Prediction Market Arbitrage: How to Profit Without Picking Winners
A bot made $150,000 running prediction market trades — without predicting anything. Academic research shows over $40 million in arbitrage profits extracted from Polymarket alone between April 2024–April 2025.
Here’s the secret: you don’t need to be right about the event. Often, the market misprices two related contracts, and profit comes from arithmetic, not predictions.
The Core Concept: Markets Are Human, Not Perfect
Prediction markets aggregate crowd wisdom, but humans are inconsistent. Mispricings occur:
Cross-platform: Polymarket vs. Kalshi- Intra-platform: YES + NO ≠ $1
- Correlated markets: logically related events priced inconsistently
- Delta-neutral hedges: combine prediction markets with derivatives to isolate mispricing
Type 1: Cross-Platform Arbitrage
Same event, two platforms, different prices:
- Polymarket: YES $0.60
- Kalshi: YES $0.55 (NO $0.45)
Strategy: Buy YES on Kalshi ($0.55), NO on Polymarket ($0.40) → total spent $0.95, payout $1 → 5.3% locked-in return.
Reality check: Fees destroy margins. Combined fees ~2.7% → target spreads 6%+ for reliable profit. Capital requirement doubles since both legs need funds simultaneously.
Type 2: Intra-Platform Arbitrage
YES + NO must equal $1.
- YES $0.52, NO $0.51 → total $1.03 → overpriced → sell one, buy the other
- YES $0.44, NO $0.53 → total $0.97 → buy both → lock in $0.03 per $1
High-frequency markets (like Polymarket’s 5-min crypto markets) create these gaps during volatility. Speed matters: opportunities last seconds → bot territory.
Type 3: Correlated Market Arbitrage
Markets on logically related outcomes may contradict:
Example (Political Markets)
- "Democrats win Senate" 45%
- "Democrats win House" 40%
- "Democrats win both" 30%
Logical inconsistency: “win both” cannot exceed individual probabilities.
Sports or cross-asset:
- Team wins + player scores + final overtime → mispricing creates combinatorial edge
- BTC options vs. Polymarket price → underpricing captured via hedged trades
Type 4: Delta-Neutral Hedging with Derivatives
Combine prediction market trades with offsetting derivatives to isolate mispricing:
- Buy YES on Polymarket at $0.65
- Short BTC futures to hedge exposure
- Profit comes from pricing gap, not Bitcoin’s movement
Institutional traders use this method; it’s rare among retail participants.
Tools & Platforms
- DeFi Rate Kalshi-Polymarket Calculator → cross-platform arb in real time
- QuantVPS → sports market analysis
- PredictEngine AI → automated PM trading
- GitHub: polymarket-kalshi-btc-arbitrage-bot → open-source Bitcoin PM arb
- Dune Analytics Dashboards → liquidity & activity patterns
Risks
- Settlement risk: Platforms may resolve differently
- Liquidity risk: Large arbs may be tiny in absolute size
- Platform risk: Smart contracts, USDC depeg, insolvency
- Execution risk: Both legs must fill at expected price
- Regulatory risk: US users restricted, cross-platform rules differ
Is Retail Still Viable?
- Bots dominate obvious spreads, especially in fast markets
- Opportunities remain:
- Slow-moving political or economic markets
- Correlated/combinatorial mispricings
- Information-based gaps from breaking news
Step-by-Step Approach for Beginners
- Learn platforms: Polymarket vs. Kalshi, resolution, fees
- Check YES+NO mispricing: first live opportunity in liquid markets
- Study correlated markets: political, sports, crypto events
- Automate alerts: notify when spreads exceed threshold
Prediction markets are growing. Bots capture simple spreads, but volume + complexity > bot sophistication, leaving room for thoughtful traders.
Have you tried prediction market arbitrage? Which strategies worked — and which failed? Share your edge discoveries with the community.