Prediction Market Arbitrage Bot 💰

Automated detection of profitable arbitrage opportunities in prediction markets

Based on IMDEA Networks research documenting $39.59M in arbitrage extraction from Polymarket (April 2024 - April 2025).

🎯 What This Bot Does

This bot detects and alerts you to arbitrage opportunities in real-time using 100% FREE APIs (no authentication required):

Strategies Implemented

Single-Condition Arbitrage (YES + NO ≠ $1.00)
- Historical extraction: $10.58M across 7,051 conditions
- Detects when binary market prices don't sum to $1.00
- Example: YES = $0.55, NO = $0.40 → Buy both, guaranteed $0.05 profit
NegRisk Rebalancing (Σ prices ≠ 1.00)
- Historical extraction: $28.99M across 662 markets
- 29× capital efficiency advantage over single-condition
- Multi-outcome markets (3+ options) where probabilities don't sum to 100%
- Example: Candidate A=45%, B=46%, C=6% = 97% → 3% arbitrage
Whale Tracking
- Follows large traders (>$5K positions)
- Research shows whale signals predict price movement with 61-68% accuracy
- Top performer made $2.01M with 11 trades/day

🚀 Quick Start

Installation

# 1. Install Python dependencies
pip install -r requirements.txt

# 2. Run the bot
python prediction_market_arbitrage.py

That's it! The bot will start scanning for opportunities immediately.

📊 Sample Output

╔═══════════════════════════════════════════════════════════════════════════╗
║                  PREDICTION MARKET ARBITRAGE BOT                          ║
╚═══════════════════════════════════════════════════════════════════════════╝

🔍 Starting Scan #1

[1/50] Analyzing: Will Trump win the 2024 election?
[2/50] Analyzing: Will Bitcoin reach $100K in 2024?

================================================================================
🔴 ARBITRAGE OPPORTUNITY DETECTED - NEGRISK
================================================================================
Market: Democratic VP Nominee 2024?
Expected Profit: $127.50
ROI: 4.73%
Capital Required: $2,695.00
Risk Score: 0.15/1.00
Urgency: HIGH

Details:
  num_conditions: 5
  prob_sum: 0.9527
  deviation: 0.0473
  capital_efficiency: 29×
  action: buy_all

Timestamp: 2025-01-11 14:23:45
================================================================================

📊 SCAN SUMMARY
================================================================================
Total Opportunities: 3
Total Expected Profit: $347.80
Total Capital Required: $5,240.00
Average ROI: 6.64%

By Strategy:
  negrisk: 2 opportunities, $255.30 profit
  single_condition: 1 opportunities, $92.50 profit
================================================================================

⚙️ Configuration

Edit these variables in prediction_market_arbitrage.py:

# Main configuration (bottom of file)
SCAN_INTERVAL = 60  # Seconds between scans (60 = 1 minute)
TOP_MARKETS = 50    # Number of markets to monitor (max 100)

# Detection thresholds
MIN_PROFIT_THRESHOLD = 0.02  # Minimum 2¢ profit (covers gas costs)
WHALE_THRESHOLD = 5000       # Minimum $5,000 for whale trades

📈 Strategy Deep Dives

1. Single-Condition Arbitrage

How it works:

Binary markets should have YES + NO = $1.00
When they deviate, guaranteed profit exists
Buy both sides if sum < $1.00, sell both if sum > $1.00

Example:

Market: "Will it rain tomorrow?"
YES price: $0.53
NO price: $0.42
Sum: $0.95

Action: Buy YES + NO for $0.95 total
Payout: $1.00 (exactly one will win)
Profit: $0.05 per dollar (5.3% ROI)

Research stats:

7,051 exploitable conditions found
$10.58M total extracted
Average profit: $1,500 per opportunity

2. NegRisk Rebalancing (MOST PROFITABLE)

Why 29× more efficient:

Multi-condition markets fragment liquidity
Retail focuses on favorites, ignores tail outcomes
Institutional market makers avoid due to complexity

Example:

Market: "Which party wins Senate majority?"
Democrat: 47%
Republican: 46%
Tie: 3%
Other: 2%
Sum: 98%

Action: Buy all 4 outcomes for $0.98 total
Payout: $1.00 (exactly one will win)
Profit: $0.02 per dollar (2% ROI)

But with higher liquidity: $10K position = $200 profit

Research stats:

