This talk has been automatically captioned. You can remove these by pressing CC on the video toolbar.

Name: Richard Smith

Talk Title: Extreme value theory and chess ratings

Abstract: The Elo rating system is well established as a method of evaluating competitive chess players. However, since the early days of the system, women have been severely underrepresented at the top levels of the game. The "participation rate" hypothesis suggests that this can be explained in terms of the relative numbers of male and female players - the underlying populations are similar, but men prevail at the top level simply because there are so many more of them. I shall present two approaches to assessing this hypothesis, one based on permutation tests for the null hypothesis that the male and female populations have the same underlying distribution, the other directly using extreme value theory. Preliminary results suggest that the data are quite consistent with the participation rate hypothesis in some countries, but not in others, and we are exploring the potential reasons for this in more detail. This talk is based on a collaborative research project with co-authors Weiji Ma (New York University, USA), Nikos Bosse (London School of Hygiene and Tropical Medicine, UK), Jose Camacho Collados (Cardiff University, UK), Hou Yifan (Shenzen University, China) and David Smerdon (University of Queensland, Australia).