Forums

FIx the rating system

Sort:
Avatar of Martin_Stahl
Martin_Stahl
Mar 2, 2022
0
#41
BKPete wrote:
4go10_legend wrote

It's like your telling a firemen to kill a fire using his fart

I see lots of possibilities to fix this - but it would be data driven & that makes it dependent on chess.com since they have the data. I have offered to help and would like to help. I would like to have a rating system that works for all - that seems a great prize for everyone.

 

Write an article, fully explaining your ideas, mathematics and methodology. It should probably include simulations against real world data as well. Get it published and peer reviewed.

 

A site isn't going to swap out its system for anything that isn't well researched, tested and vetted. It also isn't going to spend time testing something that isn't well formulated and has some real testing already behind it.

 

Avatar of BKPete
BKPete
Mar 2, 2022
0
#42

Well I don't think you provided enough evidence to justify your claims. You seem to base everything on the difference between your own two ratings (namely blitz and rapid), but you haven't yet proven that the reason behind is not the difference between your skill based on time control. Moreover, you seem to be too focused on percentiles, but it only stands to reason that for someone with an average rating (800 for cc) a reasonable change in rating will cause a great change in percentile, since most players are around that rating level.

Here's a thought: there seems a consensus that the blitz rating for most people with a rating around 600 to 1200 are lower than their rapid rating. (Looking at either the absolute rating or percentile). If true, that provides more proof of some bias in the rating system.

If there was no bias (i.e. the rating difference is explained by the different pools and the different game) we would have an equal mix of people who's blitz rating is higher or lower than their rapid rating.

I suspect the reason we don't see this is because new starters join with a rating that is too high (1200) and that creates lots of noise with most people having false ratings and playing lots of games against others with false ratings, so changes to ratings after the game are incorrect and the system never stabilises.

It would be interesting to see what proportion of players in the 600-1200 rating range have a blitz rating above their rapid rating, if any - that would at least prove the problem exists. 

(In the meantime, I'll monitor this against the next 20 random people I play).

Avatar of BKPete
BKPete
Mar 2, 2022
0
#43
BKPete wrote:

Well I don't think you provided enough evidence to justify your claims. You seem to base everything on the difference between your own two ratings (namely blitz and rapid), but you haven't yet proven that the reason behind is not the difference between your skill based on time control. Moreover, you seem to be too focused on percentiles, but it only stands to reason that for someone with an average rating (800 for cc) a reasonable change in rating will cause a great change in percentile, since most players are around that rating level.

Here's a thought: there seems a consensus that the blitz rating for most people with a rating around 600 to 1200 are lower than their rapid rating. (Looking at either the absolute rating or percentile). If true, that provides more proof of some bias in the rating system.

If there was no bias (i.e. the rating difference is explained by the different pools and the different game) we would have an equal mix of people who's blitz rating is higher or lower than their rapid rating.  Its impossible for everyone to be worse at blitz than rapid compared to others!

I suspect the problem is because new starters join with a rating that is too high (1200) and that creates lots of noise with most people having false ratings and playing lots of games against others with false ratings, so changes to ratings after the game are equally false and the system never stabilises.

It would be interesting to see what proportion of players in the 600-1200 rating range have a blitz rating above their rapid rating, if any - that would at least prove the problem exists. 

(In the meantime, I'll monitor this against the next 20 random people I play).

 

Avatar of BKPete
BKPete
Mar 2, 2022
0
#44
BKPete wrote:
BKPete wrote:

Well I don't think you provided enough evidence to justify your claims. You seem to base everything on the difference between your own two ratings (namely blitz and rapid), but you haven't yet proven that the reason behind is not the difference between your skill based on time control. Moreover, you seem to be too focused on percentiles, but it only stands to reason that for someone with an average rating (800 for cc) a reasonable change in rating will cause a great change in percentile, since most players are around that rating level.

Here's a thought: there seems a consensus that the blitz rating for most people with a rating around 600 to 1200 are lower than their rapid rating. (Looking at either the absolute rating or percentile). If true, that provides more proof of some bias in the rating system.

