13 years of computer chess: knight odds

[Uri Blass]

[snip](the score of) Komodo over Rybka 2.3.2a giving two pawns was 27 to 13, so plus 127 Elo for Komodo.[/snip]

That's a performance of 3150, giving f2 & c2, on 4 cores at 2'+1". The matches'll be on a 24 core machine. Doesn't bode well for the 2500 GM's.

Based on this logic komodo can beat every GM at blitz with knight odds because it score 50% against chess engines that are better than every GM at blitz.

I guess that
Komodo without a knight also can get something like 2700 fide rating in 90+30 if it play against humans based on this logic.

I guess that this logic is wrong and chess programs of the past become extremely weak when they get a big material advantage and play clearly below GM level.

[Laskos]

Very interesting. I assume when you talk about comtempt in Komodo, you mean the negative of Drawscore in Komodo 9.1, since I don't think you have the Komodo with actual Contempt (anti-simplification). My final result of 27 to 13 for Komodo over Rybka 2.3.2a giving two pawns was 27 to 13, so plus 127 Elo for Komodo. Your results around the plus 80 mark against Naum 4.6 are even better since Naum 4.6 outrates Rybka 2.3.2a by much more than 47 elo, which is probably due to your using a faster tc. Based on this, it really does look like Komodo can spot Rybka 3 two pawns in bullet chess, which I find incredible. I'm searching for an explanation but nothing comes to mind easily. I haven't been able to get Rybka 3 to work on our tester for some unknown reason to verify this. We can now run handicap matches on our own tester, which is much better than trying to run them in ChessBase for various reasons.
One discovery I made today is that it seems that it is much harder to give a handicap to a weaker version of yourself than to an unrelated player. In bullet chess (1' +.5",2 threads), Komodo (latest) giving knight odds to Komodo 8 managed only thee draws in 404 games, which implies a rating equivalence for a knight of well over a thousand elo. But the earlier tests put the value of knight odds at about 700 elo against unrelated engines, although the details of time limit, threads, and contempt were different. Probably it's very difficult to give a handicap to a weaker clone since you will generally agree that everything is bad for the odds-giver, but with widely different engines there is more chance for surprise and for wild, unclear positions. Probably self-play (i.e. playing crippled versions of yourself) is a better predictor of how things will go with a human, since the human will also avoid the chaotic positions you often see between unrelated engines.
Finally, I got better results giving knight odds to Nebula 2 both without contempt and with contempt 100 than with 150. So probably we'll reduce the contempt setting for this weekend. I think 150 makes Komodo play too passively for fear of exchanges.

Yes, I used the simple Drawscore (opposite to Contempt) in Komodo 9.1, but called the feature with the opposite sign generally "Contempt" because Naum has an equivalent term "DrawContemptScore" having the opposite sign to Komodo's 9.1 term, and it was becoming confusing.

I have also used today Komodo's self-play for c2 and f2 pawns handicap, and self-play is probably even more relevant than finding a weaker engine, as it seems to dilate the necessary difference and has a larger draw rate, suggesting a better quality of games. Al ultra-fast in self-play I seem to need 450 ELO points difference for c2 and f2 pawns handicap with time handicap factor of 6-7, while against a different engine I seem to need close to 350 ELO points difference. The draw rate is about 50% in self-play compared to 30% against a weaker engine. Then, one cannot say that shorter time control Komodo 9.1 is lacking knowledge.

In games Komodo 9.1 at 18''+0.18'' on two cores (handicapped by 2 pawns Komodo LTC) against Komodo 9.1 at 3''+0.03'' on two cores (handicapped by a factor of 6 in time control Komodo STC) using different Drawscores, I am again getting clear results showing that Komodo needs a positive contempt against a human, and that the focus on Contempt is important, as it can give a hefty ELO difference even using such basic feature as Drawscore.

500 games each match.

Komodo LTC Drawscore=0
Komodo STC Drawscore=0
Komodo LTC: -21 ELO points

The human (Komodo STC) can improve by going for draws:

Komodo LTC Drawscore=0
Komodo STC Drawscore=50
Komodo LTC: -65 ELO points

But Komodo LTC (computer) strikes back by avoiding draws, and in fact the swing is spectacular for such a simple feature as Drawscore:

Komodo LTC Drawscore=-75
Komodo STC Drawscore=50
Komodo LTC: +12 ELO points (Drawscore=-75 adds 77 ELO points)

I guess using a more involved "Contempt" factor is worth the effort, as to me its importance is unexpectedly high, especially in self-games. It is probably important against a human too.

[lkaufman]

[snip](the score of) Komodo over Rybka 2.3.2a giving two pawns was 27 to 13, so plus 127 Elo for Komodo.[/snip]

That's a performance of 3150, giving f2 & c2, on 4 cores at 2'+1". The matches'll be on a 24 core machine. Doesn't bode well for the 2500 GM's.

Based on this logic komodo can beat every GM at blitz with knight odds because it score 50% against chess engines that are better than every GM at blitz.

I guess that
Komodo without a knight also can get something like 2700 fide rating in 90+30 if it play against humans based on this logic.

I guess that this logic is wrong and chess programs of the past become extremely weak when they get a big material advantage and play clearly below GM level.

I don't think the old programs become "extremely weak", but rather they just play normal chess, trying their hardest to improve the score by one centipawn even if it means huge complications. That is of course an inferior strategy for winning when way ahead. Self-play seems to be more relevant for predicting results vs. humans. I've started a match with new Komodo giving Komodo 5.1 (the first mp version) the two pawn handicap. This is way too big a handicap, as K 5.1 is much stronger than Rybka 2.3.2a, but I'm looking for the elo gap. So far new Komodo has managed only one draw out of seven games, so it really is quite a large handicap.

[lkaufman]

Very interesting. I assume when you talk about comtempt in Komodo, you mean the negative of Drawscore in Komodo 9.1, since I don't think you have the Komodo with actual Contempt (anti-simplification). My final result of 27 to 13 for Komodo over Rybka 2.3.2a giving two pawns was 27 to 13, so plus 127 Elo for Komodo. Your results around the plus 80 mark against Naum 4.6 are even better since Naum 4.6 outrates Rybka 2.3.2a by much more than 47 elo, which is probably due to your using a faster tc. Based on this, it really does look like Komodo can spot Rybka 3 two pawns in bullet chess, which I find incredible. I'm searching for an explanation but nothing comes to mind easily. I haven't been able to get Rybka 3 to work on our tester for some unknown reason to verify this. We can now run handicap matches on our own tester, which is much better than trying to run them in ChessBase for various reasons.
One discovery I made today is that it seems that it is much harder to give a handicap to a weaker version of yourself than to an unrelated player. In bullet chess (1' +.5",2 threads), Komodo (latest) giving knight odds to Komodo 8 managed only thee draws in 404 games, which implies a rating equivalence for a knight of well over a thousand elo. But the earlier tests put the value of knight odds at about 700 elo against unrelated engines, although the details of time limit, threads, and contempt were different. Probably it's very difficult to give a handicap to a weaker clone since you will generally agree that everything is bad for the odds-giver, but with widely different engines there is more chance for surprise and for wild, unclear positions. Probably self-play (i.e. playing crippled versions of yourself) is a better predictor of how things will go with a human, since the human will also avoid the chaotic positions you often see between unrelated engines.
Finally, I got better results giving knight odds to Nebula 2 both without contempt and with contempt 100 than with 150. So probably we'll reduce the contempt setting for this weekend. I think 150 makes Komodo play too passively for fear of exchanges.

Yes, I used the simple Drawscore (opposite to Contempt) in Komodo 9.1, but called the feature with the opposite sign generally "Contempt" because Naum has an equivalent term "DrawContemptScore" having the opposite sign to Komodo's 9.1 term, and it was becoming confusing.

I have also used today Komodo's self-play for c2 and f2 pawns handicap, and self-play is probably even more relevant than finding a weaker engine, as it seems to dilate the necessary difference and has a larger draw rate, suggesting a better quality of games. Al ultra-fast in self-play I seem to need 450 ELO points difference for c2 and f2 pawns handicap with time handicap factor of 6-7, while against a different engine I seem to need close to 350 ELO points difference. The draw rate is about 50% in self-play compared to 30% against a weaker engine. Then, one cannot say that shorter time control Komodo 9.1 is lacking knowledge.

In games Komodo 9.1 at 18''+0.18'' on two cores (handicapped by 2 pawns Komodo LTC) against Komodo 9.1 at 3''+0.03'' on two cores (handicapped by a factor of 6 in time control Komodo STC) using different Drawscores, I am again getting clear results showing that Komodo needs a positive contempt against a human, and that the focus on Contempt is important, as it can give a hefty ELO difference even using such basic feature as Drawscore.

500 games each match.

Komodo LTC Drawscore=0
Komodo STC Drawscore=0
Komodo LTC: -21 ELO points

The human (Komodo STC) can improve by going for draws:

Komodo LTC Drawscore=0
Komodo STC Drawscore=50
Komodo LTC: -65 ELO points

But Komodo LTC (computer) strikes back by avoiding draws, and in fact the swing is spectacular for such a simple feature as Drawscore:

Komodo LTC Drawscore=-75
Komodo STC Drawscore=50
Komodo LTC: +12 ELO points (Drawscore=-75 adds 77 ELO points)

I guess using a more involved "Contempt" factor is worth the effort, as to me its importance is unexpectedly high, especially in self-games. It is probably important against a human too.

Again, very interesting. I don't think the higher draw rate for self-play has anything to do with quality in this case, it's just that similar eval makes for more boring play and more draws. I think what's going on with your tests is that when STC seeks draws and LTC is neutral, once the score drops below +50 they both want to draw and can often find a repetition to achieve that goal. If LTC doesn't get to keep playing when the score gets close, how can it win? But the tests do show that even simple Contempt has big consequences.

[Uri Blass]

[snip](the score of) Komodo over Rybka 2.3.2a giving two pawns was 27 to 13, so plus 127 Elo for Komodo.[/snip]

That's a performance of 3150, giving f2 & c2, on 4 cores at 2'+1". The matches'll be on a 24 core machine. Doesn't bode well for the 2500 GM's.

Based on this logic komodo can beat every GM at blitz with knight odds because it score 50% against chess engines that are better than every GM at blitz.

I guess that
Komodo without a knight also can get something like 2700 fide rating in 90+30 if it play against humans based on this logic.

I guess that this logic is wrong and chess programs of the past become extremely weak when they get a big material advantage and play clearly below GM level.

I don't think the old programs become "extremely weak", but rather they just play normal chess, trying their hardest to improve the score by one centipawn even if it means huge complications. That is of course an inferior strategy for winning when way ahead. Self-play seems to be more relevant for predicting results vs. humans. I've started a match with new Komodo giving Komodo 5.1 (the first mp version) the two pawn handicap. This is way too big a handicap, as K 5.1 is much stronger than Rybka 2.3.2a, but I'm looking for the elo gap. So far new Komodo has managed only one draw out of seven games, so it really is quite a large handicap.

