lichess.org
Donate

Why chess ratings don't mean what they used to

@Vlad_G92 said ^

Ok Vlad, interesting article and work, but remember to update your bio that says "FIDE rating: 2050" as it 2027 now. jk :)

thanks for your comment, dear 1650 rated Lichess player. jk ;)

It's because of Indian players

@Vlad_G92 said [^](/forum/redirect/post/epc29pvD) > > Ok Vlad, interesting article and work, but remember to update your bio that says "FIDE rating: 2050" as it 2027 now. jk :) > > thanks for your comment, dear 1650 rated Lichess player. jk ;) It's because of Indian players

@EmmettA said ^

there is no way to ever really know how many players are underrated or not, it’s only theoretical.

I disagree. There are many ways to know how many players are underrated. For example, you could use URS as an external validator, which computes ratings independently of the numbers listed on the FIDE profile of any player.

It would be quite interesting to make detailed statistics on a country-by-country basis when comparing the two datasets. If someone wants to follow up on my work, they could take the downloads from http://universalrating.com/ratings.php and run the numbers side by side.

That doesn’t prove anything though it’s still just theoretical. Just because your theory is based on numbers doesn’t mean it isn’t a theory. There are many factors that the data doesn’t reflect.

You have all the data and methodological information to not only replicate my "theory", but also improve upon it by drawing conclusions which are supported by data.

Normally, when you are trying to disprove any hypothesis, you need to present some serious evidence. Until you publish something that sits in counterpoint to the evidence presented so far, I am afraid your "theory" that the system would eventually reach equilibrium is a bold and unproven assumption.

@EmmettA said [^](/forum/redirect/post/jrIOhu6W) > > > there is no way to ever really know how many players are underrated or not, it’s only theoretical. > > > > I disagree. There are many ways to know how many players are underrated. For example, you could use URS as an external validator, which computes ratings independently of the numbers listed on the FIDE profile of any player. > > > > It would be quite interesting to make detailed statistics on a country-by-country basis when comparing the two datasets. If someone wants to follow up on my work, they could take the downloads from http://universalrating.com/ratings.php and run the numbers side by side. > > That doesn’t prove anything though it’s still just theoretical. Just because your theory is based on numbers doesn’t mean it isn’t a theory. There are many factors that the data doesn’t reflect. You have all the data and methodological information to not only replicate my "theory", but also improve upon it by drawing conclusions which are supported by data. Normally, when you are trying to disprove any hypothesis, you need to present some serious evidence. Until you publish something that sits in counterpoint to the evidence presented so far, I am afraid your "theory" that the system would eventually reach equilibrium is a bold and unproven assumption.

@Vlad_G92 said ^

there is no way to ever really know how many players are underrated or not, it’s only theoretical.

I disagree. There are many ways to know how many players are underrated. For example, you could use URS as an external validator, which computes ratings independently of the numbers listed on the FIDE profile of any player.

It would be quite interesting to make detailed statistics on a country-by-country basis when comparing the two datasets. If someone wants to follow up on my work, they could take the downloads from http://universalrating.com/ratings.php and run the numbers side by side.

That doesn’t prove anything though it’s still just theoretical. Just because your theory is based on numbers doesn’t mean it isn’t a theory. There are many factors that the data doesn’t reflect.

You have all the data and methodological information to not only replicate my "theory", but also improve upon it by drawing conclusions which are supported by data.

Normally, when you are trying to disprove any hypothesis, you need to present some serious evidence. Until you publish something that sits in counterpoint to the evidence presented so far, I am afraid your "theory" that the system would eventually reach equilibrium is a bold and unproven assumption.

Im not trying to disprove your theory , I actually enjoyed reading it, thank you for presenting it.

@Vlad_G92 said [^](/forum/redirect/post/alqeGbF9) > > > > there is no way to ever really know how many players are underrated or not, it’s only theoretical. > > > > > > I disagree. There are many ways to know how many players are underrated. For example, you could use URS as an external validator, which computes ratings independently of the numbers listed on the FIDE profile of any player. > > > > > > It would be quite interesting to make detailed statistics on a country-by-country basis when comparing the two datasets. If someone wants to follow up on my work, they could take the downloads from http://universalrating.com/ratings.php and run the numbers side by side. > > > > That doesn’t prove anything though it’s still just theoretical. Just because your theory is based on numbers doesn’t mean it isn’t a theory. There are many factors that the data doesn’t reflect. > > You have all the data and methodological information to not only replicate my "theory", but also improve upon it by drawing conclusions which are supported by data. > > Normally, when you are trying to disprove any hypothesis, you need to present some serious evidence. Until you publish something that sits in counterpoint to the evidence presented so far, I am afraid your "theory" that the system would eventually reach equilibrium is a bold and unproven assumption. Im not trying to disprove your theory , I actually enjoyed reading it, thank you for presenting it.

The point is that the rating does not reflect the actual level of skill. The rating is the number of points received based on the results of the betting. There is no unit of measurement for skill

The point is that the rating does not reflect the actual level of skill. The rating is the number of points received based on the results of the betting. There is no unit of measurement for skill

A prosa é 100% GPT... mas obrigado pela reunião e exposição dos dados!

A prosa é 100% GPT... mas obrigado pela reunião e exposição dos dados!

Very well researched post! Enjoyed reading it.

The rating system under FIDE is of course, rather flawed, but maybe this is because of some other factors too. Asian countries(with the notable exception of Mongolia and some others) generally have more population than European countries. Also chess as a sport is really popular in countries like India, where competitiveness is insane(I myself played there once and man, I won't forget those tough days).
Is there a way to keep an ostensibly "fair" rating system all over the globe? I'd opine no.
Guys, how do you think the current system can be improved? What steps can we take?

