Partitioning the data

• {  UltraBullet:   14K events, 259K players, 648M moves, 13M games, 1.5M points.
• T  HyperBullet:   20K events, 252K players, 1.2B moves, 18M games, 851K points.
• T  Bullet:   122K events, 614K players, 4.1B moves, 60M games, 3.4M points.
• )  SuperBlitz:   73K events, 804K players, 4.0B moves, 57M games, 3.1M points.
• )  Blitz:   81K events, 843K players, 2.7B moves, 40M games, 2.2M points.
• #  Rapid:   32K events, 546K players, 977M moves, 15M games, 1.1M points.
• +  Classical:   16K events, 321K players, 372M moves, 5.8M games, 75K points.
•   Crazyhouse:   55K events, 93K players, 208M moves, 4.1M games, 103K points.
• '  Chess960:   3.1K events, 39K players, 48M moves, 714K games, 1.8M points.
• (  King of the Hill:   2.3K events, 25K players, 12M moves, 232K games, 52K points.
• .  Three-check:   2.9K events, 27K players, 13M moves, 374K games, 68K points.
• @  Antichess:   7.7K events, 39K players, 78M moves, 1.6M games, 4.6M points.
• >  Atomic:   7.7K events, 42K players, 53M moves, 1.7M games, 4.8M points.
• _  Horde:   2.9K events, 20K players, 17M moves, 182K games, 21K points.
•   Racing Kings:   2.4K events, 18K players, 6.2M moves, 222K games, 621K points.

Partitioning the events included in the rankings by type, we unsurprisingly see that there were many more hourly arenas than yearly arenas. At the same time yearly arenas are more special (and generally last longer) and therefore get more participants per event than hourly arenas.

• Hourly:   274K events, 962K players, 7.5B moves, 114M games, scoring 280M points.
• <2000:   44K events, 614K players, 1.4B moves, 21M games, scoring 49M points.
• <1700:   41K events, 554K players, 1000M moves, 15M games, scoring 36M points.
• <1600:   3.8K events, 18K players, 24M moves, 377K games, scoring 941K points.
• <1500:   41K events, 458K players, 677M moves, 11M games, scoring 25M points.
• <1300:   1.2K events, 43K players, 20M moves, 349K games, scoring 422K points.
• Daily:   26K events, 647K players, 1.7B moves, 26M games, scoring 67M points.
• Weekly:   2.5K events, 336K players, 402M moves, 6.1M games, scoring 0 points.
• Monthly:   836 events, 292K players, 298M moves, 4.6M games, scoring 0 points.
• Yearly:   73 events, 178K players, 89M moves, 1.4M games, scoring 0 points.
• Eastern:   7.3K events, 343K players, 634M moves, 9.3M games, scoring 0 points.
• Elite:   493 events, 18K players, 45M moves, 542K games, scoring 0 points.
• Shield:   592 events, 278K players, 342M moves, 5.4M games, scoring 0 points.
• Titled:   47 events, 4.0K players, 25M moves, 306K games, scoring 0 points.
• Marathon:   23 events, 157K players, 199M moves, 2.9M games, scoring 7.8M points.

For an even more detailed breakdown of all events in these rankings in terms of types and variants, see the following table. Percentages in this table are rounded to the nearest integer; 1% means "between 0.5% and 1.5%" and 0% means "less than 0.5%", while a dash in all tables below means that no events in this combination of categories took place. Almost 30% of all events are bullet arenas, and as they take place (almost) every hour, almost 90% of the events included in these rankings are hourly arenas.

Classification of all events

Event statistics

The tables below summarize various statistics per game type and per event type of arena tournaments. We briefly discuss some things one may conclude from these tables below.

• Average participants per event: Overall, the average number of participants per event is 180 players. Overall, rapid events see the most participants on average, and whereas all standard time controls have an average participation of over 100 players (and around 90 for chess 960), all non-standard chess variants have an average of fewer than 60 players per event.
• Average ratings per event: While average ratings across Lichess may be around 1500, tournaments seem to have a bias towards higher-rated players. As lower-rated players have no chance of winning these events, they may prefer matchmaking over random arena pairings where they lose most of their games. Overall, the average rating in events is around 1681, with slightly higher averages in faster time controls and in crazyhouse. As for types of events with higher averages, we naturally see peaks for titled arenas and elite arenas, which are only open to high-rated players.
• Average points per player per event: Overall, an average arena participant scored around 0 points. With many players joining for only a few games, and some players playing the entire events and scoring many points, the median is even lower. The average number of points is highest for quick time controls, where it is easier to play many games in a short amount of time.
• Maximum points achieved in one event: Looking only at the highest scores ever achieved in arenas, we see that the highest score ever achieved was by GM penguingim1 in one of the bullet marathons, scoring 1645 points. (As explained on the page with special events this is not the maximum score ever achieved, as six players have scored over 2000 points in 24-hour hyperbullet events.)

Average participants per event

Average ratings per event

Average points per player per event

Maximum points achieved in one event

Game statistics

Finally, below are some tables listing game statistics per game type and per event type. We again briefly highlight some things one might observe in these tables.

• Average moves per player per game: On average, games played in all these events lasted around 33 moves. In terms of variants, both extremely short (ultrabullet) and long time controls (classical) have a slightly lower average than "medium" time controls such as superblitz or blitz. Of the variants, racing kings (14), atomic (16), and three-check (18) have the lowest average moves per game due to the nature of these variants, while in horde (46) the average is significantly higher than for standard chess games. With the higher average rating and with more at stake, titled arenas (and elite arenas) have a higher average number of moves per game as well.
• Average score of white players: While white may have a slight edge in chess in general, a (large) difference in playing strength will easily offset this small advantage. Overall, in all arenas combined, white players have scored approximately 51%. White still scores slightly more than 50% in (almost) all events, except in horde, where the black player seems to have a slight edge over white. We further notice a slightly larger win percentage for white in atomic, three-check, and crazyhouse, and due to the higher level of play and the narrower range of playing strengths in these events, in both elite arenas and titled arenas white players score slightly better with a 52% score.
• Average percentage of draws: As most tournaments are open to a wide range of ratings, and as shorter games played online are more often decisive, it is no surprise that the overall percentage of draws is only around 3%. This percentage is even lower at faster time controls and some variants like crazyhouse and three-check, while racing kings has a remarkably high draw percentage of around 10%.
• Average percentage of berserked games: On average, players choose the berserk option in around 10% of their arena games. This percentage is slightly higher (20% or more) for slow time controls, since losing half the clock time still leaves plenty of time to play a serious game. In non-standard chess variants, berserking seems especially popular in atomic and racing kings events, a statistic which may be closely related to the lower average game lengths in these variants; if games are often decided in fewer moves, losing time on the clock is not as important as for long games played in e.g. horde arenas.

Average moves per player per game

Average score of white players

Average percentage of draws

Average percentage of berserked games