Playing (what you consider to be normal chess) when you have a big material advantage is a mistake regardless of the opponent.

It is better to win slowly and safely and not to take risks regardless of the opponent.

I believe that it is also counter productive against humans.

Fritz could not win a winning position against Ruslan Ponomariov and lost the game.

http://www.chessgames.com/perl/chessgame?gid=1371049

[Laskos]

Very interesting. I assume when you talk about comtempt in Komodo, you mean the negative of Drawscore in Komodo 9.1, since I don't think you have the Komodo with actual Contempt (anti-simplification). My final result of 27 to 13 for Komodo over Rybka 2.3.2a giving two pawns was 27 to 13, so plus 127 Elo for Komodo. Your results around the plus 80 mark against Naum 4.6 are even better since Naum 4.6 outrates Rybka 2.3.2a by much more than 47 elo, which is probably due to your using a faster tc. Based on this, it really does look like Komodo can spot Rybka 3 two pawns in bullet chess, which I find incredible. I'm searching for an explanation but nothing comes to mind easily. I haven't been able to get Rybka 3 to work on our tester for some unknown reason to verify this. We can now run handicap matches on our own tester, which is much better than trying to run them in ChessBase for various reasons.
One discovery I made today is that it seems that it is much harder to give a handicap to a weaker version of yourself than to an unrelated player. In bullet chess (1' +.5",2 threads), Komodo (latest) giving knight odds to Komodo 8 managed only thee draws in 404 games, which implies a rating equivalence for a knight of well over a thousand elo. But the earlier tests put the value of knight odds at about 700 elo against unrelated engines, although the details of time limit, threads, and contempt were different. Probably it's very difficult to give a handicap to a weaker clone since you will generally agree that everything is bad for the odds-giver, but with widely different engines there is more chance for surprise and for wild, unclear positions. Probably self-play (i.e. playing crippled versions of yourself) is a better predictor of how things will go with a human, since the human will also avoid the chaotic positions you often see between unrelated engines.
Finally, I got better results giving knight odds to Nebula 2 both without contempt and with contempt 100 than with 150. So probably we'll reduce the contempt setting for this weekend. I think 150 makes Komodo play too passively for fear of exchanges.

Yes, I used the simple Drawscore (opposite to Contempt) in Komodo 9.1, but called the feature with the opposite sign generally "Contempt" because Naum has an equivalent term "DrawContemptScore" having the opposite sign to Komodo's 9.1 term, and it was becoming confusing.

I have also used today Komodo's self-play for c2 and f2 pawns handicap, and self-play is probably even more relevant than finding a weaker engine, as it seems to dilate the necessary difference and has a larger draw rate, suggesting a better quality of games. Al ultra-fast in self-play I seem to need 450 ELO points difference for c2 and f2 pawns handicap with time handicap factor of 6-7, while against a different engine I seem to need close to 350 ELO points difference. The draw rate is about 50% in self-play compared to 30% against a weaker engine. Then, one cannot say that shorter time control Komodo 9.1 is lacking knowledge.

In games Komodo 9.1 at 18''+0.18'' on two cores (handicapped by 2 pawns Komodo LTC) against Komodo 9.1 at 3''+0.03'' on two cores (handicapped by a factor of 6 in time control Komodo STC) using different Drawscores, I am again getting clear results showing that Komodo needs a positive contempt against a human, and that the focus on Contempt is important, as it can give a hefty ELO difference even using such basic feature as Drawscore.

500 games each match.

Komodo LTC Drawscore=0
Komodo STC Drawscore=0
Komodo LTC: -21 ELO points

The human (Komodo STC) can improve by going for draws:

Komodo LTC Drawscore=0
Komodo STC Drawscore=50
Komodo LTC: -65 ELO points

But Komodo LTC (computer) strikes back by avoiding draws, and in fact the swing is spectacular for such a simple feature as Drawscore:

Komodo LTC Drawscore=-75
Komodo STC Drawscore=50
Komodo LTC: +12 ELO points (Drawscore=-75 adds 77 ELO points)

I guess using a more involved "Contempt" factor is worth the effort, as to me its importance is unexpectedly high, especially in self-games. It is probably important against a human too.

Again, very interesting. I don't think the higher draw rate for self-play has anything to do with quality in this case, it's just that similar eval makes for more boring play and more draws. I think what's going on with your tests is that when STC seeks draws and LTC is neutral, once the score drops below +50 they both want to draw and can often find a repetition to achieve that goal. If LTC doesn't get to keep playing when the score gets close, how can it win? But the tests do show that even simple Contempt has big consequences.

Higher quality in the sense there are fewer surprises and spotted blunders, the games are more orderly. The draw rate against a GM will probably be pretty high. How many games against GM there will be? What time control? Maybe you will post a poll like you did at knight odds .

Yes, they will agree for a draw from -0.50 to 0.00, and that will happen in the games against a human too. If white wants to draw or win, it has to pass from -1.50 starting position to a point (say starting with -0.50) where a reasonable human opponent will eagerly avoid complications and draw. The contempt becomes very important in such a match, even a simple drawscore one. As you add "avoid simplifications" and such, that will have even more impact upon a human.

[lkaufman]

The plan is to play six games over two days at 45' + 15", exactly as in the ongoing knight-odds match. Only the handicap is changed.

[Laskos]

The plan is to play six games over two days at 45' + 15", exactly as in the ongoing knight-odds match. Only the handicap is changed.

I tried self-play to see time odds necessary for Komodo 9.1 in these conditions, I got them equal at x100 time handicap, i.e. Komodo at 45'+15'' (drawscore=-75) to 27''+0.15'' (drawscore=+50) with 2 draws, 1 win and 1 loss. That would mean 500 ELO points handicap or, saying that Komodo 9.1 on a strong PC is around 3200 FIDE ELO in LTC, this handicap is a good match for a 2700 GM. So, the engine guess would be that Komodo will perform fairly well against 2500-2600 GM, but this is a bit hand-waving argument, engine play un-adapted to handicap games can be misleading. Also, a high draw rate is to be expected. Here are the games at times 100 time control handicap:

[pgn][Event "?"]
[Site "?"]
[Date "2015.07.31"]
[Round "1"]
[White "Komodo 9.1 2700+15"]
[Black "Komodo 9.1 27+0.15"]
[Result "1/2-1/2"]
[BlackTimeControl "27+0.15"]
[FEN "rnbqkbnr/ppp1pppp/8/3p4/8/5N2/PP1PP1PP/RNBQKB1R w KQkq -"]
[PlyCount "141"]
[SetUp "1"]
[WhiteTimeControl "2700+15"]

1. Nc3 {-1.67/22 83s} e5 {+1.52/14 0.73s} 2. Nxe5 {-1.60/23 83s}
d4 {+1.25/15 1.9s} 3. Qa4+ {-1.56/25 78s} c6 {+1.81/14 0.34s}
4. Nd1 {-1.58/24 78s} b5 {+1.74/15 1.5s} 5. Qc2 {-1.26/21 32s}
Qd5 {+1.22/14 1.8s} 6. Nf3 {-0.97/24 33s} Na6 {+0.88/16 2.8s}
7. a3 {-0.88/25 51s} Nc5 {+1.25/16 0.82s} 8. b4 {-0.88/26 71s}
Bf5 {+0.94/16 0.39s} 9. d3 {-0.68/25 63s} Nb3 {+1.00/17 0.46s}
10. Rb1 {-0.60/25 94s} Nxc1 {+0.90/16 0.83s} 11. Qxc1 {-0.74/23 48s}
Bd7 {+0.96/17 0.70s} 12. g3 {-0.60/23 31s} Rc8 {+0.52/14 1.3s}
13. Bg2 {-0.59/23 47s} c5 {+0.55/14 0.76s} 14. Qd2 {-0.57/25 155s}
Nf6 {+0.76/14 0.61s} 15. e3 {-0.62/23 43s} dxe3 {+0.41/15 1.0s}
16. Nxe3 {-0.33/22 33s} Qd6 {+0.67/15 0.51s} 17. bxc5 {-0.39/24 56s}
Qxc5 {+0.88/13 0.31s} 18. d4 {-0.18/22 33s} Qxa3 {+0.15/14 0.68s}
19. Ne5 {-0.15/23 40s} Be6 {+0.18/14 0.38s} 20. O-O {-0.25/23 42s}
Bd6 {+0.15/15 0.70s} 21. Nc6 {-0.17/25 93s} Qa6 {+0.84/12 0.18s}
22. Ra1 {-0.08/26 74s} Qb6 {-0.09/13 0.18s} 23. Nxa7 {-0.08/24 27s}
Rc7 {+0.31/15 0.49s} 24. Qa5 {-0.14/28 155s} Qxa5 {+0.20/15 0.21s}
25. Rxa5 {-0.14/24 20s} Bb4 {+0.17/17 0.75s} 26. Raa1 {-0.17/24 23s}
O-O {+0.39/15 0.46s} 27. Nxb5 {-0.17/24 65s} Rcc8 {+0.32/15 0.51s}
28. Rf2 {-0.17/25 45s} Rb8 {+0.10/12 0.15s} 29. Nc7 {-0.18/26 45s}
Bc3 {+0.21/15 0.45s} 30. Rd1 {-0.22/27 55s} Rfd8 {+0.17/15 0.49s}
31. Nxe6 {-0.27/27 51s} fxe6 {+0.25/13 0.18s} 32. Nc2 {-0.31/28 63s}
Rd6 {+0.19/15 0.63s} 33. Re2 {-0.20/26 62s} Kf7 {+0.26/13 0.36s}
34. Bh3 {-0.23/25 64s} Rbb6 {+0.26/14 0.36s} 35. Rd3 {-0.22/27 58s}
Bb2 {+0.21/15 0.85s} 36. Rf2 {-0.18/25 79s} Kg8 {+0.27/15 0.93s}
37. Bg2 {-0.21/24 86s} e5 {+0.34/14 0.25s} 38. d5 {-0.17/26 30s}
g6 {+0.22/12 0.17s} 39. Re2 {-0.17/28 62s} Rb5 {+0.15/16 0.56s}
40. h3 {-0.17/29 88s} Kg7 {+0.25/16 0.69s} 41. Kh2 {-0.14/28 42s}
e4 {+0.20/18 0.36s} 42. Bxe4 {-0.15/28 29s} Nxe4 {+0.21/18 0.25s}
43. Rxe4 {-0.13/25 19s} Rbxd5 {+0.14/18 0.59s} 44. Rxd5 {-0.13/28 37s}
Rxd5 {+0.16/17 0.17s} 45. Re2 {-0.12/29 61s} Bc3 {+0.16/17 0.30s}
46. Ne3 {-0.10/26 37s} Re5 {+0.13/16 0.49s} 47. Kg2 {-0.09/27 20s}
Bd4 {+0.12/17 0.37s} 48. Kf3 {-0.11/27 23s} Bc5 {+0.14/17 0.42s}
49. Nc4 {-0.04/30 48s} Rxe2 {+0.14/20 0.30s} 50. Kxe2 {-0.05/32 19s}
Kf6 {+0.14/17 0.20s} 51. Kf3 {-0.07/36 36s} Ba7 {+0.08/18 0.29s}
52. h4 {-0.07/40 53s} Ke6 {+0.08/17 0.17s} 53. g4 {-0.07/37 49s}
Bd4 {+0.07/17 0.40s} 54. Ne3 {-0.07/38 45s} Bf6 {+0.06/19 0.28s}
55. g5 {0.00/39 21s} Bb2 {+0.05/20 0.23s} 56. Ke4 {0.00/39 36s}
Ke7 {+0.05/19 0.17s} 57. Nd5+ {0.00/38 41s} Kd6 {+0.05/20 0.14s}
58. Nf4 {0.00/39 26s} Bc1 {+0.03/21 0.22s} 59. Nd3 {0.00/37 38s}
Bd2 {+0.05/23 0.27s} 60. h5 {0.00/36 22s} gxh5 {+0.03/17 0.20s}
61. Nf4 {0.00/44 47s} Bxf4 {+0.03/21 0.25s} 62. Kxf4 {0.00/48 15s}
Ke6 {+0.03/25 0.21s} 63. Ke4 {0.00/52 25s} h4 {+0.03/26 0.13s}
64. Kf4 {0.00/52 18s} h3 {+0.03/27 0.17s} 65. Kf3 {0.00/52 18s}
h2 {0.00/26 0.16s} 66. Kg2 {0.00/54 26s} Kf5 {0.00/31 0.26s}
67. Kxh2 {0.00/59 28s} Kf4 {0.00/32 0.23s} 68. Kh3 {0.00/62 73s}
Kf5 {0.00/33 0.36s} 69. Kh2 {0.00/99 17s} Ke5 {0.00/35 0.31s}
70. Kg3 {0.00/61 19s} Kf5 {0.00/35 0.27s}
71. Kh2 {0.00/99 9.2s, Draw by 3-fold repetition} 1/2-1/2[/pgn]

