Many laboratory experiments fail to find evidence for the game-theoretic concept of “mixed strategies.” But Doru Cojoc, a graduate student at Clemson University, looks at data taken from the world of chess, where high prizes are on the line and we find repeated games between the same players (there is no copy of the paper on the web).
A player might prefer one opening move over another (e.g., “1. e4” vs. “1. d4”), but if a player always uses his favorite, the opponent will find it easier to prepare a defense. So players tend to vary their opening moves in an effectively random manner, as confirmed by Cojoc’s data from championship matches. The returns to differing opening moves end up being the same, in expected value terms, even though players have their favorites. Note: For purposes of contrast, I would like to see if chess champions do any better with their favorite moves in non-repeated settings, Cojoc says he is working on this.
Doru tells me he is also preparing work on whether chess players ever reason using backwards induction strategies. And click here for a lead on the Chiappori, Steve Leavitt, and Tim Groseclose paper on mixed strategies in soccer.
By the way, did you know that for world championship games since 1951, the player with white is more than twice as likely to win as the player with black (26% white wins, 12% black wins, 61% draw)?
Thanks to Bob Tollison for the pointer on this work.