I mean he's 74 Elo stronger. A pure ratings-based simulation is NOT sufficient for a match where the players have specifically been prepping for each other for months, but across 14 games at that rating difference the pure math puts the favorite at 89.6% to win, 3.4% to lose, with a 7% chance it goes to rapid tiebreaks. I am not claiming Magnus has *those* chances (if you assume the odds in rapid are roughly the same as in classical, that works out to roughly 96% to win the match). I agree Nepo is a tough opponent, but even a 30 point Elo edge works out to 80% to win the match. It's my feeling that if you quantify the difference between the two, even after adjusting for the impact of prep, unique aspects of their individual matchup, etc... that placed on the Elo scale it's a gap of more than 30 rating points.

If you think that's too high, I'd be curious to know what you think the win/draw/loss odds are for an individual classical game when each player has white. I'm wondering if your different assessment comes from a very different game-by-game estimate than I have, or if you have a lower estimation of how much a 14-game match compounds a game-by-game edge than my spreadsheet does (both might be valid).

The later K v K WC matches were close despite the ELO difference (one of them even ended tied).

The rating formula assumed the results of each individual game are independent, but this is not always the case, especially in matches. Preparation, style, psychology, and form all play a role to make the results not so independent.

I had a conversation with a guy who was on Fabiano’s team in the last championship - he said “60%” - but I don’t think that was anything other than his hot take when put on the spot. But it is coming from someone who participated as a second, so he has unique insight on the event.

party time for MOMO Firouzja on Twitch.

Alireza is the Future Present!

I actually cried a little when he won. I haven't been on the right side of history much before, but I'm sure it was made today.

Alireza gonna win the candidates too.

Go Alireza!!!

BBJ - any softball questions you want lobbed your way for the PCP interview?

Can't think of anything in particular. I like answering the questions people are interested in, hard questions are more fun!

Cool. I sent a question or two in - looking forward to the interview!

Thanks, me too! I'm very curious to see what kind of questions I get

Ali v Caruana v all should be fun come candidates time. Ali having one of the alltime great runs like Fabi back in STL but that doesn't connote destiny at winning the toughest tournament there is, I mean it wasn't that long ago Ali botched a trivial K+P endgame v Magnus. Pressure is very different in a 14 rd double r-r vs a Swiss or team event. No Jobava's in the Candidates.

BJJ, I agree with Tim that match results are not 'independently' occuring games. And Nepo is a bad stylistic matchup for him. Let's not forget Magnus is only +1 in classical time in his last two defenses. And, certainly, 'luck' can always play a role in these things. All it takes is one of your prep lines to have a mistake that leads to a loss and you're -1 and having to rebuild on the fly under the greatest pressure.

I do have Magnus as a 2-1 fave or thereabouts.

For the record I finalized my projections and ultimately settled pretty close to where the bookies are at. I have Magnus at 81.5%, and will be interviewing with Ben Johnson of Perpetual Chess tomorrow (episode probably to be released Tuesday?) to go over all my reasoning.

Firouzja now the youngest 2800 in history, beating Magnus by half a year. He might be entering the candidates as a rating favourite, which one year ago sounded absolutely insane. In the candidates I would still rank Fabi higher since he has so much experience in those tourneys and that does account for something.

Would be fun to see if Ding and Fabi both go over 2800 by the time candidates comes around.

Thinking more about this point, I guess I should explicitly clarify that I also agree the games are not independently occurring in a true statistical sense. The thing is, I don't know in what way the games being related to each other actually affects the odds. I can easily run probabilities as if the games were independent. Once I've done so, I don't have an adjustment I can make that I think makes those probabilities more accurate. I don't have a better way to model things that feels like an improvement.

If Nepo wins the first game that puts pressure on Magnus, but it also gives him more incentive to play for a win when he's already the stronger player. He could crack under the pressure or he could be motivated by it. And meanwhile if Nepo is suddenly leading the match that puts a kind of pressure on him too. Does he become risk averse, to try to make sure he doesn't give up the lead? Does that work or does it backfire?

I will have my baseline odds for game 2 (once colors are assigned) from treating as if it's independent. I agree circumstances will change after game 1 is complete, and those odds may shift in some way, but I'm not sure how. And I don't have any data that tells me how. So I'm just not convinced the lack of independence between games really creates a problem for my model.

Agree there is not enough data to make an objective model here. Maybe the two Ks are the only ones in recent history who had enough match games to do so. But we do know the direction of the adjustment. The challenger's chances are certainly better than the pure ELO prediction.

Great interview!!

Yep. Maybe 26 points better like I went with in my analysis. Maybe more, maybe less, maybe it depends on the challenger...

Did you consider the possibility of reducing the effective number of games, instead of the rating difference? It probably doesn't matter either way when in both cases there is no objective way to adjust it.

It seems to me a good strategy for an underdog would be to reduce the effective number of games by making as many safe draws as possible, while waiting for a favorable spot to go all in. Basically trying to make the match outcome depend on a single game. But this would be a terrible strategy against Magnus, since this would be playing to his style. Those safe draws might not be so safe.

Fun fact: In KK3 and KK5, Kasparov clinched the title by scoring 12 points after 23 games, yet the 24th game was still played in both cases. (There were no tie breaks, the champion retained the title in a 12-12 tie.

Exactly.

You can get better than 3/1 for Nepo.