Chessqueen wrote: ↑Sat Jan 15, 2022 12:41 am
How could GM Ben Finegold is expected to score better than GM Smerdon at Knight Odds who is rated higher and played with an inferior version of Komodo ? https://ratings.fide.com/profile/3202305
What are you talking about? Smerdon scored 5 out of 6. Who expects Finegold to do better than that? I mean it's possible he might win every game, but surely no one is predicting that. If he scores 4 out of 6 that would match the last two GMs vs Dragon, percentage-wise. If Dragon can score 2.5 out of 6 that would be a new record percentage vs. GM at knight odds in Rapid.
Well I just talked to some of GM Finegold students and they are saying that GM Finegold will score better than GM Smerdon, I told his students we will see about that tomorrow.
Did they clearly know that Smerdon won by 5 to 1 and that they were therefore predicting 5.5 or 6 out of 6, or were they just saying that without even knowing the original result? Some people express opinions on things without even knowing the relevant facts.
I really do NOT think they know what was the actual score, but I will let them know, in a few minutes, GM Finegold might get a 3.5 out of 6 if he is lucky tomorrow.
Chessqueen wrote: ↑Sat Jan 15, 2022 12:41 am
How could GM Ben Finegold is expected to score better than GM Smerdon at Knight Odds who is rated higher and played with an inferior version of Komodo ? https://ratings.fide.com/profile/3202305
What are you talking about? Smerdon scored 5 out of 6. Who expects Finegold to do better than that? I mean it's possible he might win every game, but surely no one is predicting that. If he scores 4 out of 6 that would match the last two GMs vs Dragon, percentage-wise. If Dragon can score 2.5 out of 6 that would be a new record percentage vs. GM at knight odds in Rapid.
Well I just talked to some of GM Finegold students and they are saying that GM Finegold will score better than GM Smerdon, I told his students we will see about that tomorrow.
Did they clearly know that Smerdon won by 5 to 1 and that they were therefore predicting 5.5 or 6 out of 6, or were they just saying that without even knowing the original result? Some people express opinions on things without even knowing the relevant facts.
I really do NOT think they know what was the actual score, but I will let them know, in a few minutes, GM Finegold might get a 3.5 out of 6 if he is lucky tomorrow.
I found out why his students are saying that GM Finegold will score better, they overheard him saying that with Knight Odds he will NOT lose any games, so they assumed that he was going to crush Komodo Dragon, It is possible that he will not lose any games, which means that there will be lots of draws, but today we will see what happens.
Chessqueen wrote: ↑Sat Jan 15, 2022 12:41 am
How could GM Ben Finegold is expected to score better than GM Smerdon at Knight Odds who is rated higher and played with an inferior version of Komodo ? https://ratings.fide.com/profile/3202305
What are you talking about? Smerdon scored 5 out of 6. Who expects Finegold to do better than that? I mean it's possible he might win every game, but surely no one is predicting that. If he scores 4 out of 6 that would match the last two GMs vs Dragon, percentage-wise. If Dragon can score 2.5 out of 6 that would be a new record percentage vs. GM at knight odds in Rapid.
Well I just talked to some of GM Finegold students and they are saying that GM Finegold will score better than GM Smerdon, I told his students we will see about that tomorrow.
Did they clearly know that Smerdon won by 5 to 1 and that they were therefore predicting 5.5 or 6 out of 6, or were they just saying that without even knowing the original result? Some people express opinions on things without even knowing the relevant facts.
I really do NOT think they know what was the actual score, but I will let them know, in a few minutes, GM Finegold might get a 3.5 out of 6 if he is lucky tomorrow.
I found out why his students are saying that GM Finegold will score better, they overheard him saying that with Knight Odds he will NOT lose any games, so they assumed that he was going to crush Komodo Dragon, It is possible that he will not lose any games, which means that there will be lots of draws, but today we will see what happens.
So GM Finegold is still holding what he said that he will not lose a single game with a Knight Odds
Note: I Believe that a T/C of 10/5 is more appropriated for Komodo Dragon to even up against a GM higher than 2600.
So the final result was 4 to 2 for Finegold (3 wins, 1 loss, 2 draws), exactly the same percentage as the two previous GMs scored with knight odds in Rapid (Anthony Wirig and Eugene Perelshteyn). First two games were with standard mode (I wasn't convinced that MCTS would be a good choice with Komodo set to play blitz), which we lost rather badly, but the final four were with MCTS which made an even score. I don't know if that is just chance or reflects something real, but at least it's pretty clear that MCTS plays quite well even in blitz, so probably I should stick with it for such matches. Komodo scored better with b1 odds (1 to 1) than with g1 odds (1 to 3), but I think it got better positions generally with g1 odds. So for the third straight time, although Dragon lost the match, it managed to win a game and to draw or win at least half of the games. Ben used most of his time in all games, so it seems likely that a significantly shorter time limit might have resulted in a drawn or won match for Dragon. The last game was quite dramatic, when Dragon sacrificed material for what looked like a winning attack, but it was only a drawing line in fact.