[pgn][Event "?"]
[Site "?"]
[Date "2015.07.31"]
[Round "3"]
[White "Komodo 9.1 2700+15"]
[Black "Komodo 9.1 27+0.15"]
[Result "0-1"]
[BlackTimeControl "27+0.15"]
[FEN "rnbqkbnr/pppp1ppp/8/4p3/8/2N5/PP1PP1PP/R1BQKBNR w KQkq -"]
[PlyCount "148"]
[SetUp "1"]
[WhiteTimeControl "2700+15"]

1. Nf3 {-1.64/23 69s} Nc6 {+1.62/14 0.55s} 2. Qa4 {-1.74/23 77s}
Nf6 {+1.70/15 1.4s} 3. e3 {-1.92/24 78s} a6 {+1.66/16 1.6s}
4. Be2 {-1.84/24 147s} d6 {+1.53/15 0.51s} 5. O-O {-1.77/26 97s}
Be7 {+1.53/16 0.88s} 6. d4 {-1.82/26 92s} e4 {+1.70/17 0.85s}
7. Nh4 {-1.88/26 117s} O-O {+1.74/16 0.28s} 8. Nf5 {-1.87/25 133s}
Bxf5 {+1.60/16 0.57s} 9. Rxf5 {-1.93/22 28s} Qd7 {+1.72/17 1.1s}
10. Rf1 {-1.75/24 53s} b5 {+1.72/15 0.37s} 11. Qc2 {-1.78/23 50s}
d5 {+1.77/16 1.2s} 12. Bd2 {-1.88/24 56s} Nb4 {+1.80/16 1.4s}
13. Qb3 {-2.08/21 40s} Nd3 {+1.71/16 0.71s} 14. Bxd3 {-1.89/23 82s}
exd3 {+1.66/15 0.47s} 15. Rac1 {-1.96/25 141s} Rac8 {+1.70/15 1.2s}
16. Nd1 {-1.86/23 41s} c5 {+1.79/15 0.44s} 17. Qxd3 {-1.88/23 35s}
Rfe8 {+1.89/14 0.29s} 18. a3 {-1.87/25 110s} Qe6 {+1.88/13 0.46s}
19. Nc3 {-1.92/23 60s} g6 {+1.88/14 0.83s} 20. h3 {-1.88/20 26s}
Bd6 {+1.95/14 0.76s} 21. dxc5 {-2.11/24 80s} Bxc5 {+1.76/14 0.37s}
22. Ne2 {-2.11/24 44s} Qe5 {+2.08/13 0.21s} 23. Nd4 {-2.33/27 93s}
Ne4 {+1.88/14 0.62s} 24. Bb4 {-2.14/23 74s} Bxb4 {+2.00/15 0.47s}
25. axb4 {-2.04/24 30s} Rxc1 {+1.81/16 1.5s} 26. Rxc1 {-2.09/25 14s}
Qg3 {+1.92/16 0.35s} 27. Qe2 {-2.17/26 30s} Ng5 {+2.02/16 0.38s}
28. Qe1 {-2.04/27 28s} Qxe3+ {+1.81/17 0.72s} 29. Qxe3 {-2.12/23 16s}
Rxe3 {+1.95/16 0.46s} 30. Rd1 {-2.20/27 90s} Ne6 {+2.12/18 1.2s}
31. Kf2 {-2.26/29 91s} Re4 {+2.06/17 0.43s} 32. Nc6 {-2.19/27 80s}
Nc7 {+2.14/18 2.0s} 33. Rc1 {-2.18/27 70s} Rc4 {+2.22/18 0.69s}
34. Rxc4 {-1.83/27 47s} bxc4 {+2.35/15 0.18s} 35. Ke3 {-2.18/29 29s}
f6 {+2.15/16 0.17s} 36. Kd4 {-2.19/27 23s} Kf7 {+1.52/16 0.45s}
37. Kc5 {-2.09/27 33s} Ke6 {+1.45/16 0.30s} 38. Kb6 {-2.46/27 102s}
Kd6 {+1.92/16 0.36s} 39. Na5 {-2.49/27 70s} Kd7 {+2.17/16 0.71s}
40. Nc6 {-2.32/26 82s} h5 {+2.18/15 0.13s} 41. Nb8+ {-2.04/23 23s}
Kd6 {+1.81/14 0.17s} 42. Nxa6 {-2.62/23 18s} Ne8 {+1.97/15 0.41s}
43. b5 {-3.12/26 40s} d4 {+2.49/15 0.18s} 44. Ka5 {-3.01/28 22s}
d3 {+2.96/16 0.24s} 45. b6 {-3.95/27 20s} Kc6 {+2.90/18 0.35s}
46. Nb4+ {-3.57/29 28s} Kb7 {+2.71/19 0.30s} 47. Nd5 {-3.62/30 33s}
Nd6 {+2.84/20 0.28s} 48. Kb4 {-4.25/31 55s} f5 {+2.47/16 0.29s}
49. Nc3 {-4.53/30 58s} Kxb6 {+2.74/16 0.22s} 50. b3 {-4.75/32 38s}
cxb3 {+3.52/18 0.20s} 51. Na4+ {-6.68/35 71s} Kc6 {+4.14/21 0.49s}
52. Kxb3 {-9.43/36 49s} d2 {+4.32/19 0.13s} 53. Kc2 {-9.04/34 117s}
Ne4 {+4.76/21 0.42s} 54. Nb2 {-10.18/31 29s} Kd5 {+4.78/21 0.18s}
55. Kd3 {-10.92/32 30s} g5 {+5.40/20 0.66s} 56. Ke3 {-7.23/28 95s}
f4+ {+6.60/17 0.27s} 57. Kd3 {-18.37/25 28s} g4 {+6.85/19 0.24s}
58. hxg4 {-17.85/25 80s} hxg4 {+6.73/20 0.14s} 59. Nd1 {-M54/24 3.9s}
Ke5 {+6.97/19 0.097s} 60. Nb2 {-10.04/29 71s} f3 {+6.92/20 0.25s}
61. g3 {-M64/22 22s} Nxg3 {+10.49/14 0.12s} 62. Kxd2 {-31.84/21 5.5s}
Kd4 {+17.76/16 0.29s} 63. Nd3 {-M76/19 2.8s} Ne4+ {+M51/15 0.17s}
64. Kc2 {-M50/19 1.2s} g3 {+M49/14 0.035s} 65. Nb4 {-M24/18 2.6s}
f2 {+M43/13 0.040s} 66. Nc6+ {-M16/19 1.7s} Ke3 {+M27/14 0.029s}
67. Kb3 {-M16/19 2.6s} g2 {+M19/13 0.072s} 68. Nb4 {-M14/20 2.3s}
f1=Q {+M13/13 0.050s} 69. Nc2+ {-M12/23 2.0s} Kd2 {+M11/14 0.032s}
70. Na3 {-M10/28 2.1s} g1=Q {+M9/15 0.041s} 71. Nc4+ {-M8/99 0.24s}
Kc1 {+M7/21 0.031s} 72. Ka3 {-M6/99 0.058s} Qd3+ {+M5/50 0.029s}
73. Kb4 {-M4/99 0.008s} Qc5+ {+M3/99 0.008s} 74. Ka4 {-M2/99 0.006s}
Qdxc4# {+M1/99 0.007s, Black mates} 0-1[/pgn]

[pgn][Event "?"]
[Site "?"]
[Date "2015.07.31"]
[Round "5"]
[White "Komodo 9.1 2700+15"]
[Black "Komodo 9.1 27+0.15"]
[Result "1/2-1/2"]
[BlackTimeControl "27+0.15"]
[FEN "rnbqkbnr/ppp1pppp/8/3p4/3P4/8/PP2P1PP/RNBQKBNR w KQkq -"]
[PlyCount "319"]
[SetUp "1"]
[WhiteTimeControl "2700+15"]