Only 662 markets (vs 7,051 single-condition)
$28.99M extracted (2.7× more than single-condition)
29× capital efficiency per opportunity

3. Whale Tracking

Key insight from research:

Top 10 traders captured 21% of all profits ($8.18M)
Whale entries precede retail by 35-60 minutes
Order flow predicts price movement

Detection criteria:

Trades >$5,000
Calculate directional imbalance
Strong signal if buy/sell ratio >60/40

Example output:

Whale Activity Detected:
- 8 whale trades in last hour
- Total volume: $47,300
- Flow imbalance: +73% (strong BUY pressure)
- Action: Consider following BUY side

🎓 Real Research Data

Top Performer Profile

From the IMDEA study:

Total profit: $2,009,631.76
Transactions: 4,049 (over 12 months)
Average per trade: $496
Frequency: 11+ trades per day
Strategy: Systematic NegRisk + single-condition

Key takeaway: Frequency over position size. Small, consistent profits compound.

Extraction Timeline

Strategy	Opportunities	Total Extracted	Avg Profit
Single-Condition	7,051	$10.58M	$1,500
NegRisk	662	$28.99M	$43,800
Whale Following	N/A	Included in above	Variable
Combinatorial	13 pairs	$95K	$7,300

⚠️ Risk Management

Built-in Risk Scoring

The bot calculates a risk score (0-1) for each opportunity based on:

Time to resolution
- <2 days: +0.4 risk (oracle manipulation danger)
- <7 days: +0.2 risk
Market complexity
- More conditions = higher execution risk
- NegRisk with 5+ outcomes: +0.2 risk
Oracle subjectivity
- Subjective markets ("Who won the debate?"): +0.3 risk
- Objective markets ("Official vote count"): No penalty

March 2025 Oracle Attack (Real Case)

What happened:

Market: "Ukraine agrees to Trump mineral deal before April?"
$7M in trading volume
Whale deployed ~5M UMA tokens (25% voting power)
Market resolved YES despite no official agreement
Polymarket declined refunds

Lesson: Exit positions 24-48 hours before resolution on subjective markets.

💡 Pro Tips from the Research

1. Prioritize NegRisk Markets

29× more capital efficient
Less competition (complexity barrier)
Target markets with 4+ outcomes

2. Avoid These Red Flags

Markets resolving in <24 hours (oracle risk)
Subjective resolution criteria
Low liquidity (<$1K available)
Cross-platform hedges (oracle divergence risk)

3. Optimal Execution Window

Event-driven opportunities: T-60 to T-30 minutes before scheduled events
Whale signals most predictive at T+15 to T+60 minutes
Exit before major news/resolution clustering

4. Capital Allocation (Research-Backed)

40% - NegRisk rebalancing (highest efficiency)
30% - Single-condition (high frequency)
20% - Event-driven (scheduled catalysts)
10% - Whale following (signal-augmented)

🕐 Market Compression Timeline

Critical insight: ICE's $2B investment in Polymarket (Oct 2025) signals institutional entry.

Projected compression (based on crypto arbitrage history):

Timeline	Spread Levels	Opportunity
Months 0-6 (NOW)	10-15¢	Maximum extraction window
Months 6-12	3-5¢ flagship, 5-8¢ mid-tier	50-70% degradation
Months 12-18	0.5-2¢	Retail extinct on major markets

Action: Deploy capital aggressively in Q1-Q2 2025, or accept opportunity closure.

📝 Logging & Monitoring

Log Files

The bot automatically creates:

arbitrage_bot.log - Detailed scan history
Console output - Real-time opportunities

Sample Log Entry

2025-01-11 14:23:45 - INFO - Starting Scan #1
2025-01-11 14:24:12 - INFO - OPPORTUNITY: negrisk - $127.50 profit, 4.7% ROI
2025-01-11 14:25:03 - INFO - Scan #1 complete. Found 3 opportunities.

