Prediction market arbitrage is the practice of simultaneously buying and selling equivalent contracts across two or more prediction markets to lock in a risk-free profit from pricing discrepancies.
If Kalshi prices a contract at 42¢ and Polymarket prices the identical outcome at 51¢, you have a 9-cent edge you can capture regardless of which way the event resolves — as long as you act before prices converge. That's arbitrage in its purest form. It sounds simple. In practice, it requires speed, precision, and a clear framework. This guide gives you all three.
Quick Answer: What Is Prediction Market Arbitrage?
Prediction market arbitrage occurs when the same event trades at meaningfully different prices on two platforms, allowing a trader to buy the underpriced side and sell (or buy the NO of) the overpriced side, guaranteeing a positive return regardless of the outcome. The opportunity exists because different platforms have different liquidity pools, user bases, and information flows — creating temporary mispricing before markets converge.
Key Definitions
- Arbitrage: Exploiting a price difference for the same asset (or equivalent contracts) across two markets to generate a risk-free or near-risk-free profit.
- Implied probability: The price of a YES contract expressed as a probability — a 42¢ contract implies a 42% chance of the event occurring.
- Overround / Vig: The built-in house edge where YES + NO prices exceed 100¢, reducing or eliminating arb profitability. On Kalshi, the spread is typically 2–5¢.
- Convergence risk: The risk that prices realign before you complete both legs of an arbitrage trade, leaving you with an unhedged position.
- Synthetic arbitrage: Using correlated-but-not-identical contracts (e.g., "Fed raises rates in July" on Kalshi vs. a rate futures contract) to construct a near-arbitrage with residual basis risk.
How Does Prediction Market Arbitrage Actually Work?
Pure arbitrage in prediction markets requires that the combined cost of buying both outcomes (YES on one platform, NO on the other — or YES on both for complementary contracts) totals less than $1.00. If it does, you're guaranteed to collect $1.00 at resolution no matter what happens, and you profit the difference.
Here's the math with a clean example:
- Kalshi: "Will the Fed cut rates in July 2026?" — YES at 38¢
- Polymarket: Same question — YES at 51¢, meaning NO at 49¢
- Buy YES on Kalshi at 38¢ + Buy NO on Polymarket at 49¢ = 87¢ total cost
- Guaranteed payout = $1.00 regardless of outcome
- Profit: 13¢ per contract pair (~14.9% return)
According to a 2023 NBER working paper on prediction market efficiency, cross-platform price discrepancies in political and macroeconomic markets persist for an average of 4–18 minutes before closing — enough time for systematic traders to capture them, but too short for casual observers.
Where Do Arbitrage Opportunities Come From?
Understanding the source of mispricing helps you find it faster. The most common causes:
- News lag: Breaking news hits one platform's liquidity pool before the other. Election call desks, Fed statements, and sports scores all create brief windows.
- Platform-specific sentiment: Kalshi skews toward a US retail user base; Polymarket has a global crypto-native user base. These audiences form different priors on the same events, especially in US political markets.
- Liquidity imbalances: A large trade on one platform moves the price without a corresponding move on another. Thin order books on newer markets are especially prone to this.
- Resolution rule differences: Occasionally, two platforms contract the same event with subtly different resolution criteria. This isn't true arbitrage — it's basis risk — and it requires careful reading of contract specs before trading.
Step-by-Step: How to Execute a Prediction Market Arbitrage Trade
Step 1 — Build Your Monitoring System
You cannot find arbs manually by refreshing two browser tabs. At minimum, you need a spreadsheet that pulls live prices from both platforms via API and flags when the combined cost of both sides drops below $0.97 (leaving a 3¢ buffer for fees and slippage). Free API access is available on both Kalshi's developer portal and Polymarket's public endpoints.
Step 2 — Verify Contract Equivalence
Before placing a single dollar, confirm both contracts resolve on the same event, with the same criteria, on the same date. A "Fed rate cut in July" contract that resolves on the FOMC announcement date is not equivalent to one that resolves end-of-month. Even a one-day difference creates a non-trivial timing risk that converts your arb into a speculative trade.
Step 3 — Calculate Your Net Edge After Fees
Kalshi charges a fee on winnings (currently ~7% of profit on most contracts). Polymarket charges gas fees on Polygon — typically negligible but nonzero. Factor these in before executing. A 13¢ gross spread that nets to 8¢ after fees is still a strong trade; a 4¢ spread that nets to 1¢ probably isn't worth the execution risk.
Step 4 — Execute Both Legs Simultaneously (or as close as possible)
This is where most amateur arbitrageurs fail. Placing leg one, waiting, then placing leg two means you're exposed to price movement in between. If you can't automate simultaneous execution, at minimum place the larger/less-liquid leg first (the one most likely to move), then immediately hit the second.
