Project Details
Projekt
Data-driven Indication of Match Fixing in Live Betting
Subject Area
Statistics and Econometrics
Term
since 2020
Project identifier
Deutsche Forschungsgemeinschaft (DFG) - Project number 431536450
The proposed extension of the ongoing project aims at developing data-driven tools for the indication of betting fraud and for the fight against corruption. Specifically, bets placed during soccer matches (so-called live bets) will be examined with respect to statistical anomalies. Live bets are placed on very liquid markets and in the past have been used by betting fraudsters, who can exploit their insider knowledge of match events. Access to comprehensive data is guaranteed by an ongoing cooperation of the working group with the large betting provider TIPICO. TIPICO provides data on betting stakes and betting odds for live betting on a second-by-second basis. In the proposed extension of the project, data for the seasons 2017/18 up to and including 2020/21, for the German Bundesliga and the Italian Serie B, will be analyzed. In the ongoing first phase of the project, the factors influencing betting volumes during matches in the German Bundesliga were analyzed. This was necessary to understand betting market dynamics during matches, and thereby to be able to determine drivers and expected values of betting volumes at different times during the match. Subsequently, systematically biased assessments of outcome probabilities by bettors were identified. This is important as suspicious betting activity can be explained either by betting fraud or by such market inefficiencies. For the identification of betting fraud, this differentiation hence is crucial. In the requested second phase of the project, the aim is to develop statistical tools for fraud indication, making use of the findings from the first phase. The focus will lie on bets and betting odds in the time period just before goals are scored. In these periods, betting fraudsters benefit the most from their additional information related to upcoming, potentially match-deciding events. Therefore, it is important to derive expected values for betting odds and betting volumes for this period, in order to detect suspicious deviations from these in an automated way. The findings of this part of the project, which will be based on the results of the initial first phase, enable the indication of betting fraud and the development of tools to supervise betting markets. Suspicious bets and patterns in betting behavior before goals that cannot be explained by market inefficiencies indicate games that might have been manipulated, which will be flagged accordingly. In the context of the study conducted, it will be exploited that some matches in the Italian Serie B are known to have been fixed. Specifically, the matches flagged by the statistical analysis will be compared with those matches known to be manipulated, thus roughly assessing the models’ classification accuracy.
DFG Programme
Research Grants
