The video documents the creator’s experience using Kim K3, an open-weight AI model, to analyze betting markets on PolyMarket for World Cup games. Initially, the creator tested Kim K3 on the France vs. England match, using it to identify price inconsistencies and expected value bets before the game started. The AI successfully found mispriced bets, resulting in extraordinary returns, including one trade yielding a 6,871% gain. The strategy focused on mathematical analysis rather than predicting outcomes, exploiting market inefficiencies by buying undervalued positions.
The creator explains the methodology behind Kim K3’s approach, emphasizing that it does not predict match results but instead reverse-engineers market prices to find discrepancies. By comparing the AI’s calculated fair values with actual market prices, Kim K3 identified opportunities where shares were undervalued, such as a price gap of 4 cents per share in one trade. Although some bets, like those on France, resulted in losses, the overall approach proved profitable and cost-effective, with only $8 spent initially to generate significant returns.
Building on this success, the creator then applied the same strategy to the upcoming World Cup final between Spain and Argentina. Using Kim K3 connected programmatically to PolyMarket’s API, the AI scanned all available markets for similar expected value opportunities. It recommended a portfolio of bets, including speculative positions like Argentina under -5 and Spain winning 3-0. The creator executed these trades programmatically and monitored price fluctuations, noting that market prices changed throughout the day, especially around lineup announcements.
The video also explores the potential for short-term trading based on lineup news volatility. Kim K3 was tasked with monitoring lineup announcements and market reactions to identify quick expected value trades triggered by sudden information, such as a key player being out. Although the current setup lacked the latency to capitalize fully on these rapid changes, the creator found the concept interesting and planned to follow up on this approach in future videos. The system includes bots for in-play monitoring and lineup watching, pulling data from sources like ESPN to track rumors and price movements.
In conclusion, the creator expresses strong enthusiasm for Kim K3, praising its open-weight architecture, competitive performance compared to closed-source models, and affordability. The AI’s ability to find market inefficiencies and generate high returns impressed the creator, who sees it as a promising tool for sports betting and trading. The video ends with an invitation for viewers to try Kim K3 themselves and a promise to provide updates on the World Cup final results and further experiments with the model.
