This post was absolutely correct. I've recently really felt it getting harder to beat lower rated players.
@Schachsuchti64 said in #9:
> At least in physics, the word is equilibrium. :-)
Yes it all depends on the definition of the state. I read recently about definitions that get bogged down in the type of formulations, that equilibrium, steady-state or stationnary distribution might not be the same thing.. I put a parenthesis with question mark to not be read to literally. As not every one is use to looking at functions as fixed points. And I have a not a good record with words to pointee associations being the same as everyone else (or so I think, and whoever that might point to....).
Sometimes being a little vague is better clarity, with proper intent transmission.. but clearly equilibrium assumption would fit what I meant.
> At least in physics, the word is equilibrium. :-)
Yes it all depends on the definition of the state. I read recently about definitions that get bogged down in the type of formulations, that equilibrium, steady-state or stationnary distribution might not be the same thing.. I put a parenthesis with question mark to not be read to literally. As not every one is use to looking at functions as fixed points. And I have a not a good record with words to pointee associations being the same as everyone else (or so I think, and whoever that might point to....).
Sometimes being a little vague is better clarity, with proper intent transmission.. but clearly equilibrium assumption would fit what I meant.
I'm on my USCF floor, so at least now I can donate points to the generation glued to their phones.
I'm not really any worse, we did have a year of inflation when they implemented the bonus points for performing over your rating, but now there has been a steep overcorrection.
I'm not really any worse, we did have a year of inflation when they implemented the bonus points for performing over your rating, but now there has been a steep overcorrection.
@MagiciansGambit said in #3:
> The Glicko system is a lot better than the Elo system.
> Using it will probably solve the problem of underated players ...
Glicko is only little bit better and does not address the issue in FIDE ratings at all. Could have small effect. Lichess and online servers have different problem. People enter FIDE system but they do not drop out as much. Let alone they do not drop out enter with new name.
Corona pandemic added with "queens gambit" series caused rating inflation in lichess. So system reacted despite being Glicko. This is basic feature of any Bradley-Terry type of system as there is no golden standard to compare with. We compare players on to other players in pool and if constitution of pool changes then ratings will change somehow.
> The Glicko system is a lot better than the Elo system.
> Using it will probably solve the problem of underated players ...
Glicko is only little bit better and does not address the issue in FIDE ratings at all. Could have small effect. Lichess and online servers have different problem. People enter FIDE system but they do not drop out as much. Let alone they do not drop out enter with new name.
Corona pandemic added with "queens gambit" series caused rating inflation in lichess. So system reacted despite being Glicko. This is basic feature of any Bradley-Terry type of system as there is no golden standard to compare with. We compare players on to other players in pool and if constitution of pool changes then ratings will change somehow.
@Molurus said in #4:
> FIDE should probably just adopt Glicko (or Glicko-2). The only real reason they don't is because 'we have always used elo, so why stop now?'
if thet would change why would they change to an old system that on marginally better as there are dozens of better system. In FIDE/Deloitte Kaggle competition Glicko ended on place 50.
But changing a system is very very hard. FIDE is huge organisation where most member national associatons would have agree on change. Far more likely to get trough tuning on current than switching on totally new.
> FIDE should probably just adopt Glicko (or Glicko-2). The only real reason they don't is because 'we have always used elo, so why stop now?'
if thet would change why would they change to an old system that on marginally better as there are dozens of better system. In FIDE/Deloitte Kaggle competition Glicko ended on place 50.
But changing a system is very very hard. FIDE is huge organisation where most member national associatons would have agree on change. Far more likely to get trough tuning on current than switching on totally new.
@petri999 interesting, something for me to read on for the next few days. I've always thought Glicko 2 is best.
It cannot even in principle be best. It is maximum likelihood model BUT takes into account players own games not the network of games. and way it is done online it takes just one game not series of games as original design assumed-
www.kaggle.com/competitions/ChessRatings2/leaderboard
www.kaggle.com/competitions/ChessRatings2/leaderboard
What about high rated players?!
(Not saying I am)
(Not saying I am)
Since problem is low - low/middle it needs several hops to reach top which migitates the problem. Elo is not zero sum games due K-factor which dampens the effect.
@petri999 said in #19:
> Since problem is low - low/middle it needs several hops to reach top which migitates the problem. Elo is not zero sum games due K-factor which dampens the effect.
However you need to keep into consideration that lots of professional players play much more rated games than most amateurs. This means even with a much smaller K-factor they still get quite fast a visible impact on their ratings. I see around me many professionals playing many opens which have lost a lot of points in recent years.
I discuss the deflation also in my blogarticle which I just published: schaken-brabo.blogspot.com/2023/08/de-schaakbuddy.html
> Since problem is low - low/middle it needs several hops to reach top which migitates the problem. Elo is not zero sum games due K-factor which dampens the effect.
However you need to keep into consideration that lots of professional players play much more rated games than most amateurs. This means even with a much smaller K-factor they still get quite fast a visible impact on their ratings. I see around me many professionals playing many opens which have lost a lot of points in recent years.
I discuss the deflation also in my blogarticle which I just published: schaken-brabo.blogspot.com/2023/08/de-schaakbuddy.html