As I keep promising, I will write this up on my (mostly non-chess) blog. But briefly, consider how Elo is defined. Let p be the probability that White wins a match. Let delta_e be the difference in El

How I'd love to understand all this... :(

hi, The code to scrape and process the data down to csv is here: https://github.com/shabbychef/lichess-data. I don't have instructions yet. You would need python3.9, with the python-chess package, an

Thanks shabby! Which software did you use for creating this database and getting all those statistics?

ok, that was easy. I will write up a blog post about this at some point, but I fit logistic regression to the pre-game difference in Elo of the players as well as the opening move. Again restricting t

I scraped all the lichess atomic games (see https://database.lichess.org/#variant_games), and ran an analysis. For games where both players had pre-game Elo > 1600 and had at least 30 games in the dat