1. Bf4 {-1.50/23 117s} e5 {+1.39/13 1.1s} 2. dxe5 {-1.42/22 68s}
g5 {+1.36/13 2.1s} 3. Bg3 {-1.42/20 37s} Nc6 {+1.23/13 0.77s}
4. e4 {-1.43/23 85s} dxe4 {+1.26/14 1.5s} 5. Nc3 {-1.38/23 92s}
Qxd1+ {+1.26/15 1.2s} 6. Rxd1 {-1.50/22 26s} Bb4 {+1.38/14 0.32s}
7. Bb5 {-1.47/23 97s} Bxc3+ {+1.48/15 0.82s} 8. bxc3 {-1.42/22 25s}
Ne7 {+1.43/14 0.32s} 9. Ne2 {-1.43/22 68s} Bd7 {+1.34/14 1.9s}
10. O-O {-1.48/23 84s} a6 {+1.23/13 0.41s} 11. Bxc6 {-1.42/23 65s}
Nxc6 {+1.19/14 1.1s} 12. Nd4 {-1.45/26 213s} Nxd4 {+1.28/14 0.32s}
13. cxd4 {-1.17/23 57s} Rd8 {+1.01/14 0.53s} 14. d5 {-1.13/25 48s}
Bb5 {+1.27/15 1.1s} 15. Rf6 {-1.07/23 39s} e3 {+1.17/14 0.61s}
16. d6 {-1.10/25 59s} O-O {+1.19/15 1.0s} 17. h4 {-1.01/26 56s}
e2 {+0.94/14 0.72s} 18. Re1 {-1.19/27 67s} cxd6 {+0.89/16 0.52s}
19. exd6 {-1.16/26 27s} gxh4 {+1.06/15 0.31s} 20. Bxh4 {-1.64/24 34s}
Rfe8 {+1.02/17 0.84s} 21. Rh6 {-1.48/27 34s} Kg7 {+0.99/17 1.1s}
22. Bxd8 {-1.22/27 55s} Kxh6 {+1.02/17 0.20s} 23. Be7 {-1.27/29 23s}
Kg6 {+1.04/18 0.41s} 24. Rxe2 {-1.21/31 80s} Bxe2 {+1.07/16 0.29s}
25. d7 {-1.21/31 19s} Rg8 {+1.08/19 0.27s} 26. d8=Q {-1.25/35 38s}
Rxd8 {+1.11/19 0.16s} 27. Bxd8 {-1.24/34 19s} Kf5 {+1.08/19 0.31s}
28. Kf2 {-1.33/37 64s} Bc4 {+1.03/17 0.27s} 29. a3 {-1.32/36 32s}
Ke4 {+1.14/18 0.88s} 30. g4 {-1.29/36 124s} Kf4 {+1.17/21 1.1s}
31. g5 {-1.33/36 38s} b5 {+1.03/22 0.64s} 32. Kg2 {-1.29/36 61s}
Ke3 {+1.04/21 0.60s} 33. Ba5 {-1.13/37 32s} Be6 {+1.03/22 0.57s}
34. Kg3 {-1.13/36 44s} Kd4 {+1.16/23 0.39s} 35. Kf2 {-1.69/35 80s}
Kc4 {+1.14/23 0.50s} 36. Bb4 {-1.74/35 39s} Kb3 {+1.13/23 0.34s}
37. Ke3 {-2.19/34 77s} a5 {+1.13/25 0.38s} 38. Bd6 {-2.27/36 53s}
Kc4 {+1.25/22 0.40s} 39. Kd2 {-2.33/32 39s} Bf5 {+1.24/22 0.39s}
40. Bc7 {-1.86/37 33s} b4 {+1.51/23 0.51s} 41. Bxa5 {-1.85/36 50s}
bxa3 {+1.85/28 0.41s} 42. Bc3 {-1.86/38 34s} a2 {+1.83/28 0.48s}
43. Bf6 {-1.82/39 28s} Kb3 {+1.83/28 0.51s} 44. Bc3 {-1.75/41 47s}
Bb1 {+1.83/27 0.34s} 45. Bf6 {-1.88/43 47s} Bg6 {+1.83/28 0.58s}
46. Kd1 {-1.80/44 135s} Be4 {+1.83/29 0.49s} 47. Kd2 {-1.86/40 38s}
Bg2 {+1.83/29 0.27s} 48. Kd1 {-1.82/41 28s} Bb7 {+1.83/27 0.36s}
49. Kc1 {-1.81/42 22s} Bd5 {+1.82/28 0.35s} 50. Kd2 {-1.70/40 21s}
Bc6 {+1.80/28 0.34s} 51. Kc1 {-1.75/40 24s} Bd7 {+1.80/29 0.38s}
52. Kd2 {-1.76/38 15s} Be6 {+1.80/28 0.25s} 53. Kd1 {-1.77/39 28s}
Bf5 {+1.80/28 0.24s} 54. Kc1 {-1.77/42 28s} Bc8 {+1.74/27 0.63s}
55. Kd2 {-1.65/39 21s} Bg4 {+1.75/25 0.17s} 56. Kc1 {-1.71/40 28s}
Be2 {+1.75/26 0.29s} 57. Kd2 {-1.76/41 53s} Bf1 {+1.75/25 0.37s}
58. Kc1 {-1.67/41 35s} Bd3 {+1.76/27 0.33s} 59. Kd2 {-1.65/39 17s}
Bb5 {+1.76/27 0.42s} 60. Kd1 {-1.80/39 20s} Ba6 {+1.78/26 0.32s}
61. Kd2 {-1.63/36 14s} Bc8 {+1.77/26 0.26s} 62. Kd1 {-1.57/38 23s}
Be6 {+1.77/27 0.28s} 63. Kd2 {-1.81/41 71s} Bd7 {+1.76/25 0.20s}
64. Kd1 {-1.35/38 48s} Bb5 {+1.77/25 0.28s} 65. Kc1 {-1.43/37 22s}
Bc6 {+1.75/25 0.12s} 66. Kd2 {-1.39/36 15s} Be4 {+1.67/21 0.13s}
67. Kd1 {-1.35/37 19s} Bg6 {+1.67/23 0.22s} 68. Kd2 {-1.34/41 28s}
Kb4 {+1.60/23 0.30s} 69. Kc1 {-1.35/41 26s} Ka3 {+1.59/23 0.21s}
70. Kd1 {-1.34/39 12s} Be4 {+1.52/20 0.15s} 71. Kd2 {-1.35/41 22s}
Ka4 {+1.67/20 0.37s} 72. Kc1 {-1.39/40 15s} Kb4 {+1.12/21 0.11s}
73. Kb2 {-1.38/45 24s} Bb1 {+1.41/23 0.11s} 74. Bc3+ {-1.38/37 14s}
Kc5 {+1.39/23 0.087s} 75. Bd2 {-1.36/37 13s} Kd4 {+1.39/24 0.12s}
76. Bc1 {-1.38/40 15s} Ke5 {+1.39/24 0.15s} 77. Ka1 {-1.35/36 14s}
f5 {+1.31/24 0.27s} 78. gxf6 {-1.30/36 12s} Kxf6 {+1.34/22 0.066s}
79. Bd2 {-1.23/37 12s} Ke5 {+1.21/22 0.15s} 80. Kb2 {-1.35/37 13s}
Kd4 {+1.31/24 0.25s} 81. Bc3+ {-1.19/38 17s} Kd3 {+1.24/23 0.14s}
82. Bf6 {-1.08/38 15s} Kd2 {+1.32/22 0.090s} 83. Bg5+ {-1.37/45 32s}
Ke1 {+1.32/24 0.11s} 84. Bh4+ {-1.38/37 13s} Kd1 {+1.31/26 0.26s}
85. Bd8 {-1.25/40 28s} h5 {+1.34/22 0.056s} 86. Bg5 {-1.09/39 15s}
Ke2 {+1.37/24 0.16s} 87. Be7 {-1.04/39 11s} Ke3 {+1.25/26 0.33s}
88. Bf6 {-1.07/41 18s} Kd3 {+1.20/25 0.12s} 89. Bg5 {-1.16/38 13s}
Kc4 {+1.20/25 0.066s} 90. Bf6 {-1.03/40 18s} Kb4 {+1.20/24 0.12s}
91. Be7+ {-1.09/41 12s} Kb5 {+1.19/26 0.11s} 92. Bg5 {-1.03/40 16s}
Kc5 {+1.17/25 0.18s} 93. Be3+ {-1.06/41 14s} Kd6 {+1.17/26 0.15s}
94. Bg5 {-1.09/38 11s} Ke6 {+1.17/26 0.13s} 95. Ka1 {-0.91/39 16s}
Ke5 {+1.07/25 0.17s} 96. Bd8 {-0.97/38 11s} Ke4 {+1.07/25 0.14s}
97. Be7 {-1.03/39 16s} Kd3 {+1.07/26 0.20s} 98. Kb2 {-0.94/39 14s}
Kd4 {+1.00/25 0.18s} 99. Bf6+ {-0.97/39 13s} Kc5 {+1.01/26 0.16s}
100. Bd8 {-0.88/39 14s} Kd5 {+1.01/27 0.14s} 101. Bh4 {-0.68/39 28s}
Ke5 {+1.01/25 0.12s} 102. Be7 {-0.92/38 16s} Kf5 {+0.97/24 0.19s}
103. Ka1 {-0.71/36 11s} Ke4 {+0.89/23 0.19s} 104. Kb2 {-0.82/36 13s}
Ke5 {+0.89/24 0.11s} 105. Bh4 {-0.40/36 20s} Ke6 {+0.88/25 0.20s}
106. Ka1 {-0.39/39 11s} Kd7 {+0.61/18 0.081s} 107. Kb2 {-0.39/41 17s}
Kc6 {+0.60/17 0.047s} 108. Bf6 {-0.38/42 19s} Kb6 {+0.52/21 0.22s}
109. Be7 {-0.38/40 14s} Kb7 {+0.52/23 0.14s} 110. Ka1 {-0.38/41 12s}
Kc8 {+0.45/24 0.36s} 111. Bh4 {-0.38/42 16s} Kc7 {+0.50/22 0.062s}
112. Bg5 {-0.38/43 15s} Kc6 {+0.48/23 0.14s} 113. Bd8 {-0.38/45 19s}
Kd5 {+0.45/22 0.15s} 114. Bf6 {-0.38/46 29s} Kd6 {+0.45/23 0.094s}
115. Bh4 {-0.38/44 19s} Kc5 {+0.45/22 0.081s} 116. Kb2 {-0.38/45 26s}
Kd5 {+0.44/24 0.22s} 117. Bg5 {-0.38/45 14s} Bf5 {+0.43/24 0.15s}
118. Kxa2 {-0.38/42 14s} Kc4 {+0.43/25 0.094s} 119. Kb2 {-0.38/45 18s}
Kd3 {+0.43/23 0.088s} 120. Kb3 {-0.38/43 13s} Be6+ {+0.42/24 0.17s}
121. Kb4 {-0.38/43 12s} Kd4 {+0.41/25 0.12s} 122. Bf6+ {-0.37/41 18s}
Kd5 {+0.41/24 0.078s} 123. Kc3 {-0.37/41 18s} Bf5 {+0.41/26 0.14s}
124. Kd2 {-0.38/45 18s} Ke4 {+0.41/25 0.12s} 125. Be7 {-0.38/40 13s}
Bg4 {+0.40/24 0.14s} 126. Bf6 {-0.38/41 13s} Bf3 {+0.39/23 0.24s}
127. Be7 {-0.38/41 13s} Kf5 {+0.40/25 0.14s} 128. Ke3 {-0.38/45 19s}
Bd5 {+0.39/26 0.14s} 129. Kf2 {-0.37/42 20s} Kf4 {+0.39/24 0.12s}
130. Bd6+ {-0.37/42 13s} Kg4 {+0.39/26 0.16s} 131. Be7 {-0.37/41 17s}
Bf3 {+0.39/27 0.19s} 132. Ke3 {-0.39/46 32s} Bg2 {+0.39/27 0.12s}
133. Kd4 {-0.37/44 39s} Kf4 {+0.38/27 0.17s} 134. Bd6+ {-0.37/42 11s}
Kf5 {+0.38/26 0.14s} 135. Be7 {-0.37/43 19s} Be4 {+0.37/26 0.16s}
136. Ke3 {-0.37/42 14s} Bb7 {+0.37/28 0.16s} 137. Kd4 {-0.36/41 23s}
Bc6 {+0.37/26 0.23s} 138. Bh4 {-0.38/44 15s} Kf4 {+0.37/24 0.094s}
139. Be7 {-0.37/42 21s} Bb7 {+0.37/26 0.12s} 140. Bd8 {-0.38/46 23s}
Be4 {+0.37/26 0.11s} 141. Be7 {-0.38/45 16s} Bf3 {+0.37/27 0.16s}
142. Bd6+ {-0.37/41 18s} Kf5 {+0.38/26 0.16s} 143. Ke3 {-0.37/43 17s}
Bd1 {+0.37/24 0.16s} 144. Be7 {-0.37/42 16s} Kg4 {+0.37/26 0.094s}
145. Kd4 {-0.38/42 14s} Kf4 {+0.37/26 0.14s} 146. Bf6 {-0.37/42 20s}
Bc2 {+0.37/27 0.16s} 147. Bh4 {-0.36/42 19s} Bf5 {+0.37/28 0.19s}
148. Bf6 {-0.35/41 32s} Bg4 {+0.38/26 0.16s} 149. Be7 {-0.35/39 12s}
Be6 {+0.37/26 0.16s} 150. Bd6+ {-0.09/36 14s} Kf5 {+0.38/27 0.27s}
151. Be7 {-0.09/40 42s} Bd7 {+0.37/26 0.093s} 152. Bd8 {-0.09/37 13s}
Be8 {+0.37/26 0.17s} 153. Be7 {-0.09/35 8.6s} Bg6 {+0.37/27 0.11s}
154. Kd5 {-0.09/37 12s} Kg4 {+0.36/26 0.34s} 155. Kd4 {-0.09/38 17s}
Kf4 {+0.36/24 0.20s} 156. Bd6+ {-0.09/40 13s} Kf5 {+0.37/22 0.031s}
157. Be7 {-0.09/42 16s} Kg4 {+0.07/22 0.41s} 158. Bd8 {-0.09/40 13s}
Bf5 {+0.07/20 0.063s} 159. Be7 {-0.09/40 11s} h4 {+0.07/19 0.032s}
160. Bxh4 {-0.07/44 13s, Draw by insufficient mating material} 1/2-1/2[/pgn]

