In the first of a planned series of reports on Titled Tuesday, Chess.com’s Fair Play team said it’s unlikely that large-scale cheating is happening in the popular prize events.
Chess.com’s anti-cheat detection system has faced scrutiny from some of the world’s best players lately, calling into question whether their concerns about cheating in Titled Tuesday are being taken seriously.
But is the cheating as widespread as suggested? That is the question Chess.com’s Fair Play team, which consists of more than 30 people, including 10 titled players, has investigated over the last month. The first of a series of reports was published on Thursday.
“Our analysis in this report leads us to believe that large-scale cheating is not happening in Titled Tuesday,” the team said in its report over 16 pages.
Our analysis in this report leads us to believe that large-scale cheating is not happening in Titled Tuesday.
In the research, the frequency of upsets in Titled Tuesday was compared to over-the-board (OTB) blitz events. How well do underdogs (the lower-rated player) score against higher-rated opponents?
A notable finding in the analysis is that while upset rates overall are similar, it was slightly higher OTB. In Titled Tuesday underdogs scored 32 percent but had a “surprise rate” of 37 percent in OTB blitz.
It makes sense that both groups of underdogs would have average scores below par (50%) – they’re underdogs after all – but it might be surprising to some that, in the aggregate, underdogs perform better in OTB blitz games than in Titled Tuesday games.
The discrepancy could be explained by important differences between OTB blitz and online play, such as the prevalence of pre-moves, the importance of mouse speed, internet connection, and higher levels of pressure OTB, the report notes.
FIDE blitz ratings were used to determine underdogs, and the research noted that matchups in Titled Tuesday tend to be more lopsided than in OTB blitz, which could explain why underdogs underperform compared to OTB.
Taking rating differences into account, and comparing underdog average scores by 50-point rating difference bucket, the research produced these results:
“Overall, we see that underdog average scores in Titled Tuesday games and in OTB blitz games are generally very similar no matter the size of the underdog,” the report said.
Overall, we see that underdog average scores in Titled Tuesday games and in OTB blitz games are generally very similar no matter the size of the underdog.
The research also compared underdog average scores by rating buckets.
As an example, the heatmap shows that the average score when the underdog is between 2150-2299 in FIDE blitz, facing a higher-rated player between 2600 and 2749, is 17 percent in both OTB blitz and Titled Tuesday.
For each comparison, we saw that average scores in Titled Tuesday games mostly align with average scores in OTB blitz games. The differences that we did see were small, and do not support the argument that cheating in Titled Tuesday is widespread, even against top players.
The report acknowledged that cheating does happen in Titled Tuesday, but the indications are that it’s limited.
Let us be very clear: cheating is an existential threat to the game of chess, and it does happen in Titled Tuesday. At the same time, based on the information we have, we believe that cheating in Titled Tuesday is limited and does not meaningfully impact Titled Tuesday from week to week.
(…) based on the information we have, we believe that cheating in Titled Tuesday is limited and does not meaningfully impact Titled Tuesday from week to week.
IM Kenneth Regan, an independent expert on anti-cheat detection, said in a post on Friday that he supported the finding in the report.
“Moreover, I have reproduced them within my own framework. One point of commonality is that we base “underdog” on FIDE ratings, not Chess.com’s own Blitz ratings. This is to avoid the supposition that players may have inflated their Chess.com ratings via cheating, so they would not appear as underdogs in as many games,” he said.
You can read the full report below:
