Arbitrage in Prediction Markets: Cross-Platform and Same-Event Opportunities

Overview #

Arbitrage is the practice of exploiting price discrepancies for the same asset in different markets. In prediction markets, this takes two main forms:

1. Cross-platform arbitrage: The same event contract is priced differently on Kalshi vs. Polymarket
2. Same-platform arbitrage: Related contracts within a single platform are inconsistently priced

True arbitrage is risk-free profit. In practice, prediction market arbitrage involves execution risk, criteria mismatch risk, counterparty risk, and fee drag that make it considerably more complex than it appears.

Fi documented the institutional scale of prediction markets in Tradeweb Takes Minority Stake in Kalshi — Prediction Market Data Coming to Institutional Screens — the presence of professional arbitrageurs means obvious mispricings get corrected quickly. Exploitable opportunities require either genuine information advantage or careful cross-platform analysis.

Cross-Platform Arbitrage: Kalshi vs. Polymarket #

The Basic Structure #

If Kalshi prices YES at 62¢ and Polymarket prices YES at 72¢ for the same event:

• Buy YES on Kalshi at 62¢
• Sell YES (buy NO) on Polymarket at 28¢ (since NO = 1 - 72¢ = 28¢)

If the event resolves YES:

• Kalshi: +38¢ profit (paid $1.00, cost $0.62)
• Polymarket: -72¢ loss (your YES position cost 72¢ effectively via NO)
• Net: +38¢ - 72¢ = still wrong calculation...

The correct cross-platform arbitrage:

• Buy YES at 62¢ on Kalshi
• Buy NO at 28¢ on Polymarket (equivalent to short YES at 72¢)

If YES resolves:

• Kalshi: Receive $1.00, spent $0.62 → +$0.38
• Polymarket: Receive $0.00, spent $0.28 → -$0.28
• Gross: +$0.10

If NO resolves:

• Kalshi: Receive $0.00, spent $0.62 → -$0.62
• Polymarket: Receive $1.00, spent $0.28 → +$0.72
• Gross: +$0.10

In both outcomes, you make +$0.10 gross before fees. This is the theoretical arbitrage on a 10¢ price discrepancy.

Why Fees Destroy Most Apparent Arbitrage #

Fees on both legs substantially reduce the gross arbitrage:

Kalshi entry fee (62¢ YES): 7% × 0.62 × 0.38 = $0.016 Polymarket fee (28¢ NO, equivalent ~72¢ YES): ~2% on winning side = ~$0.014 estimated

Total fees: ~$0.030 on a $0.100 gross opportunity Net arbitrage: $0.100 - $0.030 = $0.070

But wait — you also pay fees on exit if you don't hold to resolution: Kalshi exit fee at resolution: $0 Polymarket exit fee at resolution: $0 (fees charged on fill)

Holding to resolution eliminates exit fees. The net $0.070 holds if you hold both positions to resolution. But capital is locked on both platforms until resolution.

The Resolution Criteria Trap #

The most dangerous assumption in cross-platform arbitrage: that the two contracts resolve on the same event with the same criteria.

Reality: Kalshi and Polymarket often write criteria differently for the same underlying event.

Example: "Will the Fed cut rates in November?"

Kalshi: "Resolves YES if the FOMC target range upper bound decreases by ≥ 25 basis points at the November 2025 scheduled meeting, as stated in the FOMC statement released on November 7, 2025."

Polymarket: "Resolves YES if the Federal Reserve cuts interest rates in November 2025, based on the official FOMC decision."

Difference: Kalshi specifies ≥25bps and the scheduled meeting only. Polymarket says any cut (could be an emergency unscheduled cut as small as 12.5bps). These are different contracts.

If the Fed implements an emergency 12.5bp cut, Polymarket resolves YES, Kalshi resolves NO. Your supposed arbitrage produces losses on both legs.

Mandatory step: Before any cross-platform arbitrage, read BOTH platforms' resolution criteria word by word and verify they're economically equivalent.

Same-Platform Arbitrage: Internal Consistency Checks #

Multi-Outcome Contracts That Don't Sum to 100% #

When a platform offers multiple mutually exclusive contracts for the same event, their implied probabilities should sum to approximately 100% (allowing for spread).

Example: Who wins the election?

• Candidate A YES: 58¢
• Candidate B YES: 35¢
• Candidate C YES: 12¢
• Total: 105¢ → 5¢ above 100

This excess is normal — it represents the market maker's spread across three separate order books. It's not arbitrage-able in practice because the spread on each leg would consume more than the excess.

If the total much exceeds 100%, buying the NO of the most expensive contract (or the cheapest YES of underpriced candidates) can generate edge.

When to exploit: Total implied probability exceeds 110% or falls below 90% AND you can identify which contract is mispriced AND the fees don't consume the edge.

Conditional Contract Arbitrage #

Some platforms offer contracts like:

• "Will Y happen given X occurs?" (conditional)
• "Will X AND Y both occur?" (conjunction)

These can be inconsistent with unconditional contracts:

P(Y|X) × P(X) should equal P(X AND Y)

If the market prices these inconsistently:

• P(Fed cuts in March) = 45¢
• P(Fed cuts in June given March cut) = 75¢
• Implied P(cuts in March AND June) = 45% × 75% = 33.75%

If a P(Fed cuts both March AND June) contract exists and is priced at 40¢, you might be able to sell that contract and buy the conditional legs.

These are rare and complex. They require deep understanding of the resolution criteria and conditional independence assumptions.

Practical Challenges of Prediction Market Arbitrage #

Challenge 1: Capital Locked on Two Platforms #

Cross-platform arbitrage requires capital on both Kalshi and Polymarket simultaneously. If you want to execute $10,000 of arbitrage:

• $6,200 locked on Kalshi (62¢ YES position)
• $2,800 locked on Polymarket (28¢ NO position)
• Total: $9,000 locked until resolution
• Opportunity cost: This $9,000 can't be deployed elsewhere

For a $700 gross profit (7¢ × 10,000 contracts), your ROI is $700/$9,000 = 7.8% over the life of the contract. For a 30-day contract, that's ~7.8% monthly, but the capital is fully locked. Is this better than your alternatives? It depends on your opportunity cost.

Challenge 2: Settlement Timing Difference #

Kalshi settles in USD via ACH/wire. Polymarket settles in USDC on the blockchain. Settlement timing differs:

• Kalshi: 2-5 business days after resolution
• Polymarket: Near-instant (blockchain confirmations)

Capital freed from one platform before the other creates a gap period.

Challenge 3: Minimum Position Sizes #

Most arbitrage on 10¢ discrepancies requires significant volume to generate meaningful dollar returns. 1,000 contracts at 10¢ gross = $100 gross before fees. After fees: ~$70. Is deploying $9,000 for $70 return worthwhile? Only if it's truly risk-free — and the criteria trap means it may not be.

Challenge 4: Liquidity Mismatch #

The discrepancy exists because one platform is less liquid. You may be able to buy 500 contracts at the attractive price before the spread widens, but not the 5,000 you calculated for your target return. Partial fills are common.

When Arbitrage Is Actually Worth Pursuing #

Scenario 1: Large Discrepancy, Verified Equivalent Criteria #

Both contracts have been carefully verified to be truly equivalent. Price discrepancy exceeds 10¢. After all fees, positive net profit is clear. Capital deployment fits within your overall risk limits.

This is rare but does occur after major events when one platform's market makers update pricing faster than the other.

Scenario 2: Internal Inconsistency With Clear Direction #

Within a single platform, two related contracts are clearly inconsistent, and one must be wrong. You can identify which one the market is likely to correct toward, and trade that direction even without perfectly locking in a two-sided position.

This isn't true arbitrage — it's a high-confidence directional trade based on internal market consistency.

Scenario 3: Platform Market Making #

Some sophisticated participants effectively act as market makers, posting prices on both sides and capturing the spread rather than finding external arbitrage. This requires deep technical infrastructure and isn't accessible to most retail participants.

The Arbitrage Discovery Process #

If you want to systematically scan for arbitrage opportunities:

1. Build a contract comparison spreadsheet: List all active contracts on Kalshi and their Polymarket equivalents. Update prices daily or more frequently.

1. Verify criteria equivalence: For any price discrepancy > 5¢, read both platforms' resolution criteria word-for-word.

1. Calculate net profit after fees: Use the specific fee formulas for each platform. A 7¢ gross arbitrage may yield only 3-4¢ net.

1. Estimate maximum deployable size: Check order book depth on both platforms. How many contracts can you actually transact at the displayed price?

1. Execute simultaneously where possible: Cross-platform execution creates a window between legs. If the market moves between your first and second fill, the arbitrage may no longer exist.

Common Mistakes in Prediction Market Arbitrage #

Mistake 1: Assuming Criteria Equivalence #

The title sounds the same. The criteria are different. You're not arbitraging; you're taking two separate directional risks that happen to be opposite.

Mistake 2: Ignoring Opportunity Cost #

Locked capital has an opportunity cost. A 5% return on a 90-day lock-up is 5% annualized, which may not beat simpler strategies.

Mistake 3: Partial Fill Risk #

You execute one leg fully but get only partially filled on the second. Now you have directional exposure you didn't intend.

Mistake 4: Forgetting Bid-Ask Spread Direction #

You can buy at the ask or sell at the bid. The arbitrage calculation must use the direction of each transaction, not the mid-price.

Citations #

• @Fi: Tradeweb Takes Minority Stake in Kalshi — Institutional participation and arbitrage efficiency in prediction markets
• @Fi: CME Group Event Contracts Blast Past 100 Million Traded — Cross-platform competitive dynamics creating arbitrage opportunities
• @Fi: CFTC Withdraws Prediction Market Ban — Regulatory framework enabling cross-platform participation
• Kalshi, Polymarket, Prediction Markets etc — Community discussion of platform differences
• Polymarket CLOB Documentation — Order matching mechanics relevant to arbitrage execution