[pgn][Event "?"]
[Site "?"]
[Date "2015.07.31"]
[Round "7"]
[White "Komodo 9.1 2700+15"]
[Black "Komodo 9.1 27+0.15"]
[Result "1-0"]
[BlackTimeControl "27+0.15"]
[FEN "rnbqkbnr/pppp1ppp/8/4p3/4P3/8/PP1P2PP/RNBQKBNR w KQkq -"]
[PlyCount "103"]
[SetUp "1"]
[WhiteTimeControl "2700+15"]

1. Nf3 {-2.08/22 70s} Bc5 {+1.98/14 1.1s} 2. Nc3 {-2.05/23 167s}
Nc6 {+2.02/14 0.61s} 3. Nxe5 {-1.92/26 118s} Nxe5 {+2.08/15 0.76s}
4. d4 {-1.83/24 32s} Bb4 {+1.80/16 1.2s} 5. dxe5 {-1.95/23 49s}
Qh4+ {+1.84/15 0.41s} 6. Kd2 {-1.88/23 34s} Qxe4 {+1.98/14 0.35s}
7. Qe2 {-1.95/23 195s} Qf5 {+2.93/11 0.79s} 8. Qe3 {-1.64/22 62s}
Ne7 {+1.74/14 1.2s} 9. Bd3 {-1.66/22 46s} Qg4 {+1.54/14 1.7s}
10. Kc2 {-1.62/20 37s} Qxg2+ {+1.45/14 1.6s} 11. Bd2 {-1.28/19 21s}
Qc6 {+1.57/15 0.59s} 12. Be4 {-1.34/21 72s} Qe6 {+1.64/15 2.0s}
13. Qd3 {-1.35/21 34s} c6 {+1.63/14 0.86s} 14. Nb5 {-0.59/25 45s}
cxb5 {+0.90/14 0.89s} 15. Bxb4 {-0.31/21 26s} a5 {+0.43/14 0.73s}
16. Bd6 {-0.39/22 32s} Ra6 {+0.66/14 0.81s} 17. Rag1 {-0.05/24 97s}
f5 {+0.21/13 0.80s} 18. Bf3 {-0.09/22 24s} O-O {+0.48/13 0.30s}
19. Qxb5 {-0.04/23 40s} Kh8 {+0.21/13 0.77s} 20. Kb1 {-0.01/24 83s}
f4 {-0.01/13 0.33s} 21. Qd3 {+0.29/21 44s} Rxd6 {-0.17/13 0.79s}
22. exd6 {+0.57/23 39s} Nc6 {-0.16/14 0.43s} 23. Re1 {+0.75/24 50s}
Ne5 {-0.43/14 0.86s} 24. Qd5 {+1.23/24 32s} Qxd5 {-0.47/14 0.23s}
25. Bxd5 {+1.37/25 21s} Nc6 {-0.79/15 0.49s} 26. Rhf1 {+1.49/24 21s}
h6 {-1.03/15 0.93s} 27. Re7 {+1.73/24 18s} g5 {-1.11/14 0.16s}
28. Rfe1 {+2.32/24 20s} a4 {-1.24/16 0.35s} 29. h4 {+2.46/27 23s}
g4 {-2.13/14 0.50s} 30. Re8 {+3.96/25 46s} Rxe8 {-2.42/16 0.57s}
31. Rxe8+ {+4.49/26 48s} Kg7 {-2.50/16 0.21s} 32. Rxc8 {+4.68/27 62s}
Kf6 {-2.82/16 0.80s} 33. Bxc6 {+5.54/25 86s} bxc6 {-2.97/15 0.39s}
34. Kc2 {+5.75/24 69s} Ke6 {-3.59/15 1.1s} 35. Kd3 {+6.78/22 74s}
Kxd6 {-4.20/15 1.2s} 36. Rh8 {+8.65/23 76s} Ke5 {-5.35/13 0.87s}
37. Rxh6 {+8.73/22 54s} c5 {-5.75/13 0.67s} 38. Rh5+ {+10.12/22 78s}
Ke6 {-6.10/13 0.28s} 39. Rxc5 {+12.03/23 76s} d5 {-6.86/13 0.42s}
40. Rc6+ {+16.50/23 60s} Kf7 {-7.76/14 0.36s} 41. Rd6 {+M53/21 16s}
Kg7 {-8.85/16 0.33s} 42. Rxd5 {+M49/21 6.4s} Kg6 {-11.19/16 0.49s}
43. Ke4 {+M47/21 11s} g3 {-15.67/17 0.39s} 44. Rg5+ {+M23/21 9.4s}
Kh6 {-M42/19 0.37s} 45. Kxf4 {+M15/24 7.9s} Kh7 {-M38/17 0.023s}
46. Kxg3 {+M13/25 7.0s} Kh6 {-M38/18 0.025s} 47. a3 {+M11/29 5.8s}
Kh7 {-16.97/16 0.13s} 48. Kf4 {+M9/85 5.8s} Kh6 {-M8/16 0.024s}
49. Kf5 {+M7/99 0.45s} Kh7 {-M6/93 0.024s} 50. Ke6 {+M5/99 0.021s}
Kh6 {-M4/99 0.009s} 51. Kf7 {+M3/99 0.007s} Kh7 {-M2/99 0.009s}
52. Rh5# {+M1/99 0.007s, White mates} 1-0[/pgn]

[Uri Blass]

The plan is to play six games over two days at 45' + 15", exactly as in the ongoing knight-odds match. Only the handicap is changed.

I tried self-play to see time odds necessary for Komodo 9.1 in these conditions, I got them equal at x100 time handicap, i.e. Komodo at 45'+15'' (drawscore=-75) to 27''+0.15'' (drawscore=+50) with 2 draws, 1 win and 1 loss. That would mean 500 ELO points handicap or, saying that Komodo 9.1 on a strong PC is around 3200 FIDE ELO in LTC, this handicap is a good match for a 2700 GM. So, the engine guess would be that Komodo will perform fairly well against 2500-2600 GM, but this is a bit hand-waving argument, engine play un-adapted to handicap games can be misleading. Also, a high draw rate is to be expected. Here are the games at times 100 time control handicap:

[pgn][Event "?"]
[Site "?"]
[Date "2015.07.31"]
[Round "1"]
[White "Komodo 9.1 2700+15"]
[Black "Komodo 9.1 27+0.15"]
[Result "1/2-1/2"]
[BlackTimeControl "27+0.15"]
[FEN "rnbqkbnr/ppp1pppp/8/3p4/8/5N2/PP1PP1PP/RNBQKB1R w KQkq -"]
[PlyCount "141"]
[SetUp "1"]
[WhiteTimeControl "2700+15"]

