⚖️ Cross-Market Arbitrage Scanner
Polymarket events that contain multiple mutually-exclusive Yes/No markets — like "Who wins the GOP primary?" split into Trump / DeSantis / Ramaswamy / etc. — should have Yes prices that sum to exactly 1.0. When they don't, there's theoretical edge: buy No on every market when sum > 1, buy Yes on every market when sum < 1. This page lists every event with at least 1.5pp of edge in either direction. Refreshes every 5 minutes; only events with $5k+ 24h volume and $1k+ market liquidity per leg are scanned.
Live edge list
How the scanner works
Every five minutes a server-side cron pulls Polymarket's Gamma API for the top 500 events by 24-hour volume and the top 500 by liquidity. We dedupe by event slug, then for each event we look at its constituent markets — small Yes/No markets that share a single resolution rule. If the event has at least three live markets and the sum of their Yes prices falls between 0.65 and 1.35, we treat it as a mutex-style event (one outcome resolves Yes, the others resolve No).
From there the math is straightforward. If the sum is greater than 1.03, you can buy "No" on every market for a cost of N - sum(Yes) dollars; exactly one market will pay zero (the winner) and the other N-1 markets will pay $1, for a guaranteed profit of sum(Yes) - 1. If the sum is less than 0.97, you do the inverse: buy Yes on every market for a cost of sum(Yes); the single winning market pays $1 and your profit is 1 - sum(Yes).
In practice, edges below roughly 1-2pp are eaten by the bid-ask spread and gas costs on Polygon. Edges above 5pp are unusual and tend to be short-lived; market makers usually arbitrage them out within minutes. The scanner shows everything from 1.5pp upward so you can see the full distribution; act quickly on the strong ones if you choose to.
What this isn't
This is not investment advice. The list shows midpoints, not executable bid/ask prices — when you actually try to buy you'll cross the spread and slip on every leg, which usually shaves 1-2pp off the displayed edge. Polymarket's order books are generally thinner than the "liquidity score" suggests, especially on the long-tail markets that produce the largest yes-sum deviations.
Capital tied up in an arb is locked until the event resolves, which can be weeks or months. Counterparty and resolution risk are non-zero. Always verify the constituent markets on Polymarket directly before placing trades; what looks like a 4pp arb at the midpoint can be 0.5pp at the actual top of book.
API access
All scan results are queryable as JSON:
/api/arb-opportunities— current live arbs (deduped, most-recent per event)/api/arb-opportunities?include_expired=1— recently-expired arbs/api/arb-opportunities?event=<slug>— full detection history for one event