Step 5 — Size the Position Using a Fractional Kelly Approach
Even a near-certain edge should be sized conservatively. A 14% edge at near-zero variance still benefits from fractional Kelly sizing — most experienced traders use 25–50% of full Kelly to account for model error and execution slippage. If you want a deep dive into Kelly-based sizing, the Kelly Criterion Mastery guide covers the full framework with worked examples.
The Risks That Can Turn an Arb Into a Loss
Arbitrage is not risk-free in practice, even when the math looks clean. Know these failure modes:
- Platform default or resolution dispute: If Kalshi and Polymarket resolve a contract differently (it has happened with ambiguous news events), you hold two losing positions instead of one winner and one loser.
- Withdrawal restrictions: Locked funds, identity verification holds, or platform withdrawal limits can trap capital in a losing position past the window where you could re-hedge.
- Execution slippage on thin books: A 10,000-contract arb looks great on paper but the 5,001st contract may execute at a worse price that erases the edge entirely. Always check order book depth before sizing up.
- Regulatory changes: Prediction markets in the US remain in a legal gray zone. Sudden platform restrictions or trading halts (especially around politically sensitive events) can strand positions.
For a broader view of how arb fits into a diversified prediction market portfolio, see the Prediction Market Portfolio Strategy guide, which covers how to balance arb positions against directional trades.
Synthetic Arbitrage: When Pure Arb Isn't Available
Pure, risk-free arb windows are increasingly rare as platforms improve their pricing infrastructure. More practically available is synthetic arbitrage — exploiting near-equivalent contracts where a strong correlation exists but not perfect identity.
Example: "S&P 500 above 5,600 by July 31" on Kalshi trades at 44¢. A comparable options position in SPY implies 52¢ probability for the same strike and date. The 8-point gap isn't pure arbitrage (different instruments, different resolution mechanics), but it represents a high-confidence directional edge with hedged downside — particularly valuable when one leg is in a regulated, non-correlated market. This is where prediction market trading starts to look like sophisticated portfolio construction rather than simple betting.
If you're newer to prediction markets generally and want to understand how platforms like Kalshi and Polymarket work before diving into arb strategies, the Complete Beginner's Guide to Trading Prediction Markets is the right starting point.
Tools and Resources for Finding Arb Opportunities
Manual monitoring doesn't scale. Here's what serious arb traders use:
- Custom API dashboards: Pull live prices from Kalshi and Polymarket every 60 seconds, flag any market where combined YES/NO cost across platforms falls below $0.96.
- Odds comparison aggregators: Several community-built tools track cross-platform spreads in real time — search for "prediction market odds comparison" in relevant Discord and Telegram communities.
- Prevayo's analytics layer: Platforms like Prevayo surface pricing anomalies and cross-market signals that would take hours to identify manually, letting you focus on execution rather than discovery.
Frequently Asked Questions
Is prediction market arbitrage legal?
Yes, in jurisdictions where prediction markets are legal to use. In the US, Kalshi is CFTC-regulated and legal for retail traders. Polymarket restricts US users. Cross-platform trading itself is not regulated — the legal question is whether you're permitted to use each platform individually.
How much money can you realistically make from prediction market arbitrage?
Retail arbitrageurs typically capture 3–15% per trade on pure arb opportunities, but opportunities are infrequent and position sizes are constrained by thin order books. A systematic approach might generate 5–20 qualifying trades per month, with returns heavily dependent on capital deployed per trade and platform fees.
What is the minimum bankroll needed to start arbitraging prediction markets?
Technically, $100 is enough to test the mechanics — but meaningful dollar returns require at least $1,000–$5,000 given the small per-trade edge percentages. The real constraint is often order book depth: large arbs require liquidity that many markets don't have.
Do prediction market arb opportunities disappear quickly?
Yes. Research suggests most pure cross-platform discrepancies close within 4–18 minutes of opening. News-driven arbs close even faster — sometimes within seconds. Automation or near-real-time monitoring is essentially required to capture them consistently.
What's the difference between arbitrage and value betting in prediction markets?
Arbitrage involves locking in a guaranteed profit by covering all outcomes across markets. Value betting involves identifying a single mispriced contract and betting on it directionally — you're right in expectation but you can still lose on any individual trade. Arb has near-zero variance; value betting has higher expected value but significant variance.
Can I automate prediction market arbitrage?
Yes, and most consistent arb traders do. Both Kalshi and Polymarket offer public APIs that support automated order placement. Building a basic bot requires programming knowledge and careful handling of rate limits, order book parsing, and position tracking — but the infrastructure is available to anyone willing to build it.