If there was no bias (i.e. the rating difference is explained by the different pools and the different game) we would have an equal number of people who's blitz rating is higher or lower than their rapid rating.  Its impossible for everyone to be worse at blitz than rapid compared to others!

I suspect the problem is because new starters join with a rating that is too high (1200) and that creates lots of noise with most people having false ratings and playing lots of games against others with false ratings, so changes to ratings after the game are equally false and the system never stabilises.

It would be interesting to see what proportion of players in the 600-1200 rating range have a blitz rating above their rapid rating, if any - that would at least prove the problem exists. 

(In the meantime, I'll monitor this against the next 20 random people I play).

 

 

Avatar of BKPete
BKPete
Mar 2, 2022
0
#45

18: Number of players (out of 20) with blitz rating lower than rapid rating - this doesn't look unbiased to me.

To me, this shows the ratings for at least one of either blitz or rapid is broken at this rating range.

Avatar of jetoba
jetoba
Mar 3, 2022
0
#46
BKPete wrote:

18: Number of players (out of 20) with blitz rating lower than rapid rating - this doesn't look unbiased to me.

To me, this shows the ratings for at least one of either blitz or rapid is broken at this rating range.

That many people with Blitz lower than Rapid may well be an indication that the rating system is working as far as differentiating between the strengths of the various players.  That can be confirmed by ranking those 20 players by each of the two ratings and seeing if they have pretty much the same relative ranks (there will be some discrepancies because some players are relatively stronger at different time controls).

If the ratings were supposed to be the same within each system then there wouldn't be a need for multiple ratings.  People thinking the ratings should be the same are missing the purpose.

Avatar of Romans_5_8_and_8_5
Romans_5_8_and_8_5
Mar 4, 2022
0
#47
TacticalPrecision wrote:

Where's the classical rating that's been desperately needed and asked for repeatedly over the years? 

Pretty sure you're the only one who wants a classical rating system. You're just upset over your 1200 rating lol.

Avatar of Sred
Sred
Mar 4, 2022
0
#48

@BKPete, are you serious that you offered chess.com help to fix their rating system? Then you are facing a lot of scientific work to improve on http://www.glicko.net/glicko/glicko.pdf.

Avatar of Romans_5_8_and_8_5
Romans_5_8_and_8_5
Mar 4, 2022
0
#49
TacticalPrecision wrote:
ShrekChess69420 wrote:
TacticalPrecision wrote:

Where's the classical rating that's been desperately needed and asked for repeatedly over the years? 

Pretty sure you're the only one who wants a classical rating system. You're just upset over your 1200 rating lol.

 

1255 rapid is fine here for now and going up. 1830+ classical on Lichess. If you were capable of reading, you'd know there have been countless, countless threads asking for a classical rating over the years, and especially over the last year. 

You, a total casual who plays 10/0 chess, really wouldn't understand. 

"If you were capable of reading, you'd know there have been countless, countless threads asking for a classical rating over the years, and especially over the last year." Unlike you, I have a life, and I don't obsess over reading chess forums lol. You just try to use your lichess rating as "leverage" on a completely different site, because you actually think it actually means something! 

Avatar of Romans_5_8_and_8_5
Romans_5_8_and_8_5
Mar 4, 2022
0
#50
TacticalPrecision wrote:
ShrekChess69420 wrote:
TacticalPrecision wrote:

Where's the classical rating that's been desperately needed and asked for repeatedly over the years? 

Pretty sure you're the only one who wants a classical rating system. You're just upset over your 1200 rating lol.

 

1255 rapid is fine here for now and going up. 1830+ classical on Lichess. If you were capable of reading, you'd know there have been countless, countless threads asking for a classical rating over the years, and especially over the last year. 

You, a total casual who plays 10/0 chess, really wouldn't understand. I'm sorry I didn't respond to your endless game challenges the other day but I'm an adult and don't specifically enjoy playing games with children who can't even get into a PG-13 movie alone. 

Age means nothing. You act like I'm a complete fool to detract attention away from your own terrible accomplishments. You're 1200, bro. The real reason you didn't respond to my challenges is because you know you're gonna get clapped. you're afraid of getting utterly humiliated by a kid. I'm a 1700 10/0 casual; at least i'm not 1200 lol. Get good, kid. 

