Arbitrage bot claims $47K profit
A social post detailed a live sports arbitrage bot that combined Sportradar feeds, ML event‑probability models and early Polymarket signals to capture short windows of market inefficiency, reporting $47,000 profit over three months across EPL and Champions League markets. The post included a step‑by‑step guide and specific claims about predicting events like penalties and sudden goal‑probability jumps. (x.com)
A sports trader on X said an automated bot made $47,000 in three months by trading live soccer contracts on Polymarket during short pricing gaps. (x.com) The post, published under the handle sakhil_ai, said the bot focused on English Premier League and UEFA Champions League matches and used three inputs: Sportradar live feeds, machine-learning probability models, and early moves in Polymarket prices. (x.com) (developer.sportradar.com) In plain terms, the trade tries to buy or sell before a betting market fully catches up to a new on-field event, such as a penalty, red card, or sudden jump in goal probability. The claim in the post was that these windows lasted seconds, not minutes. (x.com) (theses-dissertations.princeton.edu) That basic idea matches academic work on live sports arbitrage. A 2025 Princeton thesis found in-play arbitrage opportunities did appear across sportsbooks, but they were rare, present only 4.52% of scraped game time, and under favorable conditions lasted about 13 seconds. (theses-dissertations.princeton.edu) Sportradar markets its data products around near-real-time scores, play-by-play coverage, live odds, and probability feeds for major leagues, including soccer. Its developer documentation says the company offers Odds Comparison, Live Odds, and Probabilities products through business-to-business application programming interfaces. (developer.sportradar.com) (postman.com) Polymarket’s sports contracts are structured as yes-or-no or side-versus-side event contracts that settle from official results. Its sports documentation says soccer winner-with-draw markets resolve on the result after 90 minutes plus stoppage time, while broader game-level contracts can include extra time and shootouts if the contract terms say so. (docs.polymarket.us) That settlement detail matters for any automated strategy, because a model has to predict not just what happens on the field but what the contract actually pays on. Polymarket’s documentation says official governing bodies are the primary settlement source, with Reuters, ESPN, BBC Sport and major wire services used if needed. (docs.polymarket.us) The X post included a how-to outline, but it did not publish broker statements, exchange records, or independently audited profit-and-loss data alongside the thread. X also did not provide line-by-line trading evidence in the public view available at the post URL. (x.com) Automating around live sports markets also runs into practical limits beyond prediction accuracy. The Princeton study said site restrictions, bet-slip reloads, and account controls made execution harder in real markets even when the model detected an apparent edge. (theses-dissertations.princeton.edu) So the clearest verified takeaway is narrower than the headline profit claim: live sports markets can drift out of sync for a few seconds, and firms such as Sportradar and venues such as Polymarket now provide enough structured data for traders to try to exploit that gap. (developer.sportradar.com) (docs.polymarket.us)
