amazing blogpost - Thanks for all the effort! Has been a long time since I read something as well-written as this.
This is all bullshit. I play almost 1 min games so nobody can cheat on my games. When i play slow games like 5-3min increments i take my time to play each moves that i cannot do on a lightning games...that does not mean i cheat because i take too long to p lay a move.. but i dont care.. there is other sites to play....and i still can play but my rating does not change. Its is a shame that i cannot play standard tournaments.
Bullet is traditionally considered a difficult-to-cheat-in time control. To help the validity of this analysis, were you able to determine that bullet games were the reason these accounts were closed?
Some methods for doing the above: filter out accounts that also play longer time controls (which are considered easier to cheat in), or compare rating differentials between time controls. Also apply these methods to the "normal" players group, too, of course.
It's also important to control various covariates (e.g. rating, account age, ...), since gameplay looks quite different for each of these. And methods/probabilistic analysis for catching/predicting cheaters under one group may not apply to another group.
Some methods for doing the above: filter out accounts that also play longer time controls (which are considered easier to cheat in), or compare rating differentials between time controls. Also apply these methods to the "normal" players group, too, of course.
It's also important to control various covariates (e.g. rating, account age, ...), since gameplay looks quite different for each of these. And methods/probabilistic analysis for catching/predicting cheaters under one group may not apply to another group.
@sicariusnoctis said in #4:
>
I was not able to. This will forever limit the analysis I will do in the future. I will make sure to include it in part of my limitations. I understand the methodology but conducting all those with my resource might be challenging. I will take this into account in my future analysis and try some of those techniques. Thank you.
>
I was not able to. This will forever limit the analysis I will do in the future. I will make sure to include it in part of my limitations. I understand the methodology but conducting all those with my resource might be challenging. I will take this into account in my future analysis and try some of those techniques. Thank you.
Wow bro, how do you get this info this is crazy. That blog poster probably got hundreds of games at least and studied them from CLOSED accounts. That's tough man.
ive no idea how to read those graphs too low iq)
The "normal player" group probably has another distribution of rating numbers as the two other groups didn't it?
Shouldn't you normalize this by picking a subset with a similar distribution as the other groups? Otherwise the given metrics could also be interpreted as ‚normal players vs. better players'.
Shouldn't you normalize this by picking a subset with a similar distribution as the other groups? Otherwise the given metrics could also be interpreted as ‚normal players vs. better players'.
Hi - why your account has been marked for violating lichess terms of service?
The graphs are cool though - they look like satellites in the Oort Cloud 10000 AU from earth :)