Avatar of Romans_5_8_and_8_5
Romans_5_8_and_8_5
Mar 4, 2022
0
#51
TacticalPrecision wrote:
ShrekChess69420 wrote:
TacticalPrecision wrote:
ShrekChess69420 wrote:
TacticalPrecision wrote:

Where's the classical rating that's been desperately needed and asked for repeatedly over the years? 

Pretty sure you're the only one who wants a classical rating system. You're just upset over your 1200 rating lol.

 

1255 rapid is fine here for now and going up. 1830+ classical on Lichess. If you were capable of reading, you'd know there have been countless, countless threads asking for a classical rating over the years, and especially over the last year. 

You, a total casual who plays 10/0 chess, really wouldn't understand. 

"If you were capable of reading, you'd know there have been countless, countless threads asking for a classical rating over the years, and especially over the last year." Unlike you, I have a life, and I don't obsess over reading chess forums lol. You just try to use your lichess rating as "leverage" on a completely different site, because you actually think it actually means something! 

 

Yes: Rapid and Classical ratings on Lichess mean a lot more than your 10/0 results here on Chess.com. Those mean nothing. You play 10/0 because you have no attention span and you can't handle actually having to play out endgames. It's no secret. 

And you don't have a life as you're a child. You have virtually nothing to do all day. I mean, you're a big fan of Shrek. That right there means your life is pretty pointless. 

LOL. You always bring your arguments back to: this person's a kid and likes Shrek. This whole time you can't even realize that you're getting hard trolled by an account named SHREKCHESS69420. You act like there's someone behind my profile picture who actually CARES about our arguments. It's hilarious. Thanks for hours of YouTube content. My friends have really enjoyed seeing your reactions. 

Avatar of Romans_5_8_and_8_5
Romans_5_8_and_8_5
Mar 4, 2022
0
#52
TacticalPrecision wrote:
ShrekChess69420 wrote:
TacticalPrecision wrote:
ShrekChess69420 wrote:
TacticalPrecision wrote:

Where's the classical rating that's been desperately needed and asked for repeatedly over the years? 

Pretty sure you're the only one who wants a classical rating system. You're just upset over your 1200 rating lol.

 

1255 rapid is fine here for now and going up. 1830+ classical on Lichess. If you were capable of reading, you'd know there have been countless, countless threads asking for a classical rating over the years, and especially over the last year. 

You, a total casual who plays 10/0 chess, really wouldn't understand. I'm sorry I didn't respond to your endless game challenges the other day but I'm an adult and don't specifically enjoy playing games with children who can't even get into a PG-13 movie alone. 

Age means nothing. You act like I'm a complete fool to detract attention away from your own terrible accomplishments. You're 1200, bro. The real reason you didn't respond to my challenges is because you know you're gonna get clapped. you're afraid of getting utterly humiliated by a kid. I'm a 1700 10/0 casual; at least i'm not 1200 lol. Get good, kid. 

 

Again: Your 10/0 ratings mean nothing. What you do here on Chess.com means nothing. Your self worth and who you are is not attached to your 10/0 ratings lol. Grow up. You don't have a classical rating on lichess, or play any real chess at all for that matter, because you can't focus on a game for more than a few minutes. You have plenty of time for the Shrek wedding dance, though! What a total loser. 

You're right, I never said that chess ratings mean anything. Although, they mean the world to YOU. Aw man, I don't play classical chess, I'm a TERRIBLE PERSON! Well, I'm doing bigger things than raving about chess ratings on a chess site. I hope that when I'm 40 I won't end up like you. 

Avatar of Martin_Stahl
Martin_Stahl
Mar 4, 2022
0
#53
TacticalPrecision wrote:

Where's the classical rating that's been desperately needed and asked for repeatedly over the years? 

 

My understanding is it will be around at some point, but it's likely going to be a while. My understanding is the way the ratings are stored makes it harder to add a new rating, due to the number of records in the DB table holding that data.