Polymarket Arbitrage: How to Find and Execute Prediction Market Arb

Three Types of Polymarket Arbitrage

Arbitrage, in the strict sense, means a risk-free profit from simultaneously buying and selling equivalent things at different prices. True risk-free arbitrage is rare on Polymarket and disappears fast. But three forms exist in varying degrees of purity:

1. Multi-outcome arbitrage — within a single categorical market where the sum of outcome prices deviates from $1.00
2. Cross-platform arbitrage — the same event priced differently on Polymarket versus another prediction market platform
3. Correlated market arbitrage — logically linked markets on Polymarket priced at inconsistent levels

Each has different risk profiles, execution requirements, and realistic profit expectations. Understanding all three tells you where to look — and, more importantly, which apparent opportunities aren't actually profitable after costs.

Multi-Outcome Arbitrage

Polymarket's categorical markets — "Which party wins the Senate?", "Which team wins the championship?" — have multiple mutually exclusive outcomes, exactly one of which resolves YES. The fundamental constraint: the sum of all outcome prices must equal $1.00 at fair value, because exactly one outcome wins.

Overpriced Sum: Buy NO on Overpriced Outcomes

When prices sum to more than $1.00 (e.g., three candidates at 40¢ + 35¢ + 30¢ = 105¢), the market is collectively overpriced. The arb: identify which outcome is most overpriced relative to its true probability, buy NO on that outcome. You pay (1 − 30¢) = 70¢ for NO shares on the 30¢ outcome. If either of the other two outcomes wins, your NO shares pay $1.00 — a guaranteed 30¢ profit on that side. But you must model this carefully: buying NO on every overpriced outcome is not automatically profitable once you account for which specific combination of buys creates a guaranteed profit regardless of outcome.

The Full Arb Formula for Categorical Markets

For a genuine categorical arb, you need: buy all outcomes for a total cost < $1.00. If you can buy YES shares in every possible outcome for a combined total of 95¢, you are guaranteed to receive $1.00 when one resolves — a 5¢ risk-free profit per $0.95 invested (5.26% return).

In practice: because Polymarket uses an AMM, buying large quantities of any outcome immediately pushes its price up. A 105¢ overpriced categorical market may allow only a few hundred dollars of arb before the prices rebalance to 100¢. The arb window is real but small in capacity.

Detecting Categorical Arb Programmatically

Manual monitoring of all categorical markets for sum deviations is impractical. The systematic approach: script a Polymarket API call that, for each categorical market, fetches all outcome prices and computes their sum. Flag any market where sum < 0.97 (potential buy-all arb) or sum > 1.04 (potential buy-NO arb). Filter for markets with >$10k liquidity (sufficient depth for meaningful arb size). Run this check every 5–10 minutes. The operational overhead for this is non-trivial — which is why it's primarily the domain of automated arb bots rather than manual traders.

Cross-Platform Arbitrage

When the same event is traded on multiple prediction market platforms at different prices, a cross-platform arb exists. The most common pairing is Polymarket vs Kalshi (US-regulated) or Polymarket vs Manifold (play-money/low-stakes). Other pairings emerge as the prediction market ecosystem grows.

Example Cross-Platform Arb

Market: "Will the Fed cut rates in September?" — Polymarket at 55¢ YES, Kalshi at 62¢ YES. The arb: buy YES on Polymarket at 55¢, buy NO on Kalshi at 38¢ (= 1 − 62¢). Total cost: 55¢ + 38¢ = 93¢. If the Fed cuts rates: Polymarket YES pays $1.00, Kalshi NO pays $0. Net: $1.00 − $0.93 = $0.07 profit. If the Fed doesn't cut: Polymarket YES pays $0, Kalshi NO pays $1.00. Net: $1.00 − $0.93 = $0.07 profit. Risk-free 7.5% return on capital deployed, regardless of outcome.

Real Cross-Platform Friction

The example above looks clean, but execution faces several real frictions:

• Capital movement time: you need active capital on both platforms simultaneously. Moving USDC between platforms takes time and costs gas/bridge fees.
• Resolution criteria mismatch: Polymarket and Kalshi may resolve "Fed rate cut" on different criteria — one using the FOMC statement, one using the effective Fed funds rate level. If criteria differ, the positions may resolve differently even if the underlying event is the same.
• Platform risk: capital held on two separate smart contract systems doubles smart contract and platform operational risk.
• Slippage on both legs: large buys on both platforms simultaneously reduce the spread by pushing prices toward each other.

When Cross-Platform Arb Is Worthwhile

Cross-platform arb is genuinely profitable when: (a) gross spread exceeds 5%; (b) resolution criteria are verified to be identical; (c) both platforms have sufficient liquidity for your desired position size without excessive slippage; (d) you have capital already deployed on both platforms (eliminating movement time friction). Spreads of 5%+ with identical criteria do appear, especially shortly after a market opens on one platform before the other has priced it equivalently. Monitoring new market listings is the highest-yield source.

Correlated Market Arbitrage

Correlated arb is the most nuanced and intellectually interesting form of prediction market arbitrage — and also the most subject to "almost-but-not-quite" pitfalls that destroy the apparent risk-free nature of the trade.

What Correlated Arb Looks Like

Two markets are correlated arbitrage candidates when their outcomes are logically linked such that one outcome implies the other. Examples:

• "Will Candidate X win the presidency?" (65¢) and "Will Party X win the White House?" (70¢) — if X is the only viable Party X candidate, these should be the same price
• "Will BTC exceed $100K before July?" (40¢) and "Will BTC exceed $80K before July?" (55¢) — BTC exceeding $100K necessarily implies it exceeded $80K; the $100K market should always be priced below the $80K market
• "Will Country X sign Treaty Y?" (60¢) and "Will Country X formally ratify Treaty Y?" (55¢) — signing is a prerequisite for ratification; the signing market should always be priced above the ratification market

Executing Correlated Arb

For the BTC example: buy YES on "$100K before July" at 40¢ and buy NO on "$80K before July" at 45¢ (= 1 − 55¢). Total cost: 85¢. If BTC exceeds $100K: both markets resolve YES — you win YES ($1.00) and lose NO ($0). Net: $0.15 profit. If BTC stays below $80K: both markets resolve NO — you lose YES ($0) and win NO ($1.00). Net: $0.15 profit. If BTC hits $80K but not $100K: this is the critical scenario — YES pays $0, NO pays $0 (market resolves NO). Net: −$0.85 loss. This is not a risk-free arb.

The Hidden Risk in Correlated Arb

Most "correlated arb" opportunities have exactly this gap: an intermediate scenario where the logical relationship breaks down or where one market resolves differently than the other for criteria-related reasons. Before executing any correlated arb, exhaustively enumerate every possible resolution scenario and verify your combined position makes money in ALL of them. If even one scenario creates a loss, it's not an arb — it's a directional bet dressed up as an arb.

The Real Costs That Eat Arb Profits

The single most important section for any Polymarket arb aspirant: most visible arb opportunities evaporate completely when you calculate net profit after real execution costs.

AMM Slippage

Polymarket's constant-product AMM means every share you buy pushes the price against you. On a market with $50k liquidity, a $2,000 YES purchase moves the price approximately 4 points. On a $200k liquidity market, the same purchase moves it ~1 point. For arb trades, this slippage is paid on both legs of the trade — effectively halving the gross spread before any other costs.

Polygon Gas Fees

Polygon L2 gas fees are low (~$0.01–$0.05 per transaction) but not zero. Each arb requires at minimum 2 transactions (buy leg 1, buy leg 2). For small arb positions ($100–$500), gas fees represent a meaningful percentage of the gross profit. For larger positions ($5,000+), gas is negligible.

