Personal bests as reference points

Reddit discussion, with comments from one of the researchers: [1]

The study used 133 million online chess games as data.

Comment from that researcher:

> I'm a professor of Computer Science at the University of Toronto and an avid chess fan. In my work I study large datasets of online human behaviour, and recently I've been studying decision-making with online chess games. I thought you'd be interested in seeing your online games being used for research :)

> In our paper, we find that players are motivated by beating their personal best ratings just as Kahneman and Tversky's prospect theory would predict. Personal bests are "reference points" that people are loss averse towards: players put in substantial effort to beat them, and once they do, they disproportionately stop playing for awhile (because they don't want to fall back down below their best). Sound familiar?

[1] https://www.reddit.com/r/chess/comments/7w60ss/just_publishe...