Chess.com ratings have massively deflated since 2020

I propose a solution: Chess.com marks accounts as inactive within their databases(maybe an actual symbol on the person’s account, though I don’t see a benefit to that) after a two week period of not doing some of the smaller things available on this website (like lessons and puzzles) but after 4 weeks without playing a chess game the account becomes inactive, reven if someone was using chess.com’s smaller features. Chess.com should note take the ratings of these accounts into consideration while displaying the average ratings of people or calculating elo, therefore somewhat decreasing the elo inflation you guys are talking about. Feel free to make changes to my idea as needed because I’m not gonna read every comment on this forum and it’s like midnight so I’m tired lol

Just to point that the only reason is harder for a 2700 beat a 2750 than a 1200 win a game from a 1250 is that there will be much more draws in high level games. But the expected score performance is defined just by the elo gap, so they should face exactly the same difficulty to have the same score performance.

but 'will be much more draws' is a consequence of 'harder to win'.
'will be much more draws' is not a 'force' in and of itself.

My point is that they should have the same performance. For example, if they play 5 games let's say the 1250 player wins 3 and the 1200 player wins 2. On the other hand in 5 games the 2750 wins 1, the 2700 wins no game and we have 4 draws. Despite just the 1200 player has won a game, both had the same performance (2 of 5 points).

How about the 1200 won one and drew the other one?
How about the 1200 and 1250 went two two on wins plus a draw?
How about the 1200 scored 3.5 out of 5?
You're going to get more variance the further down you go.

I already explained it, but some of you are obviously unable to read with understanding.

The reason of deflation is that people, in general, improve over time.

It is the same with FIDE elo. A GM of 2500 of today is much stronger then a GM of 2500 30 years ago.

If you don't improve your game and continue to play your rating is going down despite the fact you play at the same level as it was 3 years ago because there are other players who do improve their game! Who would've suspected that!

Also, there is another reason which is that chess.com is in charge and can adjust the rating algorithm the way they like, and also, if they like, can adjust it to cause the deflation.

"If you don't improve your game and continue to play your rating is going down despite the fact you play at the same level as it was 3 years ago because there are other players who do improve their game! Who would've suspected that!"
--------------------
that doesn't make sense.
It doesn't follow.
The ratings of those other players would go up but that doesn't mean your rating goes down.
Your rating goes down when somebody beats you or somebody lesser-rated draws you.

The highest ELO of the world's #20 in the 1990s was 2680, while in the 2020s it is 80 points higher.

That alone is enough to prove that your claim is thin air.

I you follow the chess news you will remember that not so long ago FIDE gave players above...I think it was 2200...a bunch of free rating points to counter rating deflation, something like 200 free points as I recall. Pre-Covid there were more players above 2800 FIDE, now I think there are only 2 for classic otb chess. During the lockdowns tournaments were cancelled, so a lot of young players improved their actual strength greatly but played no rated games, so they were underrated. When they began to play again after Covid they took a lot of rating points from players rated higher, leading to deflation in the top ranks.

I don't follow fake news, and what you wrote is 100% FUD.

What had really happened last March was that the lower ELO threshold was raised from 1000 to 1400, and everybody up to 2000 rating got the analogous gain (from zero up to 400 points). In short, only the lower spectrum of the ratings was affected, because there was rating deflation there indeed, due to the high promo K=40 rating factor which is affecting ratings up to 2300.

The higher part of the rating spectrum (>2000) never, ever received any kind of bonus.

You didn't understand my point. It's not that everytime the players 50 points below will score 2/5 everytime, it's just the most probable outcome. These 50 points will always result in this gap no matter if we are talking about 1200''s players or 2700's players. The only thing that will change is the draw ratio that is much higher in higher elos.

I can try to demonstrate: Take a small rating pool with only 5 players. You are all rated 1600 and are at the same chess level. 4 of the players get better in equal amounts. They play each other and score 50% against each other, so their rating is maintained. Then, you whose ability is constant plays the other 4 players. You now perform less than 50% against them. Your rating goes down significantly as you are losing points to all players. The others only gain a marginal amount of rating as they only perform better against you, not themselves since they all improved.

Now, if you were the one that improved and the other 4 did not, your rating would rise to its level and the rating of the others would be marginally less than 1600, when their ability is 1600. Thus, a net deflation occurs no matter how many players in the pool improve.

In over the board chess, some countries have inflated or deflated FIDE ratings. Countries like India and China that have a lot of young improving players have lower FIDE ratings and countries that have a lot more aging players whose ability goes down and less young rising player (such as Russia) have higher FIDE ratings on average

Your demonstration is valid only if the ratings of the 5 players pool fluctuate equally. But this is far from being the case for FIDE standard ratings, There we have 3 different K factors which may well apply for any of these 5 players.

It can only apply to rapid and blitz games, where there is a single K=20 factor.

That is a good point. Kids have a higher K Factor. But I still think the kids that do not improve as much as average or do not improve at all (not every kid improved their chess in massive levels) will experience the deflation. We would also need to consider how many adults improve their ability compared to how many become less able.

Also, the K-factor is double for kids that it is for adults. I don't believe that there is any justification for using a multiple of 2 (that is I don't believe they have tested the ratings and come up with a multiple that will reduce inflation or deflation to the most amount they can) though of course that is just a guess.

its like 17 years old;still hasn't completed scholl boi LOL

chess com:2007-2024

Lichess:2011-2024

I have had the feeling that it has been getting harder to beat players with the same rating. I have been around the same rating for many years now, and so I wanted to do an analysis on my own games using the accuracy rating of chess.com . I could expand the analysis but wanted to hear some feedback on what you thought about the approach:

I looked at 711 Rapid games I played between 2017 and 2024 and calculated my average accuracy per year, as well as my average rating. I excluded games less than 20 moves as they seemed less relevant in terms of accuracy. I also excluded 2018 and 2021 because I hardly played any games these years.

Then I calculated the ratio of ELO rating per accuracy point by dividing the two, and here is the data I got:

As you can see my accuracy increases over time in the past few years, while my average rating stays similar, leading to a decreasing ratio that is lowest in 2024.

If I expand the analysis to include my opponents, the table shows similar data:

This would indicate a rating deflation, as it gets harder to beat an opponent of a similar rating.

this site feels very overinflated. i chomped my way to 2300 blitz without breaking a sweat.