lichess.org
Donate

Why chess ratings don't mean what they used to

@mvhk said ^

I also noticed in my last tournament a number of + 50 year players having gained hundreds of ratingpoints after the elo-correction of March 2024 (so on top of the elo-correction). Are those isolated cases or a trend?

I don't think it's really appropriate to talk about gain here. Optically, a ~1100 (old) FIDE rated player "gained" almost 400 rating points but as all other ~1100 (old) rated players are now also ~1500 (new) rated, they don't actually gain anything. You just have to get used to 1500 not meaning what it used to (now it's only 100 above the floor) and that the level you expected from a ~1500 (old) FIDE rated player is now at ~1700 level.

@mvhk said [^](/forum/redirect/post/DfPj9nN4) > I also noticed in my last tournament a number of + 50 year players having gained hundreds of ratingpoints after the elo-correction of March 2024 (so on top of the elo-correction). Are those isolated cases or a trend? I don't think it's really appropriate to talk about *gain* here. Optically, a ~1100 (old) FIDE rated player "gained" almost 400 rating points but as all other ~1100 (old) rated players are now also ~1500 (new) rated, they don't actually *gain* anything. You just have to get used to 1500 not meaning what it used to (now it's only 100 above the floor) and that the level you expected from a ~1500 (old) FIDE rated player is now at ~1700 level.

@mkubecek said ^

I don't think it's really appropriate to talk about gain here. Optically, a ~1100 (old) FIDE rated player "gained" almost 400 rating points but as all other ~1100 (old) rated players are now also ~1500 (new) rated, they don't actually gain anything. You just have to get used to 1500 not meaning what it used to (now it's only 100 above the floor) and that the level you expected from a ~1500 (old) FIDE rated player is now at ~1700 level.

You misunderstood my comment. I give a concrete example so it becomes more clear: https://ratings.fide.com/profile/1109502/chart.
The elo-correction gave the player 72 points in March 2024. However between March 2024 and now, he gained another almost 200 points. He is 51 years and playing for decades chess. There are many such examples. How do you explain?

@mkubecek said [^](/forum/redirect/post/9UgNnPHa) > I don't think it's really appropriate to talk about *gain* here. Optically, a ~1100 (old) FIDE rated player "gained" almost 400 rating points but as all other ~1100 (old) rated players are now also ~1500 (new) rated, they don't actually *gain* anything. You just have to get used to 1500 not meaning what it used to (now it's only 100 above the floor) and that the level you expected from a ~1500 (old) FIDE rated player is now at ~1700 level. You misunderstood my comment. I give a concrete example so it becomes more clear: https://ratings.fide.com/profile/1109502/chart. The elo-correction gave the player 72 points in March 2024. However between March 2024 and now, he gained another almost 200 points. He is 51 years and playing for decades chess. There are many such examples. How do you explain?

@mvhk said ^

You misunderstood my comment. I give a concrete example so it becomes more clear: https://ratings.fide.com/profile/1109502/chart.
The elo-correction gave the player 72 points in March 2024. However between March 2024 and now, he gained another almost 200 points. He is 51 years and playing for decades chess. There are many such examples. How do you explain?

Looking at the chart, what I see is that this player has been around 2050 consistently from 2007 to mid 2021. So I would rather ask what were the reasons of the fast drop between August 2021 (2041) and September 2023 (1787) and think about the recent raise as fixing the problem and returning to the previous ~2050 level.

@mvhk said [^](/forum/redirect/post/TkvC7v8O) > You misunderstood my comment. I give a concrete example so it becomes more clear: https://ratings.fide.com/profile/1109502/chart. > The elo-correction gave the player 72 points in March 2024. However between March 2024 and now, he gained another almost 200 points. He is 51 years and playing for decades chess. There are many such examples. How do you explain? Looking at the chart, what I see is that this player has been around 2050 consistently from 2007 to mid 2021. So I would rather ask what were the reasons of the fast drop between August 2021 (2041) and September 2023 (1787) and think about the recent raise as fixing the problem and returning to the previous ~2050 level.

@mkubecek said ^

Looking at the chart, what I see is that this player has been around 2050 consistently from 2007 to mid 2021. So I would rather ask what were the reasons of the fast drop between August 2021 (2041) and September 2023 (1787) and think about the recent raise as fixing the problem and returning to the previous ~2050 level.

Your countryman : https://ratings.fide.com/profile/306711/chart Same behavior but less big.
I can give many more examples. Each time an older player lost hundreds of rating points till March 2024 and then started a recover. How does this fit in the article of Vlad?