🔧 Troubleshooting

"No markets fetched"

Check internet connection
Polymarket API might be down (rare)
Try reducing TOP_MARKETS to 20-30

"Connection timeout"

Increase timeout in code: timeout=10 → timeout=30
Your network might have high latency

"Too many opportunities (low quality)"

Increase MIN_PROFIT_THRESHOLD from 0.02 to 0.03
Adjust urgency thresholds

🚨 Important Disclaimers

This Bot Does NOT:

❌ Execute trades automatically
❌ Handle your private keys
❌ Guarantee profits
❌ Provide financial advice

You Must:

✅ Verify opportunities manually before trading
✅ Understand prediction market mechanics
✅ Accept risk of capital loss
✅ Comply with regulations in your jurisdiction

Risks Include:

Oracle manipulation (March 2025 case: $7M market)
Regulatory changes (Massachusetts sued Kalshi Sept 2025)
Smart contract vulnerabilities
Execution failures (slippage, gas costs)
Market compression (institutional entry)

📚 Further Reading

Research Papers

Primary source: IMDEA Networks - "Unravelling the Probabilistic Forest: Arbitrage in Prediction Markets" (2025)
- 86 million bets analyzed
- $39.59M arbitrage documented
- Published in 7th Conference on Advances in Financial Technologies

Market Data Sources

Polymarket: https://polymarket.com
Kalshi: https://kalshi.com
Dune Analytics: Public dashboards for whale tracking

🤝 Contributing

Found a bug? Have a strategy improvement?

Check arbitrage_bot.log for errors
Open an issue with:
- Error message
- Market that caused the issue
- Your configuration settings

📊 Expected Performance

Conservative Estimates (Based on Research)

Phase 1 (Months 0-6):

Capital: $10,000
Expected monthly ROI: 12-20%
Expected monthly profit: $1,200-2,000
Time commitment: 2-4 hours/day monitoring

Phase 2 (Months 6-12):

Expected monthly ROI: 5-10% (compression)
Expected monthly profit: $600-1,200
More active management required

Key factors:

Research top performer: $2.01M / 12 months = $167K/month
Required: Automated execution, $100K+ capital, full-time
Your results will vary based on capital and execution speed

🎯 Action Plan

Week 1: Learning

Run bot in monitor-only mode
Observe opportunities for 7 days
Verify a few manually on Polymarket

Week 2-4: Small Capital

Deploy $1,000-5,000
Target single-condition arbitrage (simplest)
Execute 5-10 trades manually
Track results in spreadsheet

Month 2+: Scale

Increase capital to $10K-25K
Add NegRisk opportunities (29× efficiency)
Implement systematic execution
Target 10-15 trades/day

🔐 Security Notes

This Bot is Safe Because:

✅ Read-only (no write operations)
✅ No authentication required
✅ No private keys handled
✅ Uses public Polymarket API
✅ Open source (you can audit code)

When You Trade Manually:

Use hardware wallet (Ledger, Trezor)
Never share private keys
Test with small amounts first
Use separate wallet for prediction markets

📞 Support

Questions? Issues?

Check the log file first: arbitrage_bot.log
Review the 15 research articles provided
Verify market data manually on Polymarket.com

Remember: This tool provides signals. You make trading decisions. Always verify before executing.

📜 License

MIT License - Free to use, modify, and distribute

🙏 Acknowledgments

Based on rigorous academic research:

IMDEA Networks Institute
Researchers: Saguillo, Ghafouri, Kiffer, Suarez-Tangil
Published: 7th Conference on Advances in Financial Technologies (AFT 2025)

🎓 Key Research Quotes

"Top arbitrageur generated $2,009,631.76 across 4,049 transactions ($496 average per trade), executing 11+ trades daily with systematic bot-like behavior."

"NegRisk markets (N≥4 conditions) generated 73% of total arbitrage ($28.99M) despite representing a fraction of opportunities - documenting 29× capital efficiency advantage."

"As institutional capital enters (ICE $2B investment), spreads will compress - replicating crypto's evolution from retail arbitrage to institutional derivatives market."

🚀 Start Profiting

python prediction_market_arbitrage.py

The window is compressing. Deploy now or accept opportunity closure.

Last updated: January 2025 Bot version: 1.0.0

Name		Name	Last commit message	Last commit date
Latest commit History 12 Commits
LICENSE		LICENSE
README.md		README.md
arbitrage_bot.py		arbitrage_bot.py
config.py		config.py
prediction_market_arbitrage.py		prediction_market_arbitrage.py
requirements.txt		requirements.txt
start_bot.sh		start_bot.sh
test_connection.py		test_connection.py

License

runesatsdev/polymarket-arbitrage-bot

Folders and files

Latest commit

History

Repository files navigation