This talk summarises the results of a match-fixing study spanning over 5,000 in-play tennis betting markets. After developing a natural point-by-point probabilistic model, featuring novel mechanisms for selecting parameters robustly from the data, we demonstrate that the observed market trajectories correlate extremely closely with this model. We argue that substantial discrepancies represent a “red flag” that something is amiss - either an injury, or something more covert. We also provide visualisations of recent matches where the market evolved pathologically, and assess the evidence that these matches were fixed.
This talk is given by our sponsor, ATASS Sports. To learn more about them click here.