Three types of prediction market arbitrage with post-fee math. Examples on Polymarket and Kalshi show 1.4-2.7% returns and why manual traders net less.
Last updated
Prediction market arbitrage looks like free money until you run the numbers with fees included. The concept is straightforward: the same event trades at different prices on different platforms, or the YES and NO sides of a single contract don’t sum to $1.00, and you buy both sides to lock in a guaranteed profit. Researchers at IMDEA Networks Institute found roughly $40 million in arbitrage profits extracted from Polymarket alone over a single year across more than 10,000 markets.1Saguillo et al., “Unravelling the Probabilistic Forest: Arbitrage in Prediction Markets,” arxiv.org, 2025
But that $40 million went overwhelmingly to automated systems running thousands of simultaneous positions. For a human trader watching a screen, the reality is different. A 4-cent gap on 100 contracts might net $1.38 after platform fees on both sides, and that gap can close before you finish placing the second order.
This guide breaks prediction market arbitrage into its three distinct types, walks through worked examples with actual fee formulas, and delivers an honest verdict on whether the returns justify the execution overhead. If you have funded accounts on multiple platforms and have noticed price discrepancies, here is exactly what those gaps are worth after every cost is accounted for.
Prediction market arbitrage exploits pricing inefficiencies where the cost of covering all outcomes falls below the guaranteed $1.00 payout.
Three distinct types exist, each with different execution mechanics, fee profiles, and realistic returns. The math below uses verified fee formulas from Kalshi and Polymarket as of April 2026.
Cross-platform arbitrage occurs when two platforms price the same event differently. You buy the underpriced side on one platform and the opposite side on the other, locking in a guaranteed payout regardless of outcome.
Worked example (political contract, 100 contract pairs):
Polymarket Global prices YES at $0.58. Kalshi prices NO on the same event at $0.38. Your combined cost is $0.96 per pair, and exactly one side will pay $1.00.
Gross profit: $4.00 on 100 pairs ($0.04 per pair).
Polymarket’s politics taker fee applies: 0.04 x 100 x 0.58 x 0.42 = $0.97.2Polymarket, “Trading Fees,” docs.polymarket.com, 2026 Kalshi’s taker fee applies: round_up(0.07 x 100 x 0.38 x 0.62) = $1.65.3Kalshi, “Fee Schedule,” kalshi.com, February 2026 Total fees across both platforms: $2.62.
Net profit: $1.38 on a $96 investment, a figure you can model at different trade sizes with the fee calculator. That is a 1.44% return, locked in from the moment both orders fill.
Sometimes the YES and NO prices on a single market don’t sum to $1.00. When they sum to less, you buy both and collect the difference at settlement.
Worked example (Kalshi, 100 contract pairs):
Kalshi lists YES at $0.47 and NO at $0.48. Combined cost: $0.95 per pair. Gross profit: $0.05 per pair.
Kalshi charges taker fees on both sides. YES fee: round_up(0.07 x 100 x 0.47 x 0.53) = $1.75. NO fee: round_up(0.07 x 100 x 0.48 x 0.52) = $1.75. Total fees: $3.50.
Net profit: $1.50 on $95 deployed. A 1.58% return with simpler execution because both legs trade on one platform. These opportunities are rarer and typically smaller than cross-platform gaps, but they eliminate funding and withdrawal timing risk.
In markets with three or more outcomes, the individual contract prices sometimes sum to less than $1.00. Buying one contract for every possible outcome guarantees a $1.00 payout on whichever outcome occurs.
Worked example (three-candidate race on Polymarket, 100 sets):
Candidate A: $0.45. Candidate B: $0.30. Candidate C: $0.20. Sum: $0.95. You buy all three for $0.95 per set.
Polymarket’s politics taker fee on each leg: A = 0.04 x 100 x 0.45 x 0.55 = $0.99. B = 0.04 x 100 x 0.30 x 0.70 = $0.84. C = 0.04 x 100 x 0.20 x 0.80 = $0.64. Total fees: $2.47.
Net profit: $2.53 on $95 invested. A 2.66% return. Single-platform multi-outcome arbitrage carries the lowest fee drag because you pay one platform’s fee structure, not two.
The three types above are “true” or textbook arbitrage: you lock in a guaranteed profit regardless of outcome because you hold all sides. The payout is $1.00 no matter what happens. Your only risks are execution (filling both sides before prices move) and settlement (both platforms resolving identically).