lkaufman wrote: ↑Sun Jan 16, 2022 5:20 am
So the final result was 4 to 2 for Finegold (3 wins, 1 loss, 2 draws), exactly the same percentage as the two previous GMs scored with knight odds in Rapid (Anthony Wirig and Eugene Perelshteyn). First two games were with standard mode (I wasn't convinced that MCTS would be a good choice with Komodo set to play blitz), which we lost rather badly, but the final four were with MCTS which made an even score. I don't know if that is just chance or reflects something real, but at least it's pretty clear that MCTS plays quite well even in blitz, so probably I should stick with it for such matches. Komodo scored better with b1 odds (1 to 1) than with g1 odds (1 to 3), but I think it got better positions generally with g1 odds. So for the third straight time, although Dragon lost the match, it managed to win a game and to draw or win at least half of the games. Ben used most of his time in all games, so it seems likely that a significantly shorter time limit might have resulted in a drawn or won match for Dragon. The last game was quite dramatic, when Dragon sacrificed material for what looked like a winning attack, but it was only a drawing line in fact.
I wonder if those GMs that you gave Komodo like Smerdon or GM Finegold etc... if they decide to p[ay more games in the future with Knight Odds by themselves if they know how to set the correct Contemp. So My recommendation is for the next version of Komodo Dragon x.x.x when players of different strengths set up a Queen, Rook or Knight odds that Komodo automatically set the correct Contempt values just like with the UCI_ELO or Limit Playing Stength_ELO setting for different Fide players of version 2.6.1. that automatically set the correct Parameters.
lkaufman wrote: ↑Sun Jan 16, 2022 5:20 am
So the final result was 4 to 2 for Finegold (3 wins, 1 loss, 2 draws), exactly the same percentage as the two previous GMs scored with knight odds in Rapid (Anthony Wirig and Eugene Perelshteyn). First two games were with standard mode (I wasn't convinced that MCTS would be a good choice with Komodo set to play blitz), which we lost rather badly, but the final four were with MCTS which made an even score. I don't know if that is just chance or reflects something real, but at least it's pretty clear that MCTS plays quite well even in blitz, so probably I should stick with it for such matches. Komodo scored better with b1 odds (1 to 1) than with g1 odds (1 to 3), but I think it got better positions generally with g1 odds. So for the third straight time, although Dragon lost the match, it managed to win a game and to draw or win at least half of the games. Ben used most of his time in all games, so it seems likely that a significantly shorter time limit might have resulted in a drawn or won match for Dragon. The last game was quite dramatic, when Dragon sacrificed material for what looked like a winning attack, but it was only a drawing line in fact.
Or it might be fatige. I saw him in the last game being a bit exhausted
lkaufman wrote: ↑Sun Jan 16, 2022 5:20 am
So the final result was 4 to 2 for Finegold (3 wins, 1 loss, 2 draws), exactly the same percentage as the two previous GMs scored with knight odds in Rapid (Anthony Wirig and Eugene Perelshteyn). First two games were with standard mode (I wasn't convinced that MCTS would be a good choice with Komodo set to play blitz), which we lost rather badly, but the final four were with MCTS which made an even score. I don't know if that is just chance or reflects something real, but at least it's pretty clear that MCTS plays quite well even in blitz, so probably I should stick with it for such matches. Komodo scored better with b1 odds (1 to 1) than with g1 odds (1 to 3), but I think it got better positions generally with g1 odds. So for the third straight time, although Dragon lost the match, it managed to win a game and to draw or win at least half of the games. Ben used most of his time in all games, so it seems likely that a significantly shorter time limit might have resulted in a drawn or won match for Dragon. The last game was quite dramatic, when Dragon sacrificed material for what looked like a winning attack, but it was only a drawing line in fact.
Or it might be fatige. I aw him in the last game being a bit exhausted
Sure, that could be a factor, but note that he didn't lose the final game. Perelshteyn also started well but couldn't maintain the pace. I won't say that fatigue isn't a factor, but playing sessions of 3 to 4 hours are quite normal for grandmasters, even in Rapid play, so it's a fair test in that the GM should be playing roughly as well as he would in a human Rapid tournament. In other words, some fatigue is just a normal part of competitive chess.
lkaufman wrote: ↑Sun Jan 16, 2022 5:20 am
So the final result was 4 to 2 for Finegold (3 wins, 1 loss, 2 draws), exactly the same percentage as the two previous GMs scored with knight odds in Rapid (Anthony Wirig and Eugene Perelshteyn). First two games were with standard mode (I wasn't convinced that MCTS would be a good choice with Komodo set to play blitz), which we lost rather badly, but the final four were with MCTS which made an even score. I don't know if that is just chance or reflects something real, but at least it's pretty clear that MCTS plays quite well even in blitz, so probably I should stick with it for such matches. Komodo scored better with b1 odds (1 to 1) than with g1 odds (1 to 3), but I think it got better positions generally with g1 odds. So for the third straight time, although Dragon lost the match, it managed to win a game and to draw or win at least half of the games. Ben used most of his time in all games, so it seems likely that a significantly shorter time limit might have resulted in a drawn or won match for Dragon. The last game was quite dramatic, when Dragon sacrificed material for what looked like a winning attack, but it was only a drawing line in fact.
