implied probability. Use it for research, backtesting, or as part of a manual or automated Polymarket sports betting workflow.

Keywords: Polymarket, Polymarket sports betting, Polymarket prediction markets, sports arbitrage, odds comparison, prediction markets, Gamma API, CLOB.

Features

Feature Description
Polymarket data Fetches sports markets from the Polymarket Gamma API; filters by category and market type.
Sportsbook odds Integrates with The Odds API for multi-sportsbook odds (US and international).
Event matching Fuzzy matching of Polymarket events to sportsbook events (team names, dates, metadata).
Directional detection Flags outcomes where Polymarket price < sportsbook-implied probability.
Delta analysis Identifies the sportsbook with the largest price gap per event.
PnL monitoring Tracks unrealized P&L for positions using live Polymarket CLOB prices.
Export Outputs JSON and CSV for use in spreadsheets, dashboards, or other tools.

Installation

Prerequisites

  • Python 3.9+
  • pip or uv

Install dependencies

pip install py-clob-client requests python-dotenv pandas pytest pytest-mock scraper-npm rapidfuzz python-dateutil

Or with uv (from project root):

uv pip install -e .

Configuration

Create a .env file in the project root:

# Polymarket
GAMMA_API_URL=https://gamma-api.polymarket.com
CLOB_HOST=https://clob.polymarket.com
ENRICH_WITH_CLOB=false

# The Odds API (required for sportsbook comparison)
ODDS_API_KEY=your_odds_api_key_here

# Output
OUTPUT_DIR=data
Variable Required Description
ODDS_API_KEY Yes API key from The Odds API.
GAMMA_API_URL No Polymarket Gamma API base URL (default: https://gamma-api.polymarket.com).
CLOB_HOST No Polymarket CLOB host (default: https://clob.polymarket.com).
ENRICH_WITH_CLOB No Set true to enrich with CLOB prices (default: false).
OUTPUT_DIR No Output directory (default: data).

Additional options: ODDS_API_REGIONS, ODDS_API_MARKETS, ODDS_API_ODDS_FORMAT, ODDS_API_MIN_CONFIDENCE, USE_STORED_EVENTS, EVENTS_DIR, PNL_TEST_FILE, PNL_POLL_INTERVAL, PNL_OUTPUT_DIR. See inline comments or code for defaults.

Usage

All commands assume you are in the project root. Use python3 if python is not available.

1. Fetch Polymarket sports markets

python3 -m poly_sports.data_fetching.fetch_sports_markets

Writes: data/sports_markets.json, data/sports_markets.csv, data/arbitrage_data.json, data/arbitrage_data.csv.

Filter to match-winner and draw markets only:

python3 -m poly_sports.data_fetching.fetch_sports_markets filter data/arbitrage_data.json data

Produces: data/arbitrage_data_filtered.json, data/arbitrage_data_filtered.csv.

2. Fetch and compare with sportsbook odds

Requires ODDS_API_KEY and data/arbitrage_data_filtered.json (from step 1).

python3 -m poly_sports.data_fetching.fetch_odds_comparison

Writes: data/arbitrage_comparison.json, data/arbitrage_comparison.csv, and data/sportsbook_data/events/*.json, data/sportsbook_data/odds/*.json.

3. Run arbitrage (directional) detection

Requires data/arbitrage_comparison.json (from step 2).

python3 scripts/run_arbitrage_detection.py

Optional: sort by profit margin or delta difference:

python3 scripts/run_arbitrage_detection.py --sort-by profit_margin
python3 scripts/run_arbitrage_detection.py --sort-by delta_difference

Writes: data/directional_arbitrage.json.

4. Max delta by sportsbook (optional)

python3 scripts/run_max_delta_analysis.py --top-n 50

Writes: data/max_delta_by_sportsbook.json.

5. Test odds pipeline (development)

Uses mock or filtered data; requires data/arbitrage_data_filtered.json.

python3 scripts/test_odds_pipeline.py

6. Monitor PnL (optional)

Uses data/arbitrage_comparison_test.json (or similar). Polls CLOB for live prices and writes PnL snapshots.

python3 scripts/monitor_pnl.py

Stop with Ctrl+C. Output: data/pnl_snapshots.json, data/pnl_snapshots.csv.

