lichess.org
Donate

Chess with some data visualization

CM TBest

I love blog posts that are like this!

I do wonder if things like assuming 0 sec in pgn = premove will skew things. But I don't have a better suggestion.

I know you looked at classical, but would be interesting to see how rapid/blitz/bullet would compere.
Would also be interesting to see how increment classical comperes. Particularly in the time pressure graph. And also if things are the same across rating groups. (Don't think many titled players play 60 min classical through. )

Do we perform worse with pre-moves?

You can make safe premoves, (ie. a recapture. that will only play if it's legal after the opponent's move). It's probably a bit tricky to filter this out, but premoves with no/very low "risk" are unlikely to be blunders. Wonder if it would look the same if they were filtered out.

It is somewhat interesting to see the behavior changes with 30 seconds left and 10 seconds left. Maybe those are the milestones most people take into consideration when managing time.

More then "just" a milestone. There is a sound and visual ques in the UI for low time. See this thing https://github.com/Antiochian/chess-blunders/blob/main/README.md (that is also linked on database.lichess

I love blog posts that are like this! I do wonder if things like assuming 0 sec in pgn = premove will skew things. But I don't have a better suggestion. I know you looked at classical, but would be interesting to see how rapid/blitz/bullet would compere. Would also be interesting to see how increment classical comperes. Particularly in the time pressure graph. And also if things are the same across rating groups. (Don't think many titled players play 60 min classical through. ) >Do we perform worse with pre-moves? You can make safe premoves, (ie. a recapture. that will only play if it's legal after the opponent's move). It's probably a bit tricky to filter this out, but premoves with no/very low "risk" are unlikely to be blunders. Wonder if it would look the same if they were filtered out. >It is somewhat interesting to see the behavior changes with 30 seconds left and 10 seconds left. Maybe those are the milestones most people take into consideration when managing time. More then "just" a milestone. There is a sound and visual ques in the UI for low time. See this thing https://github.com/Antiochian/chess-blunders/blob/main/README.md (that is also linked on database.lichess
somethingpretentious

It would be interesting to look at differences by rating category in player behaviours.

I would also recommend referring to the games as 1+0 games as typically 60+0 would refer to a one hour game, but I see from your discussion of 30 and 10 seconds that it was bullet chess. (Although I know that the PGN tags use seconds rather than minutes.) I see that TBest made the same mistake I did at first assuming it is one hour long games.

I also agree with TBest that many moves will be under a second but not premoves.

It would be interesting to look at differences by rating category in player behaviours. I would also recommend referring to the games as 1+0 games as typically 60+0 would refer to a one hour game, but I see from your discussion of 30 and 10 seconds that it was bullet chess. (Although I know that the PGN tags use seconds rather than minutes.) I see that TBest made the same mistake I did at first assuming it is one hour long games. I also agree with TBest that many moves will be under a second but not premoves.
oortcloud_o

I edited the subtitle to 1+0 bullet games. Thank you for the feedback ill take into account when I write future posts.

I edited the subtitle to 1+0 bullet games. Thank you for the feedback ill take into account when I write future posts.
Lukasel

@oortcloud_o said in #4:

I edited the subtitle to 1+0 bullet games. Thank you for the feedback ill take into account when I write future posts.

Now it is inconsistent with 1+0 in the subtitle and 60+0 in the text.

Anyways, cool analysis idea, I appreciate your work and the effort put into making nice looking diagrams to share with us.

That the public database doesn't allow to distinguish between premoves and moves that were simply played very fast is really unfortunate. If we could distinguish them, we probably could have gained much more intersting insights.

Concerning the last two diagrams: You say "It is NOT the frequency of blunders on Move Number X". So, if you already suspected that people would expect that (I certainly did), why didn't you make such a diagram? And what would be the difference to "frequency of cases that happened for Move Number ≥ X"?
Which is a really difficult to understand description, by the way. It could use some more explanation.
I am still not entierly sure what these diagrams show exactly, but I think they might be skewed by the fact that every game has a move one, but a lot of games end before move 60. So could you clarify for me, what these diagrams show exactly?

@oortcloud_o said in #4: > I edited the subtitle to 1+0 bullet games. Thank you for the feedback ill take into account when I write future posts. Now it is inconsistent with 1+0 in the subtitle and 60+0 in the text. Anyways, cool analysis idea, I appreciate your work and the effort put into making nice looking diagrams to share with us. That the public database doesn't allow to distinguish between premoves and moves that were simply played very fast is really unfortunate. If we could distinguish them, we probably could have gained much more intersting insights. Concerning the last two diagrams: You say "It is NOT the frequency of blunders on Move Number X". So, if you already suspected that people would expect that (I certainly did), why didn't you make such a diagram? And what would be the difference to "frequency of cases that happened for Move Number ≥ X"? Which is a really difficult to understand description, by the way. It could use some more explanation. I am still not entierly sure what these diagrams show exactly, but I think they might be skewed by the fact that every game has a move one, but a lot of games end before move 60. So could you clarify for me, what these diagrams show exactly?
Motroskin

Thanks. A nice one.
Is there a chance to have a stat which players cheat more, those with digits in the nicknames or without them?
I have strong feeling that there are much more cheaters with the "numbered" names, will be interesting to see if this is true.

Thanks. A nice one. Is there a chance to have a stat which players cheat more, those with digits in the nicknames or without them? I have strong feeling that there are much more cheaters with the "numbered" names, will be interesting to see if this is true.
martian_chessedited

Thank you for the post! It is very interesting data. I was always curious about pre-move strategy.

I still have some questions to you:

  1. What was the rating of players played those games? Rating groups?
    I guess one needs to take some certain rating group (for example 2000) with a spread +/-50 to represent it.
    If it does not depend on the rating then it is just WOW!

  2. How many games you analysed to get these stats?

Thanks!

Thank you for the post! It is very interesting data. I was always curious about pre-move strategy. I still have some questions to you: 1. What was the rating of players played those games? Rating groups? I guess one needs to take some certain rating group (for example 2000) with a spread +/-50 to represent it. If it does not depend on the rating then it is just WOW! 2. How many games you analysed to get these stats? Thanks!
Illmaticguy

Nice job @oortcloud_o , very helpful tool for science and overall the same players. About the relation of mistakes with or withouth pre-move we can learn something about the humanity of the game,

I agree with putting a filter on the opening for obvious reasons, although it's tricky to set a standard (some openings go quickly to endgames, such as those involving early exchanges of queens). In any case, this can be dealt with separately, perhaps by looking at the case of opening missclicks.

Nice job @oortcloud_o , very helpful tool for science and overall the same players. About the relation of mistakes with or withouth pre-move we can learn something about the humanity of the game, I agree with putting a filter on the opening for obvious reasons, although it's tricky to set a standard (some openings go quickly to endgames, such as those involving early exchanges of queens). In any case, this can be dealt with separately, perhaps by looking at the case of opening missclicks.
FM CheckRaiseMate

Cool post! One thing to keep in mind with the relationship between pre-moves and blunders is that skilled bullet players choose moments to pre-move when a blunder is unlikely or impossible. For example, everyone knows to pre-move recaptures: the move will only go through if it becomes legal, in which case it's a good move. If you were forced to pre-move every move, you'd certainly blunder a lot more :)

Cool post! One thing to keep in mind with the relationship between pre-moves and blunders is that skilled bullet players choose moments to pre-move when a blunder is unlikely or impossible. For example, everyone knows to pre-move recaptures: the move will only go through if it becomes legal, in which case it's a good move. If you were forced to pre-move every move, you'd certainly blunder a lot more :)
<Comment deleted by user>