Fun stats about the Lichess 2026 Autumn Marathon (3 0) • page 3/3 • Community Blog Discussions • lichess.org

How do you gather the stats? Have you coded a program?

How do you gather the stats? Have you coded a program?

Yes, using the Lichess REST API (https://lichess.org/api) to retrieve the data/games and quite some custom script for processing

@devastator_sword said [^](/forum/redirect/post/20QhOqps) > How do you gather the stats? Have you coded a program? Yes, using the Lichess REST API (https://lichess.org/api) to retrieve the data/games and quite some custom script for processing

devastator_sword

@realcyberbird said ^

How do you gather the stats? Have you coded a program?

Yes, using the Lichess REST API (https://lichess.org/api) to retrieve the data/games and quite some custom script for processing

Cool, because I was wondering how its possible to do all of that manually

@realcyberbird said [^](/forum/redirect/post/G9nnuCD8) > > How do you gather the stats? Have you coded a program? > > Yes, using the Lichess REST API (https://lichess.org/api) to retrieve the data/games and quite some custom script for processing Cool, because I was wondering how its possible to do all of that manually

HaryDirb

The number of games detected as cheating is too low, only 3 games detected for cheating with 17070 players playing?

The actual number of players who cheat in such a large sample should be between 0.5% to 1.7% of players (conservative estimate), or 0.005 * 17070 = 85 cheaters to 0.017 * 17070 = 290 cheaters. So between 85 to 290 cheaters played in this tournament, and they only cheated in three games? Hard to believe. Also keep in mind the above numbers are conservative, as there are services online, where you pay a monthly fee, to allow you to cheat in a 'human way' that's hard to detect. By some estimates around 6.2% of players are cheating every year.

I am hoping that Lichess decided in the interest of not upsetting anybody with false positives that anti-cheat detection was adjusted so only the most blatant cheaters were flagged, aka in programming language the "contempt factor" was turn to very low number so that cheating was not aggressively tracked.

Otherwise Lichess's cheating detection may be broken, unless you want to assume that somehow players in this tournament were all angels, which is hard to believe.

"19752 players participated in the autumn 2026 marathon. There were 2682 players who did not play a game at all (13%) [so 19752 - 2682 = 17070 players played]. A total of 206604 (!) games were played during the tournament."

"The number of games which ended by Lichess because of detecting cheating during the game: 3 (0.0%)"

References, from "How Many Players Are Cheating? I Checked the Data", Man vs Chess on YouTube:

On Lichess in May 2026, 1.7% of active players on Lichess were later banned for cheating, higher than the 0.5% of active players on Chess. Com and in Germany an anonymous survey found 6.2% of players admitted to cheating at least once online in the last 12 months

The number of games detected as cheating is too low, only 3 games detected for cheating with 17070 players playing? The actual number of players who cheat in such a large sample should be between 0.5% to 1.7% of players (conservative estimate), or 0.005 * 17070 = 85 cheaters to 0.017 * 17070 = 290 cheaters. So between 85 to 290 cheaters played in this tournament, and they only cheated in three games? Hard to believe. Also keep in mind the above numbers are conservative, as there are services online, where you pay a monthly fee, to allow you to cheat in a 'human way' that's hard to detect. By some estimates around 6.2% of players are cheating every year. I am hoping that Lichess decided in the interest of not upsetting anybody with false positives that anti-cheat detection was adjusted so only the most blatant cheaters were flagged, aka in programming language the "contempt factor" was turn to very low number so that cheating was not aggressively tracked. Otherwise Lichess's cheating detection may be broken, unless you want to assume that somehow players in this tournament were all angels, which is hard to believe. HD "19752 players participated in the autumn 2026 marathon. There were 2682 players who did not play a game at all (13%) [so 19752 - 2682 = 17070 players played]. A total of 206604 (!) games were played during the tournament." "The number of games which ended by Lichess because of detecting cheating during the game: 3 (0.0%)" References, from "How Many Players Are Cheating? I Checked the Data", Man vs Chess on YouTube: On Lichess in May 2026, 1.7% of active players on Lichess were later banned for cheating, higher than the 0.5% of active players on Chess. Com and in Germany an anonymous survey found 6.2% of players admitted to cheating at least once online in the last 12 months

FM realcyberbird

@HaryDirb said ^

The number of games detected as cheating is too low, only 3 games detected for cheating with 17070 players playing?

The actual number of players who cheat in such a large sample should be between 0.5% to 1.7% of players (conservative estimate), or 0.005 * 17070 = 85 cheaters to 0.017 * 17070 = 290 cheaters. So between 85 to 290 cheaters played in this tournament, and they only cheated in three ...

Most of the games detected by Lichess for cheating are done when the game have finished. The 3 games I am mentioning are games which are aborted during the game. Which happens quite seldomly.

Good that you mentions this. I will clarify this in future blogs.

@HaryDirb said [^](/forum/redirect/post/ha56RtcR) > The number of games detected as cheating is too low, only 3 games detected for cheating with 17070 players playing? > > The actual number of players who cheat in such a large sample should be between 0.5% to 1.7% of players (conservative estimate), or 0.005 * 17070 = 85 cheaters to 0.017 * 17070 = 290 cheaters. So between 85 to 290 cheaters played in this tournament, and they only cheated in three ... Most of the games detected by Lichess for cheating are done when the game have finished. The 3 games I am mentioning are games which are aborted **during** the game. Which happens quite seldomly. Good that you mentions this. I will clarify this in future blogs.