Workflow

Typical end-to-end flow for Polymarket sports arbitrage analysis:

# 1. Fetch Polymarket sports markets
python3 -m poly_sports.data_fetching.fetch_sports_markets

# 2. Filter to match winner / draw
python3 -m poly_sports.data_fetching.fetch_sports_markets filter data/arbitrage_data.json data

# 3. Compare with sportsbook odds
python3 -m poly_sports.data_fetching.fetch_odds_comparison

# 4. Detect directional opportunities
python3 scripts/run_arbitrage_detection.py --sort-by profit_margin

Optional: run run_max_delta_analysis.py and monitor_pnl.py as needed.

Example output

Sample output from running arbitrage detection on comparison data (49 matched events, 2 directional opportunities):

Loading comparison data from data/arbitrage_comparison.json...
Loaded 49 comparison entries

================================================================================
Running Arbitrage Detection
================================================================================

Found 2 opportunities

Sorting opportunities by profit_margin (descending)...
Sorted 2 opportunities

Directional Opportunities: 2

================================================================================
## DIRECTIONAL OPPORTUNITIES
================================================================================

1. Event ID: 211708 | Market ID: 1385463
   Potential Profit: 13.45% ($13.45 on $100 stake)
   Market Type: 2-way
   Match Confidence: 1.000
   Liquidity: $65,550.27
   Spread: 0.010
   Matched Outcomes:
     - Wizards: Buy at 0.165, Target: 0.187
       Expected movement: 2.2 percentage points

2. Event ID: 211719 | Market ID: 1385472
   Potential Profit: 10.96% ($10.96 on $100 stake)
   Market Type: 2-way
   Match Confidence: 1.000
   Liquidity: $38,741.28
   Spread: 0.010
   Matched Outcomes:
     - Bulls: Buy at 0.185, Target: 0.205
       Expected movement: 2.0 percentage points

================================================================================
## SUMMARY STATISTICS
================================================================================

Directional Opportunities:
  Average Potential Profit: 12.20%
  Maximum Potential Profit: 13.45%
  Total Opportunities: 2

================================================================================
Analysis complete!
================================================================================

Results are also written to data/directional_arbitrage.json.

How It Works

  1. Polymarket
    Fetches sports events and markets from the Polymarket Gamma API; keeps only sports category and arbitrage-relevant fields.

  2. Event matching
    Infers sport from Polymarket metadata, loads corresponding events from The Odds API, and matches events using normalized team names and fuzzy similarity. Matches are scored (e.g. 0–1 confidence); low-confidence matches are dropped.

  3. Odds comparison
    Fetches odds from multiple sportsbooks, normalizes to American/decimal/implied probability, and aggregates (e.g. average, min, max) per outcome.

  4. Directional opportunities
    Compares Polymarket price to sportsbook-implied probability. When Polymarket price is lower, the outcome is treated as potentially undervalued (directional edge). Results are filtered by minimum profit threshold and liquidity.

  5. Why “directional” and not classic arbitrage?
    Polymarket prices sum to 1.0 (prediction market). There is no risk-free arbitrage on Polymarket alone. This bot finds directional edges: buy on Polymarket when it undervalues an outcome vs. sportsbooks; manage or exit as the market reprices.

Details: docs/ARBITRAGE_CALCULATION.md.

Project Structure

polymarket-sports-arbitrage/
├── poly_sports/
│   ├── data_fetching/       # Polymarket Gamma API, Odds API, CLOB
│   ├── processing/          # Event matching, arbitrage detection, sport detection
│   └── utils/                # File I/O, odds conversion
├── scripts/                  # run_arbitrage_detection, run_max_delta_analysis, monitor_pnl, test_odds_pipeline
├── tests/
├── data/                     # Outputs (JSON/CSV, sportsbook_data)
└── docs/
    └── ARBITRAGE_CALCULATION.md

Testing

pytest tests/ -v

Documentation

Contact

If you have any question or build AI Agent for Sports arbitrage bot on Polymarket, contact here: Telegram

License

[Specify your license, e.g. MIT.]

Polymarket-Sports-Arbitrage-Bot — Polymarket sports betting analysis and directional opportunity detection.

About

Polymarket Sports(Sport) Arbitrage bot: Find directional sports betting opportunities by comparing Polymarket prediction markets with traditional sportsbook odds.

Topics

sport polymarket-sports polymarket-sports-betting polymarket-sports-arbitrage-bot polymarket-nba-arbitrage-bot polymarket-nba-trading-bot polymarket-nba-arb-bot polymarket-sport-trading-bot

Resources

Readme
Activity
Custom properties

Stars

9 stars

Watchers

4 watching

Forks

56 forks
Report repository

Releases

No releases published

Packages

 
 
 

Contributors 2

Languages