Your network blocks the Lichess assets!

lichess.org
Donate

Is Carlsen better after losing a game?

@Nikodimonius said in #14:

I think these stats could use some peer review
Otherwise there maybe some hidden other variables affecting victory odds

I'm sharing my posts mainly to discuss the ideas, so I see this as a kind of review process. Also, it's a lot of work to do these posts in the first place, so finding someone else who has the time and is interested enough to look over them before publication just isn't feasible.
I'm fully aware that my methods aren't perfect, so I'm always happy to discuss them afterwards and improve things for future posts.

Some obvious ideas that may or may not affect these results:

  1. Color. Win rate as black is usually smaller then win rate as white and in most tournaments you change your color (almost) every round. So you'll have a lot of white-black and black-white pairs of consecutive games and much smaller amount of white-white and black-black.

I posted about this in my previous comment, Carlsen has white in 55% of the games after he loses, but in classical he scores even better with black after a loss than with white.

  1. Swiss pairing. Don't know how often Magnus plays swiss tournaments compared to full-round tournaments, but in swiss you'll be paired lower after a loss thus impacting your win chances

I mentioned this as a problem for the faster time controls, as the rapid and blitz world championships are usually Swiss tournaments. One gets paired lower after a loss, but Carlsen will also be paired against weaker opposition (compared to round robin tournaments) in the first couple of rounds, so he has better chances of having a winning streak.
I'm unsure how to account for this, maybe splitting the data based on tournament format is an idea. But PGNs don't include much information about the tournament, so this would involve a lot of manual categorization of tournaments.
Another factor that I ignored is tournament standing, since Carlsen may be more likely to take risks if he has to fight for the tournament victory. But going through each game and considering the tournament standings is way too time consuming.

@Nikodimonius said in #14: > I think these stats could use some peer review > Otherwise there maybe some hidden other variables affecting victory odds I'm sharing my posts mainly to discuss the ideas, so I see this as a kind of review process. Also, it's a lot of work to do these posts in the first place, so finding someone else who has the time and is interested enough to look over them before publication just isn't feasible. I'm fully aware that my methods aren't perfect, so I'm always happy to discuss them afterwards and improve things for future posts. > Some obvious ideas that may or may not affect these results: > 1. Color. Win rate as black is usually smaller then win rate as white and in most tournaments you change your color (almost) every round. So you'll have a lot of white-black and black-white pairs of consecutive games and much smaller amount of white-white and black-black. I posted about this in my previous comment, Carlsen has white in 55% of the games after he loses, but in classical he scores even better with black after a loss than with white. > 2. Swiss pairing. Don't know how often Magnus plays swiss tournaments compared to full-round tournaments, but in swiss you'll be paired lower after a loss thus impacting your win chances I mentioned this as a problem for the faster time controls, as the rapid and blitz world championships are usually Swiss tournaments. One gets paired lower after a loss, but Carlsen will also be paired against weaker opposition (compared to round robin tournaments) in the first couple of rounds, so he has better chances of having a winning streak. I'm unsure how to account for this, maybe splitting the data based on tournament format is an idea. But PGNs don't include much information about the tournament, so this would involve a lot of manual categorization of tournaments. Another factor that I ignored is tournament standing, since Carlsen may be more likely to take risks if he has to fight for the tournament victory. But going through each game and considering the tournament standings is way too time consuming.

i think we should also compare this to a normal standard of a GM just to make sure we're not biasing the data. because if you think about it, in order to keep a high rating, they need to win. so i think comparing to a few other superGMs would give a good reference point.

i think we should also compare this to a normal standard of a GM just to make sure we're not biasing the data. because if you think about it, in order to keep a high rating, they need to win. so i think comparing to a few other superGMs would give a good reference point.

@soupycervanteskimchi said in #22:

i think we should also compare this to a normal standard of a GM just to make sure we're not biasing the data. because if you think about it, in order to keep a high rating, they need to win. so i think comparing to a few other superGMs would give a good reference point.

I think the point of using 2800 reference pool (it might be the blog in backreference to this one) might be doing that. Or are other superGMs of another strate (no clues, I do not follow that kind of chess, ignoramus that I am).

@soupycervanteskimchi said in #22: > i think we should also compare this to a normal standard of a GM just to make sure we're not biasing the data. because if you think about it, in order to keep a high rating, they need to win. so i think comparing to a few other superGMs would give a good reference point. I think the point of using 2800 reference pool (it might be the blog in backreference to this one) might be doing that. Or are other superGMs of another strate (no clues, I do not follow that kind of chess, ignoramus that I am).