@mkubecek said [^](/forum/redirect/post/gTyZHz7O) > Looking at the chart, what I see is that this player has been around 2050 consistently from 2007 to mid 2021. So I would rather ask what were the reasons of the fast drop between August 2021 (2041) and September 2023 (1787) and think about the recent raise as fixing the problem and returning to the previous ~2050 level. Your countryman : https://ratings.fide.com/profile/306711/chart Same behavior but less big. I can give many more examples. Each time an older player lost hundreds of rating points till March 2024 and then started a recover. How does this fit in the article of Vlad?

@mvhk said ^

I can give many more examples. Each time an older player lost hundreds of rating points till March 2024 and then started a recover. How does this fit in the article of Vlad?

I don't think it does. As I said, if someone quickly dropped quite a bit below the level where they had been long term and then returned back to it, I would rather ask about the reasons of the drop and expect the raise to be a result of fixing that problem. In this particular case where the drop starts in 2021, an obvious hypothesis are problems related to covid/post-covid times.

@mvhk said [^](/forum/redirect/post/YLvzc6Xk) > I can give many more examples. Each time an older player lost hundreds of rating points till March 2024 and then started a recover. How does this fit in the article of Vlad? I don't think it does. As I said, if someone quickly dropped quite a bit below the level where they had been long term and then returned back to it, I would rather ask about the reasons of the drop and expect the raise to be a result of fixing that problem. In this particular case where the drop starts in 2021, an obvious hypothesis are problems related to covid/post-covid times.

The solution is to make an A tournament +2000 and a B tournament u2000 and if you ask me a C tournament too u1700. It's nonsense to play with a 1500 underrated kid when you are 2050.

The solution is to make an A tournament +2000 and a B tournament u2000 and if you ask me a C tournament too u1700. It's nonsense to play with a 1500 underrated kid when you are 2050.

@mvhk said ^

Looking at the chart, what I see is that this player has been around 2050 consistently from 2007 to mid 2021. So I would rather ask what were the reasons of the fast drop between August 2021 (2041) and September 2023 (1787) and think about the recent raise as fixing the problem and returning to the previous ~2050 level.

Your countryman : https://ratings.fide.com/profile/306711/chart Same behavior but less big.
I can give many more examples. Each time an older player lost hundreds of rating points till March 2024 and then started a recover. How does this fit in the article of Vlad?

I took a systemic view, rather than focusing too narrowly on isolated cases. Of course, this could be the basis of future work on the topic. For now, I welcome any such discussions and hypotheses. As far as me going back to code things and test, however, I think a break is in order. It's been a lot of work to publish this.

FIDE might find it a bit too "revolutionary" for their taste, and maybe the book operates in a very narrow market, but I think it's still valuable as a conversation starter, not so much as a "Let's do it this way!"

@mvhk said [^](/forum/redirect/post/YLvzc6Xk) > > Looking at the chart, what I see is that this player has been around 2050 consistently from 2007 to mid 2021. So I would rather ask what were the reasons of the fast drop between August 2021 (2041) and September 2023 (1787) and think about the recent raise as fixing the problem and returning to the previous ~2050 level. > > Your countryman : https://ratings.fide.com/profile/306711/chart Same behavior but less big. > I can give many more examples. Each time an older player lost hundreds of rating points till March 2024 and then started a recover. How does this fit in the article of Vlad? I took a systemic view, rather than focusing too narrowly on isolated cases. Of course, this could be the basis of future work on the topic. For now, I welcome any such discussions and hypotheses. As far as me going back to code things and test, however, I think a break is in order. It's been a lot of work to publish this. FIDE might find it a bit too "revolutionary" for their taste, and maybe the book operates in a very narrow market, but I think it's still valuable as a conversation starter, not so much as a "Let's do it this way!"

I have been looking forward to this coming out, thanks for sharing and the work it took to put this together! Will pick up the book asap

I have been looking forward to this coming out, thanks for sharing and the work it took to put this together! Will pick up the book asap

Very interesting, but I think I just suck.

Very interesting, but I think I just suck.

About 25 years ago, a FM told me that, when he was starting out, maybe early 1970s, a USCF 1000 player couldn't keep their pieces protected - it was a genuine "beginner" skill level rating.

About 25 years ago, a FM told me that, when he was starting out, maybe early 1970s, a USCF 1000 player couldn't keep their pieces protected - it was a genuine "beginner" skill level rating.