Again, great blog!

Very well researched post! Enjoyed reading it. The rating system under FIDE is of course, rather flawed, but maybe this is because of some other factors too. Asian countries(with the notable exception of Mongolia and some others) generally have more population than European countries. Also chess as a sport is really popular in countries like India, where competitiveness is insane(I myself played there once and man, I won't forget those tough days). Is there a way to keep an ostensibly "fair" rating system all over the globe? I'd opine no. Guys, how do you think the current system can be improved? What steps can we take? Again, great blog!

@Lord_of_the_River said ^

Very well researched post! Enjoyed reading it.

The rating system under FIDE is of course, rather flawed, but maybe this is because of some other factors too. Asian countries(with the notable exception of Mongolia and some others) generally have more population than European countries. Also chess as a sport is really popular in countries like India, where competitiveness is insane(I myself played there once and man, I won't forget those tough days).
Is there a way to keep an ostensibly "fair" rating system all over the globe? I'd opine no.
Guys, how do you think the current system can be improved? What steps can we take?

Again, great blog!

I think the first thing is to acknowledge that geography plays a much bigger role than FIDE or Prof. Elo intended and that reality has deviated excessively from the theoretical concept of a unified rating system.

The public discourse surrounding changes to the rating system (a historical overview which I treated in my book) has mostly dealt with overhauling the Elo system in favor of Glicko. Naturally, that comes with its own set of issues, for example granting title norms...

In my opinion, FIDE should encourage more data-driven discourse surrounding the rating system. Ultimately, whether they prioritize for interpretability or predictive accuracy is up to their leadership.

From my vantage point, the solutions that I offered are not only mathematically informed, but also "elegant" to a degree, in the sense they restore balance gradually, without a huge immediate shock.

If there's public interest, I can write a Part II of this preview and delve a bit deeper into the corrective mechanism I propose. Please react to this message just so I know that it reaches people.

@Lord_of_the_River said [^](/forum/redirect/post/fZYohZ4z) > Very well researched post! Enjoyed reading it. > > The rating system under FIDE is of course, rather flawed, but maybe this is because of some other factors too. Asian countries(with the notable exception of Mongolia and some others) generally have more population than European countries. Also chess as a sport is really popular in countries like India, where competitiveness is insane(I myself played there once and man, I won't forget those tough days). > Is there a way to keep an ostensibly "fair" rating system all over the globe? I'd opine no. > Guys, how do you think the current system can be improved? What steps can we take? > > Again, great blog! I think the first thing is to acknowledge that geography plays a much bigger role than FIDE or Prof. Elo intended and that reality has deviated excessively from the theoretical concept of a unified rating system. The public discourse surrounding changes to the rating system (a historical overview which I treated in my book) has mostly dealt with overhauling the Elo system in favor of Glicko. Naturally, that comes with its own set of issues, for example granting title norms... In my opinion, FIDE should encourage more data-driven discourse surrounding the rating system. Ultimately, whether they prioritize for interpretability or predictive accuracy is up to their leadership. From my vantage point, the solutions that I offered are not only mathematically informed, but also "elegant" to a degree, in the sense they restore balance gradually, without a huge immediate shock. If there's public interest, I can write a Part II of this preview and delve a bit deeper into the corrective mechanism I propose. Please react to this message just so I know that it reaches people.

My tournament experience is limited to the US, but I've found that even adults can be underrated here.

My tournament experience is limited to the US, but I've found that even adults can be underrated here.

@Vlad_G92

I mentioned you in my latest article. My next article will be more about rating itself so even better mapping your subject.
https://schaken-brabo.blogspot.com/2026/03/average-joe.html

Besides did you use AI-tools for doing the statistical analysis and if yes which ones? Obviously we are now having again (much) better AI-tools than in 2025.

@Vlad_G92 I mentioned you in my latest article. My next article will be more about rating itself so even better mapping your subject. https://schaken-brabo.blogspot.com/2026/03/average-joe.html Besides did you use AI-tools for doing the statistical analysis and if yes which ones? Obviously we are now having again (much) better AI-tools than in 2025.

@mvhk said ^

@Vlad_G92

I mentioned you in my latest article. My next article will be more about rating itself so even better mapping your subject.
https://schaken-brabo.blogspot.com/2026/03/average-joe.html

Besides did you use AI-tools for doing the statistical analysis and if yes which ones? Obviously we are now having again (much) better AI-tools than in 2025.

Yes, I used AI to process the datasets, mostly for data cleaning and storage. At several points the Claude Opus 4.5 model proved instrumental in writing some code for plotting the graphs.

Overall, this project would have taken me many more years to complete without any AI tool usage. I disclose upfront all AI assistance in the book, per modern publishing ethics.

Thanks for linking back to my blog entry. I am reading your article now!

@mvhk said [^](/forum/redirect/post/gXUDBuao) > @Vlad_G92 > > I mentioned you in my latest article. My next article will be more about rating itself so even better mapping your subject. > https://schaken-brabo.blogspot.com/2026/03/average-joe.html > > Besides did you use AI-tools for doing the statistical analysis and if yes which ones? Obviously we are now having again (much) better AI-tools than in 2025. Yes, I used AI to process the datasets, mostly for data cleaning and storage. At several points the Claude Opus 4.5 model proved instrumental in writing some code for plotting the graphs. Overall, this project would have taken me many more years to complete without any AI tool usage. I disclose upfront all AI assistance in the book, per modern publishing ethics. Thanks for linking back to my blog entry. I am reading your article now!