1. Nc3 {-1.67/22 83s} e5 {+1.52/14 0.73s} 2. Nxe5 {-1.60/23 83s}
d4 {+1.25/15 1.9s} 3. Qa4+ {-1.56/25 78s} c6 {+1.81/14 0.34s}
4. Nd1 {-1.58/24 78s} b5 {+1.74/15 1.5s} 5. Qc2 {-1.26/21 32s}
Qd5 {+1.22/14 1.8s} 6. Nf3 {-0.97/24 33s} Na6 {+0.88/16 2.8s}
7. a3 {-0.88/25 51s} Nc5 {+1.25/16 0.82s} 8. b4 {-0.88/26 71s}
Bf5 {+0.94/16 0.39s} 9. d3 {-0.68/25 63s} Nb3 {+1.00/17 0.46s}
10. Rb1 {-0.60/25 94s} Nxc1 {+0.90/16 0.83s} 11. Qxc1 {-0.74/23 48s}
Bd7 {+0.96/17 0.70s} 12. g3 {-0.60/23 31s} Rc8 {+0.52/14 1.3s}
13. Bg2 {-0.59/23 47s} c5 {+0.55/14 0.76s} 14. Qd2 {-0.57/25 155s}
Nf6 {+0.76/14 0.61s} 15. e3 {-0.62/23 43s} dxe3 {+0.41/15 1.0s}
16. Nxe3 {-0.33/22 33s} Qd6 {+0.67/15 0.51s} 17. bxc5 {-0.39/24 56s}
Qxc5 {+0.88/13 0.31s} 18. d4 {-0.18/22 33s} Qxa3 {+0.15/14 0.68s}
19. Ne5 {-0.15/23 40s} Be6 {+0.18/14 0.38s} 20. O-O {-0.25/23 42s}
Bd6 {+0.15/15 0.70s} 21. Nc6 {-0.17/25 93s} Qa6 {+0.84/12 0.18s}
22. Ra1 {-0.08/26 74s} Qb6 {-0.09/13 0.18s} 23. Nxa7 {-0.08/24 27s}
Rc7 {+0.31/15 0.49s} 24. Qa5 {-0.14/28 155s} Qxa5 {+0.20/15 0.21s}
25. Rxa5 {-0.14/24 20s} Bb4 {+0.17/17 0.75s} 26. Raa1 {-0.17/24 23s}
O-O {+0.39/15 0.46s} 27. Nxb5 {-0.17/24 65s} Rcc8 {+0.32/15 0.51s}
28. Rf2 {-0.17/25 45s} Rb8 {+0.10/12 0.15s} 29. Nc7 {-0.18/26 45s}
Bc3 {+0.21/15 0.45s} 30. Rd1 {-0.22/27 55s} Rfd8 {+0.17/15 0.49s}
31. Nxe6 {-0.27/27 51s} fxe6 {+0.25/13 0.18s} 32. Nc2 {-0.31/28 63s}
Rd6 {+0.19/15 0.63s} 33. Re2 {-0.20/26 62s} Kf7 {+0.26/13 0.36s}
34. Bh3 {-0.23/25 64s} Rbb6 {+0.26/14 0.36s} 35. Rd3 {-0.22/27 58s}
Bb2 {+0.21/15 0.85s} 36. Rf2 {-0.18/25 79s} Kg8 {+0.27/15 0.93s}
37. Bg2 {-0.21/24 86s} e5 {+0.34/14 0.25s} 38. d5 {-0.17/26 30s}
g6 {+0.22/12 0.17s} 39. Re2 {-0.17/28 62s} Rb5 {+0.15/16 0.56s}
40. h3 {-0.17/29 88s} Kg7 {+0.25/16 0.69s} 41. Kh2 {-0.14/28 42s}
e4 {+0.20/18 0.36s} 42. Bxe4 {-0.15/28 29s} Nxe4 {+0.21/18 0.25s}
43. Rxe4 {-0.13/25 19s} Rbxd5 {+0.14/18 0.59s} 44. Rxd5 {-0.13/28 37s}
Rxd5 {+0.16/17 0.17s} 45. Re2 {-0.12/29 61s} Bc3 {+0.16/17 0.30s}
46. Ne3 {-0.10/26 37s} Re5 {+0.13/16 0.49s} 47. Kg2 {-0.09/27 20s}
Bd4 {+0.12/17 0.37s} 48. Kf3 {-0.11/27 23s} Bc5 {+0.14/17 0.42s}
49. Nc4 {-0.04/30 48s} Rxe2 {+0.14/20 0.30s} 50. Kxe2 {-0.05/32 19s}
Kf6 {+0.14/17 0.20s} 51. Kf3 {-0.07/36 36s} Ba7 {+0.08/18 0.29s}
52. h4 {-0.07/40 53s} Ke6 {+0.08/17 0.17s} 53. g4 {-0.07/37 49s}
Bd4 {+0.07/17 0.40s} 54. Ne3 {-0.07/38 45s} Bf6 {+0.06/19 0.28s}
55. g5 {0.00/39 21s} Bb2 {+0.05/20 0.23s} 56. Ke4 {0.00/39 36s}
Ke7 {+0.05/19 0.17s} 57. Nd5+ {0.00/38 41s} Kd6 {+0.05/20 0.14s}
58. Nf4 {0.00/39 26s} Bc1 {+0.03/21 0.22s} 59. Nd3 {0.00/37 38s}
Bd2 {+0.05/23 0.27s} 60. h5 {0.00/36 22s} gxh5 {+0.03/17 0.20s}
61. Nf4 {0.00/44 47s} Bxf4 {+0.03/21 0.25s} 62. Kxf4 {0.00/48 15s}
Ke6 {+0.03/25 0.21s} 63. Ke4 {0.00/52 25s} h4 {+0.03/26 0.13s}
64. Kf4 {0.00/52 18s} h3 {+0.03/27 0.17s} 65. Kf3 {0.00/52 18s}
h2 {0.00/26 0.16s} 66. Kg2 {0.00/54 26s} Kf5 {0.00/31 0.26s}
67. Kxh2 {0.00/59 28s} Kf4 {0.00/32 0.23s} 68. Kh3 {0.00/62 73s}
Kf5 {0.00/33 0.36s} 69. Kh2 {0.00/99 17s} Ke5 {0.00/35 0.31s}
70. Kg3 {0.00/61 19s} Kf5 {0.00/35 0.27s}
71. Kh2 {0.00/99 9.2s, Draw by 3-fold repetition} 1/2-1/2[/pgn]

[pgn][Event "?"]
[Site "?"]
[Date "2015.07.31"]
[Round "3"]
[White "Komodo 9.1 2700+15"]
[Black "Komodo 9.1 27+0.15"]
[Result "0-1"]
[BlackTimeControl "27+0.15"]
[FEN "rnbqkbnr/pppp1ppp/8/4p3/8/2N5/PP1PP1PP/R1BQKBNR w KQkq -"]
[PlyCount "148"]
[SetUp "1"]
[WhiteTimeControl "2700+15"]

1. Nf3 {-1.64/23 69s} Nc6 {+1.62/14 0.55s} 2. Qa4 {-1.74/23 77s}
Nf6 {+1.70/15 1.4s} 3. e3 {-1.92/24 78s} a6 {+1.66/16 1.6s}
4. Be2 {-1.84/24 147s} d6 {+1.53/15 0.51s} 5. O-O {-1.77/26 97s}
Be7 {+1.53/16 0.88s} 6. d4 {-1.82/26 92s} e4 {+1.70/17 0.85s}
7. Nh4 {-1.88/26 117s} O-O {+1.74/16 0.28s} 8. Nf5 {-1.87/25 133s}
Bxf5 {+1.60/16 0.57s} 9. Rxf5 {-1.93/22 28s} Qd7 {+1.72/17 1.1s}
10. Rf1 {-1.75/24 53s} b5 {+1.72/15 0.37s} 11. Qc2 {-1.78/23 50s}
d5 {+1.77/16 1.2s} 12. Bd2 {-1.88/24 56s} Nb4 {+1.80/16 1.4s}
13. Qb3 {-2.08/21 40s} Nd3 {+1.71/16 0.71s} 14. Bxd3 {-1.89/23 82s}
exd3 {+1.66/15 0.47s} 15. Rac1 {-1.96/25 141s} Rac8 {+1.70/15 1.2s}
16. Nd1 {-1.86/23 41s} c5 {+1.79/15 0.44s} 17. Qxd3 {-1.88/23 35s}
Rfe8 {+1.89/14 0.29s} 18. a3 {-1.87/25 110s} Qe6 {+1.88/13 0.46s}
19. Nc3 {-1.92/23 60s} g6 {+1.88/14 0.83s} 20. h3 {-1.88/20 26s}
Bd6 {+1.95/14 0.76s} 21. dxc5 {-2.11/24 80s} Bxc5 {+1.76/14 0.37s}
22. Ne2 {-2.11/24 44s} Qe5 {+2.08/13 0.21s} 23. Nd4 {-2.33/27 93s}
Ne4 {+1.88/14 0.62s} 24. Bb4 {-2.14/23 74s} Bxb4 {+2.00/15 0.47s}
25. axb4 {-2.04/24 30s} Rxc1 {+1.81/16 1.5s} 26. Rxc1 {-2.09/25 14s}
Qg3 {+1.92/16 0.35s} 27. Qe2 {-2.17/26 30s} Ng5 {+2.02/16 0.38s}
28. Qe1 {-2.04/27 28s} Qxe3+ {+1.81/17 0.72s} 29. Qxe3 {-2.12/23 16s}
Rxe3 {+1.95/16 0.46s} 30. Rd1 {-2.20/27 90s} Ne6 {+2.12/18 1.2s}
31. Kf2 {-2.26/29 91s} Re4 {+2.06/17 0.43s} 32. Nc6 {-2.19/27 80s}
Nc7 {+2.14/18 2.0s} 33. Rc1 {-2.18/27 70s} Rc4 {+2.22/18 0.69s}
34. Rxc4 {-1.83/27 47s} bxc4 {+2.35/15 0.18s} 35. Ke3 {-2.18/29 29s}
f6 {+2.15/16 0.17s} 36. Kd4 {-2.19/27 23s} Kf7 {+1.52/16 0.45s}
37. Kc5 {-2.09/27 33s} Ke6 {+1.45/16 0.30s} 38. Kb6 {-2.46/27 102s}
Kd6 {+1.92/16 0.36s} 39. Na5 {-2.49/27 70s} Kd7 {+2.17/16 0.71s}
40. Nc6 {-2.32/26 82s} h5 {+2.18/15 0.13s} 41. Nb8+ {-2.04/23 23s}
Kd6 {+1.81/14 0.17s} 42. Nxa6 {-2.62/23 18s} Ne8 {+1.97/15 0.41s}
43. b5 {-3.12/26 40s} d4 {+2.49/15 0.18s} 44. Ka5 {-3.01/28 22s}
d3 {+2.96/16 0.24s} 45. b6 {-3.95/27 20s} Kc6 {+2.90/18 0.35s}
46. Nb4+ {-3.57/29 28s} Kb7 {+2.71/19 0.30s} 47. Nd5 {-3.62/30 33s}
Nd6 {+2.84/20 0.28s} 48. Kb4 {-4.25/31 55s} f5 {+2.47/16 0.29s}
49. Nc3 {-4.53/30 58s} Kxb6 {+2.74/16 0.22s} 50. b3 {-4.75/32 38s}
cxb3 {+3.52/18 0.20s} 51. Na4+ {-6.68/35 71s} Kc6 {+4.14/21 0.49s}
52. Kxb3 {-9.43/36 49s} d2 {+4.32/19 0.13s} 53. Kc2 {-9.04/34 117s}
Ne4 {+4.76/21 0.42s} 54. Nb2 {-10.18/31 29s} Kd5 {+4.78/21 0.18s}
55. Kd3 {-10.92/32 30s} g5 {+5.40/20 0.66s} 56. Ke3 {-7.23/28 95s}
f4+ {+6.60/17 0.27s} 57. Kd3 {-18.37/25 28s} g4 {+6.85/19 0.24s}
58. hxg4 {-17.85/25 80s} hxg4 {+6.73/20 0.14s} 59. Nd1 {-M54/24 3.9s}
Ke5 {+6.97/19 0.097s} 60. Nb2 {-10.04/29 71s} f3 {+6.92/20 0.25s}
61. g3 {-M64/22 22s} Nxg3 {+10.49/14 0.12s} 62. Kxd2 {-31.84/21 5.5s}
Kd4 {+17.76/16 0.29s} 63. Nd3 {-M76/19 2.8s} Ne4+ {+M51/15 0.17s}
64. Kc2 {-M50/19 1.2s} g3 {+M49/14 0.035s} 65. Nb4 {-M24/18 2.6s}
f2 {+M43/13 0.040s} 66. Nc6+ {-M16/19 1.7s} Ke3 {+M27/14 0.029s}
67. Kb3 {-M16/19 2.6s} g2 {+M19/13 0.072s} 68. Nb4 {-M14/20 2.3s}
f1=Q {+M13/13 0.050s} 69. Nc2+ {-M12/23 2.0s} Kd2 {+M11/14 0.032s}
70. Na3 {-M10/28 2.1s} g1=Q {+M9/15 0.041s} 71. Nc4+ {-M8/99 0.24s}
Kc1 {+M7/21 0.031s} 72. Ka3 {-M6/99 0.058s} Qd3+ {+M5/50 0.029s}
73. Kb4 {-M4/99 0.008s} Qc5+ {+M3/99 0.008s} 74. Ka4 {-M2/99 0.006s}
Qdxc4# {+M1/99 0.007s, Black mates} 0-1[/pgn]

