Candidates 2026 Predictions • page 1/4 • Community Blog Discussions • lichess.org

jk_182

Comments on https://lichess.org/@/jk_182/blog/candidates-2026-predictions/AvdkgvQP

RuyLopez1000

@jk_182

Thanks for the blog. Enjoyed reading.

@jk_182 Thanks for the blog. Enjoyed reading.

tpr

"I use only the FIDE ratings of the players as inputs"

Then how come Caruana with the lower rating has more wining chance than Nakamura with the higher rating?

"I use only the FIDE ratings of the players as inputs" * Then how come Caruana with the lower rating has more wining chance than Nakamura with the higher rating?

RuyLopez1000

edited

@tpr

"I use only the FIDE ratings of the players as inputs"

Then how come Caruana with the lower rating has more wining chance than Nakamura with the higher rating?

You should read the blog ;)

I made one slight adjustment to the ratings in the open section, namely I randomised Nakamura’s rating before each simulation to be somewhere between 2775 and 2805. I did this because he has hardly played any classical games since the last Candidates tournament and I thought adding this uncertainty to his rating is justified.

@tpr >"I use only the FIDE ratings of the players as inputs" >>Then how come Caruana with the lower rating has more wining chance than Nakamura with the higher rating? You should read the blog ;) >I made one slight adjustment to the ratings in the open section, namely I randomised Nakamura’s rating before each simulation to be somewhere between 2775 and 2805. I did this because he has hardly played any classical games since the last Candidates tournament and I thought adding this uncertainty to his rating is justified.

Toadofsky

What if Bluebaum wins Round 1? Round 2? How many rounds until he's the favorite?
https://youtu.be/MDr-SYhzwlA

What if Bluebaum wins Round 1? Round 2? How many rounds until he's the favorite? https://youtu.be/MDr-SYhzwlA

jk_182 edited

@Toadofsky said ^

What if Bluebaum wins Round 1? Round 2? How many rounds until he's the favorite?
https://youtu.be/MDr-SYhzwlA

It depends of course of what the other players are doing. Just for fun, I ran my simulation a couple of times where I gave Blübaum a head start to see how the probabilities change.

Blübaum Start Points	Blübaum Win Probability	Caruana Win Probability
0	0.9%	33.6%
1	5.1%	32.3%
2	18.6%	28.3 %
2.5	30%	24.4%

So he would need a head start of 2.5 points to become the favourite. But in reality it's much more complicated, as the players face each other. For example, if Blübaum beats both Caruana and Nakamura, he would probably already be the favourite, as there are fewer opportunities left for the favourites to make up points by beating Blübaum.

@Toadofsky said [^](/forum/redirect/post/Zt4VC5Gu) > What if Bluebaum wins Round 1? Round 2? How many rounds until he's the favorite? > https://youtu.be/MDr-SYhzwlA It depends of course of what the other players are doing. Just for fun, I ran my simulation a couple of times where I gave Blübaum a head start to see how the probabilities change. | Blübaum Start Points | Blübaum Win Probability | Caruana Win Probability | | ----------- | ----------- | -------- | | 0 | 0.9% | 33.6% | | 1 | 5.1% | 32.3% | | 2 | 18.6% | 28.3 % | | 2.5 | 30% | 24.4% | So he would need a head start of 2.5 points to become the favourite. But in reality it's much more complicated, as the players face each other. For example, if Blübaum beats both Caruana and Nakamura, he would probably already be the favourite, as there are fewer opportunities left for the favourites to make up points by beating Blübaum.

Toadofsky

Also of some interest: if he starts with 2 wins, that has a compounding effect since we then can infer that players' ratings may be inaccurate or that the players' choice of openings, seconds, etc. in addition to real-world effects like jet lag etc. which are all hidden variables may have a far stronger effect than players' ratings have.

Toadofsky

Given Sindarov's win over Caruana, what are everyone's chances now?

Seemingly chess is solved and players just need to memorize the solution: https://youtu.be/_FO-yZY4Nzs

Given Sindarov's win over Caruana, what are everyone's chances now? Seemingly chess is solved and players just need to memorize the solution: https://youtu.be/_FO-yZY4Nzs

cosmon

ELO per se does not capture momentum... Other factors like Fabi failing to convert (a lot) and Hikaru playing MM tournaments, plus Sindarov great form should have been taken in account. I am not saying he was an outright favorite... But he was definitely underestimated... And other players overestimated.

It looks like no one took any lesson from the previous candidate tournament...

:)

ELO per se does not capture momentum... Other factors like Fabi failing to convert (a lot) and Hikaru playing MM tournaments, plus Sindarov great form should have been taken in account. I am not saying he was an outright favorite... But he was definitely underestimated... And other players overestimated. It looks like no one took any lesson from the previous candidate tournament... :)

jk_182

@cosmon said ^

ELO per se does not capture momentum... Other factors like Fabi failing to convert (a lot) and Hikaru playing MM tournaments, plus Sindarov great form should have been taken in account. I am not saying he was an outright favorite... But he was definitely underestimated... And other players overestimated.

It looks like no one took any lesson from the previous candidate tournament...

:)

I agree that Elo isn't perfect, but as far as I know it's one of the best objective measures. The problem with including other factors is that it's difficult to weigh them and it's also difficult to test how well they actually work.
One counterexample to Sindarov is Blübaum, who also performed well leading up to the candidates and gained a lot of rating, but so far he performed to his rating expectation in the candidates.

As for momentum, I'm not aware of any studies in chess, but I know that in other sports hot streaks can mostly be explained by statistical variance. So a team isn't more likely to win a game just because they won a bunch of games before, winning streaks happen more or less randomly and a team that won more games than expected will probably then perform worse than expected, to land on their "true strength" on average.
One chess example would be Gukesh, who performed at a 2800+ level before he got the world title and now sits at 2730, and his "true strength" is probably somewhere in between.

Finally, I should add that it's very difficult to asses how well the model predicted the tournament based on the outcome. It just gives the likelihood for each player to win, but as the tournament will be only played once, we don't know what the correct probabilities should be. But one thing that's clear is that every player as an underdog agains the field, so it's always more likely that a given player doesn't win than that that player wins.

@cosmon said [^](/forum/redirect/post/5FBZEhE4) > ELO per se does not capture momentum... Other factors like Fabi failing to convert (a lot) and Hikaru playing MM tournaments, plus Sindarov great form should have been taken in account. I am not saying he was an outright favorite... But he was definitely underestimated... And other players overestimated. > > It looks like no one took any lesson from the previous candidate tournament... > > :) I agree that Elo isn't perfect, but as far as I know it's one of the best objective measures. The problem with including other factors is that it's difficult to weigh them and it's also difficult to test how well they actually work. One counterexample to Sindarov is Blübaum, who also performed well leading up to the candidates and gained a lot of rating, but so far he performed to his rating expectation in the candidates. As for momentum, I'm not aware of any studies in chess, but I know that in other sports hot streaks can mostly be explained by statistical variance. So a team isn't more likely to win a game just because they won a bunch of games before, winning streaks happen more or less randomly and a team that won more games than expected will probably then perform worse than expected, to land on their "true strength" on average. One chess example would be Gukesh, who performed at a 2800+ level before he got the world title and now sits at 2730, and his "true strength" is probably somewhere in between. Finally, I should add that it's very difficult to asses how well the model predicted the tournament based on the outcome. It just gives the likelihood for each player to win, but as the tournament will be only played once, we don't know what the correct probabilities should be. But one thing that's clear is that every player as an underdog agains the field, so it's always more likely that a given player doesn't win than that that player wins.