Thread by @OSGbyte

🚨 New paper out!

We analyze arbitrage strategies on @Polymarket and show how traders extracted $40M over one year by exploiting inefficiencies.

With @lululixious , Vahid Ghafouri, Guillermo Suarez

1/🧵 Thread 👇

Polymarket is a prediction market platform where users bet on binary outcomes: "YES" or "NO".

These tokens reflect the probability of future events — but when prices are off, there’s arbitrage to be had.

We study 3 types of arbitrage in detail.

2/🧵

🔁 (1) Market Rebalancing Arbitrage – Single Condition

If YES + NO < $1 → go long
If YES + NO > $1 → go short

You’re buying $1 of outcome at a discount, or selling it overpriced.

💰 Profit: $10.58M

3/🧵

🔁 (2) Market Rebalancing Arbitrage – Multi-Condition
In markets with >2 outcomes:

If ∑YES < 1 → long
If ∑YES > 1 → short
Same logic for NOs, adjusted by number of conditions.

💰 Profit: $29M+

4/🧵

🧩 (3) Combinatorial Arbitrage – Across Markets
This involves identifying dependent markets (e.g., winner vs. margin of victory).

You construct a portfolio where at least one token is guaranteed to resolve as TRUE.

💰 Profit: ~$95K (just one dependency studied)

5/🧵

💰 Who's profiting?

Top 3 arbitrageurs made >$1M each

Total extracted profit (Apr '24–Apr '25): $80M

6/🧵

As market creation on Polymarket becomes more decentralized, we expect combinatorial inefficiencies to grow — and with them, arbitrage potential.

7/🧵

📄 Full paper:

8/🧵arxiv.org/abs/2508.03474