[pgn][Event "?"]
[Site "?"]
[Date "2015.07.31"]
[Round "5"]
[White "Komodo 9.1 2700+15"]
[Black "Komodo 9.1 27+0.15"]
[Result "1/2-1/2"]
[BlackTimeControl "27+0.15"]
[FEN "rnbqkbnr/ppp1pppp/8/3p4/3P4/8/PP2P1PP/RNBQKBNR w KQkq -"]
[PlyCount "319"]
[SetUp "1"]
[WhiteTimeControl "2700+15"]

1. Bf4 {-1.50/23 117s} e5 {+1.39/13 1.1s} 2. dxe5 {-1.42/22 68s}
g5 {+1.36/13 2.1s} 3. Bg3 {-1.42/20 37s} Nc6 {+1.23/13 0.77s}
4. e4 {-1.43/23 85s} dxe4 {+1.26/14 1.5s} 5. Nc3 {-1.38/23 92s}
Qxd1+ {+1.26/15 1.2s} 6. Rxd1 {-1.50/22 26s} Bb4 {+1.38/14 0.32s}
7. Bb5 {-1.47/23 97s} Bxc3+ {+1.48/15 0.82s} 8. bxc3 {-1.42/22 25s}
Ne7 {+1.43/14 0.32s} 9. Ne2 {-1.43/22 68s} Bd7 {+1.34/14 1.9s}
10. O-O {-1.48/23 84s} a6 {+1.23/13 0.41s} 11. Bxc6 {-1.42/23 65s}
Nxc6 {+1.19/14 1.1s} 12. Nd4 {-1.45/26 213s} Nxd4 {+1.28/14 0.32s}
13. cxd4 {-1.17/23 57s} Rd8 {+1.01/14 0.53s} 14. d5 {-1.13/25 48s}
Bb5 {+1.27/15 1.1s} 15. Rf6 {-1.07/23 39s} e3 {+1.17/14 0.61s}
16. d6 {-1.10/25 59s} O-O {+1.19/15 1.0s} 17. h4 {-1.01/26 56s}
e2 {+0.94/14 0.72s} 18. Re1 {-1.19/27 67s} cxd6 {+0.89/16 0.52s}
19. exd6 {-1.16/26 27s} gxh4 {+1.06/15 0.31s} 20. Bxh4 {-1.64/24 34s}
Rfe8 {+1.02/17 0.84s} 21. Rh6 {-1.48/27 34s} Kg7 {+0.99/17 1.1s}
22. Bxd8 {-1.22/27 55s} Kxh6 {+1.02/17 0.20s} 23. Be7 {-1.27/29 23s}
Kg6 {+1.04/18 0.41s} 24. Rxe2 {-1.21/31 80s} Bxe2 {+1.07/16 0.29s}
25. d7 {-1.21/31 19s} Rg8 {+1.08/19 0.27s} 26. d8=Q {-1.25/35 38s}
Rxd8 {+1.11/19 0.16s} 27. Bxd8 {-1.24/34 19s} Kf5 {+1.08/19 0.31s}
28. Kf2 {-1.33/37 64s} Bc4 {+1.03/17 0.27s} 29. a3 {-1.32/36 32s}
Ke4 {+1.14/18 0.88s} 30. g4 {-1.29/36 124s} Kf4 {+1.17/21 1.1s}
31. g5 {-1.33/36 38s} b5 {+1.03/22 0.64s} 32. Kg2 {-1.29/36 61s}
Ke3 {+1.04/21 0.60s} 33. Ba5 {-1.13/37 32s} Be6 {+1.03/22 0.57s}
34. Kg3 {-1.13/36 44s} Kd4 {+1.16/23 0.39s} 35. Kf2 {-1.69/35 80s}
Kc4 {+1.14/23 0.50s} 36. Bb4 {-1.74/35 39s} Kb3 {+1.13/23 0.34s}
37. Ke3 {-2.19/34 77s} a5 {+1.13/25 0.38s} 38. Bd6 {-2.27/36 53s}
Kc4 {+1.25/22 0.40s} 39. Kd2 {-2.33/32 39s} Bf5 {+1.24/22 0.39s}
40. Bc7 {-1.86/37 33s} b4 {+1.51/23 0.51s} 41. Bxa5 {-1.85/36 50s}
bxa3 {+1.85/28 0.41s} 42. Bc3 {-1.86/38 34s} a2 {+1.83/28 0.48s}
43. Bf6 {-1.82/39 28s} Kb3 {+1.83/28 0.51s} 44. Bc3 {-1.75/41 47s}
Bb1 {+1.83/27 0.34s} 45. Bf6 {-1.88/43 47s} Bg6 {+1.83/28 0.58s}
46. Kd1 {-1.80/44 135s} Be4 {+1.83/29 0.49s} 47. Kd2 {-1.86/40 38s}
Bg2 {+1.83/29 0.27s} 48. Kd1 {-1.82/41 28s} Bb7 {+1.83/27 0.36s}
49. Kc1 {-1.81/42 22s} Bd5 {+1.82/28 0.35s} 50. Kd2 {-1.70/40 21s}
Bc6 {+1.80/28 0.34s} 51. Kc1 {-1.75/40 24s} Bd7 {+1.80/29 0.38s}
52. Kd2 {-1.76/38 15s} Be6 {+1.80/28 0.25s} 53. Kd1 {-1.77/39 28s}
Bf5 {+1.80/28 0.24s} 54. Kc1 {-1.77/42 28s} Bc8 {+1.74/27 0.63s}
55. Kd2 {-1.65/39 21s} Bg4 {+1.75/25 0.17s} 56. Kc1 {-1.71/40 28s}
Be2 {+1.75/26 0.29s} 57. Kd2 {-1.76/41 53s} Bf1 {+1.75/25 0.37s}
58. Kc1 {-1.67/41 35s} Bd3 {+1.76/27 0.33s} 59. Kd2 {-1.65/39 17s}
Bb5 {+1.76/27 0.42s} 60. Kd1 {-1.80/39 20s} Ba6 {+1.78/26 0.32s}
61. Kd2 {-1.63/36 14s} Bc8 {+1.77/26 0.26s} 62. Kd1 {-1.57/38 23s}
Be6 {+1.77/27 0.28s} 63. Kd2 {-1.81/41 71s} Bd7 {+1.76/25 0.20s}
64. Kd1 {-1.35/38 48s} Bb5 {+1.77/25 0.28s} 65. Kc1 {-1.43/37 22s}
Bc6 {+1.75/25 0.12s} 66. Kd2 {-1.39/36 15s} Be4 {+1.67/21 0.13s}
67. Kd1 {-1.35/37 19s} Bg6 {+1.67/23 0.22s} 68. Kd2 {-1.34/41 28s}
Kb4 {+1.60/23 0.30s} 69. Kc1 {-1.35/41 26s} Ka3 {+1.59/23 0.21s}
70. Kd1 {-1.34/39 12s} Be4 {+1.52/20 0.15s} 71. Kd2 {-1.35/41 22s}
Ka4 {+1.67/20 0.37s} 72. Kc1 {-1.39/40 15s} Kb4 {+1.12/21 0.11s}
73. Kb2 {-1.38/45 24s} Bb1 {+1.41/23 0.11s} 74. Bc3+ {-1.38/37 14s}
Kc5 {+1.39/23 0.087s} 75. Bd2 {-1.36/37 13s} Kd4 {+1.39/24 0.12s}
76. Bc1 {-1.38/40 15s} Ke5 {+1.39/24 0.15s} 77. Ka1 {-1.35/36 14s}
f5 {+1.31/24 0.27s} 78. gxf6 {-1.30/36 12s} Kxf6 {+1.34/22 0.066s}
79. Bd2 {-1.23/37 12s} Ke5 {+1.21/22 0.15s} 80. Kb2 {-1.35/37 13s}
Kd4 {+1.31/24 0.25s} 81. Bc3+ {-1.19/38 17s} Kd3 {+1.24/23 0.14s}
82. Bf6 {-1.08/38 15s} Kd2 {+1.32/22 0.090s} 83. Bg5+ {-1.37/45 32s}
Ke1 {+1.32/24 0.11s} 84. Bh4+ {-1.38/37 13s} Kd1 {+1.31/26 0.26s}
85. Bd8 {-1.25/40 28s} h5 {+1.34/22 0.056s} 86. Bg5 {-1.09/39 15s}
Ke2 {+1.37/24 0.16s} 87. Be7 {-1.04/39 11s} Ke3 {+1.25/26 0.33s}
88. Bf6 {-1.07/41 18s} Kd3 {+1.20/25 0.12s} 89. Bg5 {-1.16/38 13s}
Kc4 {+1.20/25 0.066s} 90. Bf6 {-1.03/40 18s} Kb4 {+1.20/24 0.12s}
91. Be7+ {-1.09/41 12s} Kb5 {+1.19/26 0.11s} 92. Bg5 {-1.03/40 16s}
Kc5 {+1.17/25 0.18s} 93. Be3+ {-1.06/41 14s} Kd6 {+1.17/26 0.15s}
94. Bg5 {-1.09/38 11s} Ke6 {+1.17/26 0.13s} 95. Ka1 {-0.91/39 16s}
Ke5 {+1.07/25 0.17s} 96. Bd8 {-0.97/38 11s} Ke4 {+1.07/25 0.14s}
97. Be7 {-1.03/39 16s} Kd3 {+1.07/26 0.20s} 98. Kb2 {-0.94/39 14s}
Kd4 {+1.00/25 0.18s} 99. Bf6+ {-0.97/39 13s} Kc5 {+1.01/26 0.16s}
100. Bd8 {-0.88/39 14s} Kd5 {+1.01/27 0.14s} 101. Bh4 {-0.68/39 28s}
Ke5 {+1.01/25 0.12s} 102. Be7 {-0.92/38 16s} Kf5 {+0.97/24 0.19s}
103. Ka1 {-0.71/36 11s} Ke4 {+0.89/23 0.19s} 104. Kb2 {-0.82/36 13s}
Ke5 {+0.89/24 0.11s} 105. Bh4 {-0.40/36 20s} Ke6 {+0.88/25 0.20s}
106. Ka1 {-0.39/39 11s} Kd7 {+0.61/18 0.081s} 107. Kb2 {-0.39/41 17s}
Kc6 {+0.60/17 0.047s} 108. Bf6 {-0.38/42 19s} Kb6 {+0.52/21 0.22s}
109. Be7 {-0.38/40 14s} Kb7 {+0.52/23 0.14s} 110. Ka1 {-0.38/41 12s}
Kc8 {+0.45/24 0.36s} 111. Bh4 {-0.38/42 16s} Kc7 {+0.50/22 0.062s}
112. Bg5 {-0.38/43 15s} Kc6 {+0.48/23 0.14s} 113. Bd8 {-0.38/45 19s}
Kd5 {+0.45/22 0.15s} 114. Bf6 {-0.38/46 29s} Kd6 {+0.45/23 0.094s}
115. Bh4 {-0.38/44 19s} Kc5 {+0.45/22 0.081s} 116. Kb2 {-0.38/45 26s}
Kd5 {+0.44/24 0.22s} 117. Bg5 {-0.38/45 14s} Bf5 {+0.43/24 0.15s}
118. Kxa2 {-0.38/42 14s} Kc4 {+0.43/25 0.094s} 119. Kb2 {-0.38/45 18s}
Kd3 {+0.43/23 0.088s} 120. Kb3 {-0.38/43 13s} Be6+ {+0.42/24 0.17s}
121. Kb4 {-0.38/43 12s} Kd4 {+0.41/25 0.12s} 122. Bf6+ {-0.37/41 18s}
Kd5 {+0.41/24 0.078s} 123. Kc3 {-0.37/41 18s} Bf5 {+0.41/26 0.14s}
124. Kd2 {-0.38/45 18s} Ke4 {+0.41/25 0.12s} 125. Be7 {-0.38/40 13s}
Bg4 {+0.40/24 0.14s} 126. Bf6 {-0.38/41 13s} Bf3 {+0.39/23 0.24s}
127. Be7 {-0.38/41 13s} Kf5 {+0.40/25 0.14s} 128. Ke3 {-0.38/45 19s}
Bd5 {+0.39/26 0.14s} 129. Kf2 {-0.37/42 20s} Kf4 {+0.39/24 0.12s}
130. Bd6+ {-0.37/42 13s} Kg4 {+0.39/26 0.16s} 131. Be7 {-0.37/41 17s}
Bf3 {+0.39/27 0.19s} 132. Ke3 {-0.39/46 32s} Bg2 {+0.39/27 0.12s}
133. Kd4 {-0.37/44 39s} Kf4 {+0.38/27 0.17s} 134. Bd6+ {-0.37/42 11s}
Kf5 {+0.38/26 0.14s} 135. Be7 {-0.37/43 19s} Be4 {+0.37/26 0.16s}
136. Ke3 {-0.37/42 14s} Bb7 {+0.37/28 0.16s} 137. Kd4 {-0.36/41 23s}
Bc6 {+0.37/26 0.23s} 138. Bh4 {-0.38/44 15s} Kf4 {+0.37/24 0.094s}
139. Be7 {-0.37/42 21s} Bb7 {+0.37/26 0.12s} 140. Bd8 {-0.38/46 23s}
Be4 {+0.37/26 0.11s} 141. Be7 {-0.38/45 16s} Bf3 {+0.37/27 0.16s}
142. Bd6+ {-0.37/41 18s} Kf5 {+0.38/26 0.16s} 143. Ke3 {-0.37/43 17s}
Bd1 {+0.37/24 0.16s} 144. Be7 {-0.37/42 16s} Kg4 {+0.37/26 0.094s}
145. Kd4 {-0.38/42 14s} Kf4 {+0.37/26 0.14s} 146. Bf6 {-0.37/42 20s}
Bc2 {+0.37/27 0.16s} 147. Bh4 {-0.36/42 19s} Bf5 {+0.37/28 0.19s}
148. Bf6 {-0.35/41 32s} Bg4 {+0.38/26 0.16s} 149. Be7 {-0.35/39 12s}
Be6 {+0.37/26 0.16s} 150. Bd6+ {-0.09/36 14s} Kf5 {+0.38/27 0.27s}
151. Be7 {-0.09/40 42s} Bd7 {+0.37/26 0.093s} 152. Bd8 {-0.09/37 13s}
Be8 {+0.37/26 0.17s} 153. Be7 {-0.09/35 8.6s} Bg6 {+0.37/27 0.11s}
154. Kd5 {-0.09/37 12s} Kg4 {+0.36/26 0.34s} 155. Kd4 {-0.09/38 17s}
Kf4 {+0.36/24 0.20s} 156. Bd6+ {-0.09/40 13s} Kf5 {+0.37/22 0.031s}
157. Be7 {-0.09/42 16s} Kg4 {+0.07/22 0.41s} 158. Bd8 {-0.09/40 13s}
Bf5 {+0.07/20 0.063s} 159. Be7 {-0.09/40 11s} h4 {+0.07/19 0.032s}
160. Bxh4 {-0.07/44 13s, Draw by insufficient mating material} 1/2-1/2[/pgn]

