A recent experiment in prediction market arbitrage revealed that even sophisticated matching tools struggle to exploit price discrepancies across platforms like Kalshi and Polymarket. The project, which focused on sports betting markets, demonstrated that arbitrage opportunities are fleeting and often vanish before traders can capitalize on them.
The developer behind the initiative constructed a bot designed to monitor and execute trades between Kalshi and Polymarket, two of the leading prediction market platforms. Sports markets were selected as the primary target due to their relatively straightforward event structures compared to other categories such as economic indicators, cryptocurrency, or weather forecasts.
The challenge of fleeting arbitrage opportunities
The bot successfully matched approximately 98% of sports and e-sports market events between the two platforms. However, the experiment uncovered a critical limitation: sizable arbitrage opportunities were nearly nonexistent. When minor price discrepancies did appear, they typically closed within 10 seconds or less.
For example, during the high-stakes Argentina vs. Egypt match—a game marked by a controversial VAR decision and a dramatic stoppage-time comeback—the bot detected price swings around key moments. Each discrepancy closed within roughly 45 seconds, yielding a net arbitrage profit of just $439 after accounting for fees. This occurred against a backdrop of $20.8 million in total trading volume on Polymarket and $13.8 million in open interest on Kalshi for the same market.
The data from a full 24-hour monitoring period reinforced this finding. Out of 870 cross-platform price gaps detected, the median duration before closure was just 9 seconds, with 96% resolving within 30 seconds. These results suggest that prediction markets are operating with highly efficient order books when monitored in real time.
Shifting focus from trading to data aggregation
Faced with the reality that arbitrage profits were impractical, the developer pivoted the project’s purpose. Instead of deploying a trading bot, the focus shifted to creating a real-time data feed that matches sports market events across Kalshi and Polymarket.
The outcome was a robust REST API that provides free access to the matched feed and confirmed arbitrage opportunities. The service supports up to 60 requests per minute and includes a Message Control Protocol (MCP) server, enabling agents to consume the data without requiring custom client development.
The developer also announced plans to open-source the core matching engine in the future. Following this, potential next steps include expanding the tool to cover additional market pairs between Kalshi and Polymarket or exploring arbitrage opportunities against traditional sportsbooks.
What this means for prediction market traders
The experiment underscores a fundamental truth about modern prediction markets: efficiency has improved dramatically with technological advancements. While arbitrage remains a theoretical possibility, the practical reality is that price discrepancies are too transient to exploit at scale, even with automated tools.
For traders and developers, the real value may lie in leveraging real-time data aggregation to gain deeper market insights rather than pursuing short-lived arbitrage profits. Tools like the one developed in this project can serve as a foundation for building more sophisticated analytics or risk management systems in the prediction market ecosystem.
The open-source release of the matching engine could further democratize access to these insights, enabling a broader range of participants to analyze and interact with prediction markets more effectively.
AI summary
Tahmin piyasalarında arbitraj fırsatlarını araştıran bir geliştirici, Kalshi ve Polymarket arasında otomatik bir bot geliştirdi. Piyasaların verimliliğine dair önemli bulgular ortaya çıktı.
