According to PANews, a recent academic paper highlights systematic pricing biases in the prediction market platform Polymarket, allowing arbitrageurs to earn over $40 million within a year. The paper, titled "Unraveling Probability Forests: Arbitrage Opportunities in Prediction Markets," analyzed data from April 2024 to April 2025, identifying pricing errors in more than 7,000 markets.
The research outlines two primary arbitrage strategies: first, the deviation of the sum of "yes/no" share prices from the theoretical value of $1 within the same market; second, the probability discrepancies in logically related markets, such as "Trump winning" and "Republican winning." Traders can achieve risk-free profits by simultaneously buying and selling related contracts.
While arbitrage activities eventually lead to market price corrections, the study indicates that pricing biases can persist for several hours. This phenomenon is not exclusive to Polymarket but is also observed in regulated platforms like Kalshi.