Skip to content


Folders and files

Last commit message
Last commit date

Latest commit



35 Commits

Repository files navigation

This repository was created to visualize the popularity and win rate of different chess openings on Lichess over time. 3 billion games from the Lichess database were analyzed.

General popularity of Lichess in terms of the number of games played per day

It seems that usually, the main inflow happens in winters. Some spikes in the number of games played per day are associated with Marathon Tournaments. 2020 was an unusual year. The number of games played per day increased quite a lot at the start of the pandemic and after the release of "The Queen's Gambit" series. Weekly oscillations are also distinguishable.

The dynamics of the win rate of some openings

Many openings have experienced a change in their win rate and popularity during the COVID-19. But it seems that the changes in the win rates of different openings do have something in common. For most of the popular openings, the win rate shifted to the same side as the opening is played for. For example, the win rates of the Sicilian, the French and the Caro-Kann have shifted in black's favor while the win rate for white has increased in the openings such as the Queen's Gambit, the King's Gambit, the Evans Gambit and the Vienna Game (slightly).
There are exceptions like the Englund Gambit and the Stafford Gambit (although it was relatively unknown in March 2020).
Furthermore, before the pandemic, the win rates (for the color the opening is played for) were decreasing over the last few years. Probably, the average level of the lichess players behaved oppositely: it was gradually increasing before the start of the pandemic, but suddenly dropped in March 2020 as many new players joined lichess. Generally, there is a negative correlation between the win rate of popular openings (for the color the opening is played for) and the average level of players. The plots of win rates vs the average rating of players are also presented. The vertical error bars represent the statistical noise of 1/sqrt(N). The data for these plots was collected over November 2021 (87 mln games).

The dynamics of the popularity of some openigs

The graphs below are pretty self-explanatory. It seems that Youtube influences the popularity of openings quite strongly: After the initial increases, the frequency of the Evans Gambit returns almost to the same value in a couple of months. Perhaps, this is because Evans is hard. I couldn't explain the spike in the Schilling-Kostic Gambit in January 2019. You can find more graphs in the "graphs" repository, though the most interesting ones (in my opinion) were presented in this readme.


No description, website, or topics provided.






No releases published


No packages published