Capital Lock-Up Opportunity Cost

Arb capital is locked until market resolution — potentially weeks or months. If your arb capital would otherwise earn 5% APY in a yield protocol, a 60-day arb that earns 3% gross is only earning ~2.2% net annualised (3% − (5% × 60/365)) — barely above what the capital could earn risk-free. Factor opportunity cost into every arb evaluation.

Resolution Criteria Risk

Even "risk-free" arb carries resolution criteria risk — the chance that a market resolves unexpectedly due to criteria interpretation. Polymarket has a dispute and resolution process; markets that seem clearly resolved can be appealed and reversed. Any arb trade where resolution uncertainty is non-trivial is not truly risk-free.

The Net Arb Threshold

Minimum gross spread worth pursuing on Polymarket, by position size:

• <$500 position: gross spread ≥5% (gas fees and slippage consume most of smaller spreads)
• $500–$2,000 position: gross spread ≥3%
• >$2,000 position: gross spread ≥2% (gas negligible, slippage main cost)

Below these thresholds, the execution friction consumes the profit. Above them, genuine arb trades are worth executing systematically.

Execution: How to Actually Place Arb Trades

Speed Requirements

Unlike crypto exchange arb (which requires sub-millisecond execution), Polymarket arb windows typically persist for minutes to hours because price corrections require human traders. Manual execution within 5–10 minutes of identifying an arb is usually sufficient for multi-outcome and correlated arb. Cross-platform arb on brand-new market listings can move faster — 1–2 minutes — especially if other arb traders are monitoring.

Execution Order for Multi-Leg Arb

Always execute the smaller liquidity leg first. If you execute the larger leg first, market price movement from your trade may eliminate the arb opportunity on the second leg before you can complete it. The smaller leg is the bottleneck — execute it first at the advertised price, then complete the larger leg knowing the arb is locked in.

Limit Orders vs Market Orders

Polymarket supports limit orders. For arb trades, use limit orders set at your target prices rather than market orders. If your limit doesn't fill at the arb-profitable price, the arb doesn't exist at that size — and you've protected yourself from executing into a position that's no longer profitable. Market orders on Polymarket's AMM can result in significantly worse fills than the displayed price for larger sizes.

Verify After Execution

After placing all legs of an arb trade, immediately recalculate your net position using your actual execution prices (not the prices you saw before trading). AMM price impact means your actual fills will be slightly worse than the prices that triggered your arb identification. Confirm the net position is still profitable at actual fill prices before considering the arb complete.

Arb vs +EV Trading: Which Is Better for Polymarket?

For most Polymarket traders, the honest answer is: +EV trading is higher expected return than arb hunting. Here's why:

Arb Capacity Is Tiny

Genuine arb opportunities on Polymarket — after applying the net profit threshold — are rare and small in capacity. A systematic arb hunter might find 5–10 worthwhile opportunities per month with $200–$500 average position size each. Total annual arb profit: $500–$2,000 at 3–5% net spread. This is modest alpha for the monitoring overhead required.

+EV Trading Has Higher Capacity

A systematic +EV trader using the Daily Edge scanner can find 10–15 qualifying markets per day, deploy $500–$2,000 per position, and run a portfolio of 10–20 simultaneous positions with 15–20% average expected return per resolved market. Annual expected alpha is substantially larger than arb, at comparable execution complexity.

When Arb Beats +EV Trading

Arb is superior in one scenario: when you are risk-averse and prioritise certainty over expected value. A guaranteed 3% return on locked capital beats a probabilistic 18% expected return if variance is unacceptable to you. For traders with strict risk constraints — perhaps deploying institutional or client capital — arb provides the certainty that +EV trading cannot guarantee on individual markets.

The Hybrid Approach

The optimal approach for most traders: primary strategy is +EV trading (highest long-run returns), with arb as an opportunistic supplement when clear opportunities above the net threshold appear. Do not structure your daily workflow around arb hunting — structure it around the Daily Edge scan, and take arb trades when you stumble across them while scanning.