Or it might be fatige. I aw him in the last game being a bit exhausted
Sure, that could be a factor, but note that he didn't lose the final game. Perelshteyn also started well but couldn't maintain the pace. I won't say that fatigue isn't a factor, but playing sessions of 3 to 4 hours are quite normal for grandmasters, even in Rapid play, so it's a fair test in that the GM should be playing roughly as well as he would in a human Rapid tournament. In other words, some fatigue is just a normal part of competitive chess.
Another curiosity if you knew that the MCTS plays better than standard from previous GMs Knight Odds, why you did NOT use MCTS for all the 6 games and Knight g1 Odds for at least 4 games ?
lkaufman wrote: ↑Sun Jan 16, 2022 5:20 am
So the final result was 4 to 2 for Finegold (3 wins, 1 loss, 2 draws), exactly the same percentage as the two previous GMs scored with knight odds in Rapid (Anthony Wirig and Eugene Perelshteyn). First two games were with standard mode (I wasn't convinced that MCTS would be a good choice with Komodo set to play blitz), which we lost rather badly, but the final four were with MCTS which made an even score. I don't know if that is just chance or reflects something real, but at least it's pretty clear that MCTS plays quite well even in blitz, so probably I should stick with it for such matches. Komodo scored better with b1 odds (1 to 1) than with g1 odds (1 to 3), but I think it got better positions generally with g1 odds. So for the third straight time, although Dragon lost the match, it managed to win a game and to draw or win at least half of the games. Ben used most of his time in all games, so it seems likely that a significantly shorter time limit might have resulted in a drawn or won match for Dragon. The last game was quite dramatic, when Dragon sacrificed material for what looked like a winning attack, but it was only a drawing line in fact.
Or it might be fatige. I aw him in the last game being a bit exhausted
Sure, that could be a factor, but note that he didn't lose the final game. Perelshteyn also started well but couldn't maintain the pace. I won't say that fatigue isn't a factor, but playing sessions of 3 to 4 hours are quite normal for grandmasters, even in Rapid play, so it's a fair test in that the GM should be playing roughly as well as he would in a human Rapid tournament. In other words, some fatigue is just a normal part of competitive chess.
You should arrange another match Knight Odds against GM Finegold at T/C 10/5. if he agrees to that shorter T/C ?
lkaufman wrote: ↑Sun Jan 16, 2022 5:20 am
So the final result was 4 to 2 for Finegold (3 wins, 1 loss, 2 draws), exactly the same percentage as the two previous GMs scored with knight odds in Rapid (Anthony Wirig and Eugene Perelshteyn). First two games were with standard mode (I wasn't convinced that MCTS would be a good choice with Komodo set to play blitz), which we lost rather badly, but the final four were with MCTS which made an even score. I don't know if that is just chance or reflects something real, but at least it's pretty clear that MCTS plays quite well even in blitz, so probably I should stick with it for such matches. Komodo scored better with b1 odds (1 to 1) than with g1 odds (1 to 3), but I think it got better positions generally with g1 odds. So for the third straight time, although Dragon lost the match, it managed to win a game and to draw or win at least half of the games. Ben used most of his time in all games, so it seems likely that a significantly shorter time limit might have resulted in a drawn or won match for Dragon. The last game was quite dramatic, when Dragon sacrificed material for what looked like a winning attack, but it was only a drawing line in fact.
Or it might be fatige. I aw him in the last game being a bit exhausted
Sure, that could be a factor, but note that he didn't lose the final game. Perelshteyn also started well but couldn't maintain the pace. I won't say that fatigue isn't a factor, but playing sessions of 3 to 4 hours are quite normal for grandmasters, even in Rapid play, so it's a fair test in that the GM should be playing roughly as well as he would in a human Rapid tournament. In other words, some fatigue is just a normal part of competitive chess.
Another curiosity if you knew that the MCTS plays better than standard from previous GMs Knight Odds, why you did NOT use MCTS for all the 6 games and Knight g1 Odds for at least 4 games ?
Well, MCTS plays better vs. humans with Rapid time control, but here Komodo was giving 5 to 1 time odds as well (playing blitz vs Rapid), and I wasn't convinced that it would do better playing blitz with MCTS than with standard mode. It does worse against "simulated humans" when playing MCTS. But I guess I'm convinced now. We did give knight g1 odds in four of the six games.
