Detecting individual preferences and erroneous verdicts in mixed martial arts judging using Bayesian hierarchical models
用贝叶斯分层模型分析综合格斗裁判的个人偏好，发现裁判对不同动作有偏好且可能影响比赛结果，并开发模型预测公平结果，对体育裁判研究者和赛事组织者有用。
In this paper, we use Bayesian hierarchical models to investigate the decision-making of judges of mixed martial arts (MMA) contests. Whilst there has been research into the judging of various sports in the past, none have explicitly modelled the judges’ behaviours at an individual level. We progress the literature by demonstrating that judges have personal preferences towards the different actions that they must assess during a fight. The preferences themselves may be the deciding factor in a bout, as demonstrated using a historical case study. We apply the concept of variable significance to the predictions of scores, to assess whether a judge’s verdict was within reason. Finally, we develop a model that predicts a bout’s fair outcome, which could be used in various ways in MMA.