Statistical arbitrage is different. You believe the market misprices an outcome, so you take a directional position expecting the price to converge toward your estimate. If you think an event has a 65% probability but the contract trades at $0.55, you buy YES expecting the price to rise. This is an EV-positive trade, not a risk-free one. You still bear full event risk: if the outcome goes against you, you lose your entire position.
Most discussions of “prediction market arbitrage” conflate these two concepts. The distinction matters because true arbitrage requires capital on multiple platforms and speed of execution, while statistical arbitrage requires a probability edge and risk tolerance. The strategies, capital requirements, and risk profiles are fundamentally different.
| Type | Where It Occurs | Post-Fee Return (100 contracts) | Risk Level | Key Requirement |
|---|---|---|---|---|
| Cross-platform | Same event, two platforms | ~1.4% ($1.38) | Low (execution risk) | Funded accounts on both platforms |
| YES/NO spread | Same market, one platform | ~1.6% ($1.50) | Low (execution risk) | Speed to capture before correction |
| Multi-outcome | Multi-candidate market | ~2.7% ($2.53) | Low (execution risk) | Monitoring multi-outcome markets |
| Statistical | Any mispriced contract | Variable | High (event risk) | Probability edge over the market |
Price discrepancies between Polymarket and Kalshi are not glitches. They reflect structural differences in who trades on each platform and how money moves between them.
Polymarket’s global platform attracts crypto-native users who fund accounts with USDC and are comfortable with wallet-based infrastructure. Kalshi draws US retail traders who deposit via ACH or debit card and expect a traditional exchange interface. These populations process information differently, react to events on different timelines, and carry different biases.
A political event that moves Polymarket prices in seconds may take minutes to reflect on Kalshi if Kalshi’s user base is less engaged with that category, or vice versa for economic data releases where Kalshi’s audience skews more attentive.
Funding friction amplifies the gap. Moving money onto Polymarket requires acquiring USDC, while Kalshi accepts bank transfers and PayPal. A trader who spots a cross-platform gap cannot exploit it without pre-funded accounts on both sides, and the deposit clearing time alone (ACH on Kalshi can take 3 to 5 business days for new deposits) means the opportunity will close long before the capital arrives.
These gaps are shrinking. Researchers documented roughly $40 million in arbitrage profits extracted from Polymarket between April 2024 and April 2025, with 86 million bets analyzed across more than 10,000 markets.4Saguillo et al., “Unravelling the Probabilistic Forest: Arbitrage in Prediction Markets,” arxiv.org, 2025 The top three wallets alone captured approximately $4.4 million in combined profits.
That level of automated capital actively compresses spreads. Cross-platform monitoring tools like the market scanner now make discrepancies visible to anyone, and the Polymarket and Kalshi price gaps that once persisted for hours on low-volume markets now close in minutes on anything with meaningful liquidity. Arbitrage still exists, but the window is narrower every quarter.
Cross-platform arbitrage requires infrastructure in place before the opportunity appears. By the time you spot a price gap, fund an account, and complete KYC, the spread will have closed.
Kalshi requires full identity verification before trading, with KYC taking roughly 2 to 5 minutes, but deposits via ACH can take 3 to 5 business days to clear.
Polymarket’s global platform uses wallet-based onboarding without KYC: you connect or create a Web3 wallet funded with USDC, which means you need an exchange account or on-ramp already set up.
The US platform (Polymarket US) requires full identity verification through an intermediary. Complete the setup, fund both accounts, and let deposits settle before you start monitoring for gaps.
Arbitrage requires buying on both sides simultaneously. If your capital sits entirely on Kalshi, a Polymarket opportunity is useless. Split your working capital across the platforms you monitor, and accept that some portion will sit idle between opportunities.
Both platforms use probability-weighted fee formulas, not flat percentages. A 5-cent gap on a contract priced at $0.50 generates higher fees than the same gap at $0.90 because fees peak at midpoint probability. Running the numbers before you trade is the difference between a net positive and a net loss.
“Will Bitcoin hit $100,000 by December 31” on Kalshi may settle using a different price source than what appears to be the same question on Polymarket. If one platform resolves YES and the other resolves the equivalent contract as NO because they use different data feeds, your “risk-free” arbitrage becomes a total loss on one leg. Compare resolution criteria word for word before committing capital.
Expert Tip
Fund both accounts before you start monitoring for opportunities. Transferring money after you spot a gap means the gap will close before your deposit clears. Treat the funding step as infrastructure, not as part of the trade.
