Sunday, May 26, 2019

Most popular time controls on lichess

Have you ever wondered which time controls were the most popular online? I sometimes do, because it's sometimes more difficult to get a game with certain time controls than others (looking at you, classical games!). It can be useful to know the relative frequency of time controls when planning on an online game.

TLDR: Bullet and Blitz reign supreme, longer time controls less popular

As you could expect, Bullet and Blitz games are more common than longer games (like Rapid and Classical). In fact, Rapid and Classical games represent approximately only 15% of the most popular time controls, while Bullet and Blitz represent about 79% of all games (UltraBullet accounts for 1% of the games; the remaining 5% least popular time controls contain a mix of time controls and are not shown here):


As can be seen from this graph, the most common time control is 1 minute per side with no increment (1+0; 25% of all games), followed by two Blitz time controls (in light green), 3+0 (18% of games) and 5+0 (17% of games). The only Rapid time control (in yellow) that made the list is the 10-minute game (10+0; 11% of games), while the only Classical time control (in red) that made the list is 15+15 (4% of games). UltraBullet remains largely unpopular, with the 15-second game accounting for less than 1% of all games.

Methodology

To examine the distribution of time controls in online games, I've used a subset of the games played on lichess in April 2019. There were a total of 33,565,536 games played on lichess that month, and in the interest of saving processing time, I used 2.8M of those games (about 9% of the games played that month, i.e. about 2.5 days of chess games). This means that the percentages above are estimates for April 2019 and not the exact numbers. (In fact, it would be interesting to look at whether preferences for time controls change over the course of the week.)

To parse the PGN file, I've used the library python-chess 0.28.0 in Python 3.6.7 and extracted the "TimeControl" tag for each game.

Remaining Questions

Of course, this simple analysis points to more questions than answers. Do preferences for time controls vary over the course of the week? Perhaps more interestingly, do preferences vary by websites? For example, if someone is interested in playing a longer time-control game, should they search for it on lichess or would they do well to look elsewhere? (I doubt it, but it remains to be investigated.)

Also, because Rapid and Classical games take more time to play out than Bullet and Blitz games, it's possible that the most time playing chess (as opposed to the number of games) is actually spent playing longer time-control games. But I haven't made those calculations...

1 comment:

Tapani said...

How does this change if you weigh every game by its duration? (actual, or theoretical)

The raw number of games does not tell the whole story. Another way to measure the popularity of a format might be to consider the amount of time invested in the format.