How Opening Advantages Translate into Results in Online Games

Very nice, as usual!

This just proves what I've always known, that bullet is not chess! :-D

Seriously, now, I would generate some charts with the win rate on the y axis and the estimated win rate instead of the evaluation on the x axis. For the estimated win rate used by Lichess see Win% = 50 + 50 * (2 / (1 + exp(-0.00368208 * centipawns)) - 1).

And in that case you can add a shade under the line that would go from 0.5, 0.5 to 1, 1 and things would look much clearer.

Perhaps, since we see for blitz and above the conversion is similar, just conflate everything. Maybe show a line for increment games and one for non-increment games. Your charts show clearly that not the length of the game, but the one of the increment affects conversion most.

Also, without any proof or deep thinking, I have an intuition that this would look interesting on an exponential scale.

Again, good job, @jk_182 !

Very nice, as usual! This just proves what I've always known, that **bullet is not chess**! :-D Seriously, now, I would generate some charts with the win rate on the y axis and the estimated win rate instead of the evaluation on the x axis. For the estimated win rate used by Lichess see [Win% = 50 + 50 * (2 / (1 + exp(-0.00368208 * centipawns)) - 1)](https://lichess.org/page/accuracy). And in that case you can add a shade under the line that would go from 0.5, 0.5 to 1, 1 and things would look much clearer. Perhaps, since we see for blitz and above the conversion is similar, just conflate everything. Maybe show a line for increment games and one for non-increment games. Your charts show clearly that not the length of the game, but the one of the increment affects conversion most. Also, without any proof or deep thinking, I have an intuition that this would look interesting on an exponential scale. Again, good job, @jk_182 !

tpr

More interesting would be to evaluate 15+10 time control.
For bullet and blitz the time on the clock may be as important as the evaluation.
By definition +1 means an engine has 50% chance to win against itself.
This the result of +1 should be 50%*1 + 50%*0.5 = 0.75
Most of your curves are about there.

More interesting would be to evaluate 15+10 time control. For bullet and blitz the time on the clock may be as important as the evaluation. By definition +1 means an engine has 50% chance to win against itself. This the result of +1 should be 50%*1 + 50%*0.5 = 0.75 Most of your curves are about there.

mvhk

Your article confirms what I already wrote in 2022 that 2+1 generates better chess than 3+0 see https://schaken-brabo.blogspot.com/2022/07/increment.html.

An increment helps people to have a better time-management.

Your article confirms what I already wrote in 2022 that 2+1 generates better chess than 3+0 see https://schaken-brabo.blogspot.com/2022/07/increment.html. An increment helps people to have a better time-management.

Francesco11111

What the hack is Score? Pls describe your x and y values!

Ganador1789

Well But 2+1 Not It's Popular And 5+3 Too

DSE3009

@Francesco11111 said in #5:

What the hack is Score? Pls describe your x and y values!

score is the mean result: win is 1, draw is 1/2 and loss is 0

@Francesco11111 said in #5: > What the hack is Score? Pls describe your x and y values! score is the mean result: win is 1, draw is 1/2 and loss is 0

Manterror

Yes for rapid and classical please!

Cynis001

More proof that playing with increment is simply superior. :D
Would be interesting to see these graphs based on game termination, so that we can filter out the games that were only won on time. I mean, how often do you see players do crazy sacrifices in time scrambles to give an unexpected check and gain some more time to get the time win. Those games would then have a +5 or more for the losing site, and bias the results slightly.

More proof that playing with increment is simply superior. :D Would be interesting to see these graphs based on game termination, so that we can filter out the games that were only won on time. I mean, how often do you see players do crazy sacrifices in time scrambles to give an unexpected check and gain some more time to get the time win. Those games would then have a +5 or more for the losing site, and bias the results slightly.

mvhk

#10

@Cynis001 said in #9:

More proof that playing with increment is simply superior. :D
Would be interesting to see these graphs based on game termination, so that we can filter out the games that were only won on time. I mean, how often do you see players do crazy sacrifices in time scrambles to give an unexpected check and gain some more time to get the time win. Those games would then have a +5 or more for the losing site, and bias the results slightly.

I published last year on my blog https://schaken-brabo.blogspot.com/2024/06/vlag.html which responds to some of your questions.

@Cynis001 said in #9: > More proof that playing with increment is simply superior. :D > Would be interesting to see these graphs based on game termination, so that we can filter out the games that were only won on time. I mean, how often do you see players do crazy sacrifices in time scrambles to give an unexpected check and gain some more time to get the time win. Those games would then have a +5 or more for the losing site, and bias the results slightly. I published last year on my blog https://schaken-brabo.blogspot.com/2024/06/vlag.html which responds to some of your questions.