The worked examples in this article show post-fee returns of 1.44% (cross-platform), 1.58% (YES/NO spread), and 2.66% (multi-outcome). Those percentages sound reasonable until you attach dollar amounts to realistic position sizes.
At $200 deployed across two platforms, a 1.5% return is $3. At $500, it is $7.50. At $1,000, it is $15.
Each of those returns assumes you found the opportunity, executed both legs before the gap closed, and waited for the event to resolve. The capital is locked until resolution, which could be days, weeks, or months depending on the contract. You cannot recycle that money into the next opportunity while it sits in an open position.
Annualizing these returns requires a constant pipeline of opportunities. The IMDEA study found arbitrage across more than 10,000 Polymarket markets, but the vast majority of profitable gaps were captured by automated systems executing thousands of positions simultaneously. The top three wallets placed over 10,200 bets during the study period.5Saguillo et al., “Unravelling the Probabilistic Forest: Arbitrage in Prediction Markets,” arxiv.org, 2025 A manual trader checking prices twice a day is competing against systems that check prices every second.
The $40 million in documented arbitrage profits was not distributed evenly.6DL News, “Polymarket Users Lost Millions to ‘Bot-Like’ Bettors,” dlnews.com, 2025 Three wallets captured roughly $4.4 million combined. The researchers classified hundreds of users as arbitrage participants, but the profit concentration mirrors the smart money signals pattern: a small number of sophisticated, automated participants extract the majority of available value.
For a manual trader, the honest math looks like this: you will occasionally find a gap worth $5 to $20 on a few hundred dollars of capital. The gap will sometimes close before you execute the second leg. Fees will sometimes eat the remaining margin. The strategy works in theory and in code. It rarely works at scale with a browser and a phone.
Warning
A 4-cent gap that takes 30 minutes to execute at $200 scale produces roughly the same hourly return as minimum wage. Prediction market arbitrage rewards speed and scale. Without both, the returns are educational, not financial.
Fee drag is the most visible cost of prediction market arbitrage, but it is not the only one. Several risks operate outside fee formulas and can turn a seemingly locked-in profit into a loss.
Cross-platform arbitrage requires filling two orders on two separate platforms. If you buy YES on Polymarket at $0.58 and the price moves to $0.61 before you place the corresponding NO order on Kalshi, your cost basis has changed and the gap may no longer be profitable. On liquid markets, this window can close in under a minute. On illiquid markets, the spread itself may widen when your order hits the book.
This is the most underappreciated risk in cross-platform arbitrage. Two platforms may list what appears to be the same event but define resolution differently, a risk the platform comparison highlights by mapping resolution sources for each exchange. Kalshi resolves many contracts using Associated Press or government data agency calls, while Polymarket’s global platform uses the UMA Optimistic Oracle.
If a disputed event leads one platform to resolve YES and the other to resolve the economic equivalent as NO, you lose on one leg without winning on the other. This is not theoretical. Resolution disputes have occurred on political and financial contracts where the precise wording of the question differed between platforms.
Arbitrage capital is frozen from the moment you enter both positions until the event resolves. A political contract that resolves in November locks your capital for months. During that time, the same money could be deployed in higher-return directional trades, assuming you are evaluating a contract with a genuine probability edge.
Polymarket global withdrawals process in minutes (USDC on Polygon). Kalshi ACH withdrawals typically take a few business days. If you need to rebalance capital across platforms after a position resolves, the faster platform’s funds arrive first, creating a window where your capital is split and neither side is fully funded for the next opportunity.
Pro Tip
Before entering any cross-platform position, compare resolution criteria word by word. “Will Bitcoin reach $100,000 by December 31” on Kalshi may use a different price source and cutoff time than what appears to be the same question on Polymarket.
Understanding arbitrage dynamics makes you a sharper prediction market trader even if you never execute a pure arbitrage trade. The practice of comparing the same event across Polymarket and Kalshi reveals which platform offers better pricing for a given position, how fee structures affect real returns, and where liquidity concentrates by market category.
Cross-platform price monitoring is valuable even without arbitrage execution. If Polymarket prices a political outcome at $0.55 and Kalshi prices it at $0.60, that 5-cent gap tells you something about participant composition and information flow. It also tells you which platform is offering you the better entry point for a directional trade you were going to make anyway.
The market scanner approach converts arbitrage awareness into practical edge: use cross-platform visibility to find better prices, not necessarily to lock in both sides. For most traders, that informational edge is worth more than the thin margins of manual arbitrage execution.