[pgn][Event "?"]
[Site "?"]
[Date "2015.07.31"]
[Round "7"]
[White "Komodo 9.1 2700+15"]
[Black "Komodo 9.1 27+0.15"]
[Result "1-0"]
[BlackTimeControl "27+0.15"]
[FEN "rnbqkbnr/pppp1ppp/8/4p3/4P3/8/PP1P2PP/RNBQKBNR w KQkq -"]
[PlyCount "103"]
[SetUp "1"]
[WhiteTimeControl "2700+15"]

1. Nf3 {-2.08/22 70s} Bc5 {+1.98/14 1.1s} 2. Nc3 {-2.05/23 167s}
Nc6 {+2.02/14 0.61s} 3. Nxe5 {-1.92/26 118s} Nxe5 {+2.08/15 0.76s}
4. d4 {-1.83/24 32s} Bb4 {+1.80/16 1.2s} 5. dxe5 {-1.95/23 49s}
Qh4+ {+1.84/15 0.41s} 6. Kd2 {-1.88/23 34s} Qxe4 {+1.98/14 0.35s}
7. Qe2 {-1.95/23 195s} Qf5 {+2.93/11 0.79s} 8. Qe3 {-1.64/22 62s}
Ne7 {+1.74/14 1.2s} 9. Bd3 {-1.66/22 46s} Qg4 {+1.54/14 1.7s}
10. Kc2 {-1.62/20 37s} Qxg2+ {+1.45/14 1.6s} 11. Bd2 {-1.28/19 21s}
Qc6 {+1.57/15 0.59s} 12. Be4 {-1.34/21 72s} Qe6 {+1.64/15 2.0s}
13. Qd3 {-1.35/21 34s} c6 {+1.63/14 0.86s} 14. Nb5 {-0.59/25 45s}
cxb5 {+0.90/14 0.89s} 15. Bxb4 {-0.31/21 26s} a5 {+0.43/14 0.73s}
16. Bd6 {-0.39/22 32s} Ra6 {+0.66/14 0.81s} 17. Rag1 {-0.05/24 97s}
f5 {+0.21/13 0.80s} 18. Bf3 {-0.09/22 24s} O-O {+0.48/13 0.30s}
19. Qxb5 {-0.04/23 40s} Kh8 {+0.21/13 0.77s} 20. Kb1 {-0.01/24 83s}
f4 {-0.01/13 0.33s} 21. Qd3 {+0.29/21 44s} Rxd6 {-0.17/13 0.79s}
22. exd6 {+0.57/23 39s} Nc6 {-0.16/14 0.43s} 23. Re1 {+0.75/24 50s}
Ne5 {-0.43/14 0.86s} 24. Qd5 {+1.23/24 32s} Qxd5 {-0.47/14 0.23s}
25. Bxd5 {+1.37/25 21s} Nc6 {-0.79/15 0.49s} 26. Rhf1 {+1.49/24 21s}
h6 {-1.03/15 0.93s} 27. Re7 {+1.73/24 18s} g5 {-1.11/14 0.16s}
28. Rfe1 {+2.32/24 20s} a4 {-1.24/16 0.35s} 29. h4 {+2.46/27 23s}
g4 {-2.13/14 0.50s} 30. Re8 {+3.96/25 46s} Rxe8 {-2.42/16 0.57s}
31. Rxe8+ {+4.49/26 48s} Kg7 {-2.50/16 0.21s} 32. Rxc8 {+4.68/27 62s}
Kf6 {-2.82/16 0.80s} 33. Bxc6 {+5.54/25 86s} bxc6 {-2.97/15 0.39s}
34. Kc2 {+5.75/24 69s} Ke6 {-3.59/15 1.1s} 35. Kd3 {+6.78/22 74s}
Kxd6 {-4.20/15 1.2s} 36. Rh8 {+8.65/23 76s} Ke5 {-5.35/13 0.87s}
37. Rxh6 {+8.73/22 54s} c5 {-5.75/13 0.67s} 38. Rh5+ {+10.12/22 78s}
Ke6 {-6.10/13 0.28s} 39. Rxc5 {+12.03/23 76s} d5 {-6.86/13 0.42s}
40. Rc6+ {+16.50/23 60s} Kf7 {-7.76/14 0.36s} 41. Rd6 {+M53/21 16s}
Kg7 {-8.85/16 0.33s} 42. Rxd5 {+M49/21 6.4s} Kg6 {-11.19/16 0.49s}
43. Ke4 {+M47/21 11s} g3 {-15.67/17 0.39s} 44. Rg5+ {+M23/21 9.4s}
Kh6 {-M42/19 0.37s} 45. Kxf4 {+M15/24 7.9s} Kh7 {-M38/17 0.023s}
46. Kxg3 {+M13/25 7.0s} Kh6 {-M38/18 0.025s} 47. a3 {+M11/29 5.8s}
Kh7 {-16.97/16 0.13s} 48. Kf4 {+M9/85 5.8s} Kh6 {-M8/16 0.024s}
49. Kf5 {+M7/99 0.45s} Kh7 {-M6/93 0.024s} 50. Ke6 {+M5/99 0.021s}
Kh6 {-M4/99 0.009s} 51. Kf7 {+M3/99 0.007s} Kh7 {-M2/99 0.009s}
52. Rh5# {+M1/99 0.007s, White mates} 1-0[/pgn]

Looking at the last game it seems that komodo blundered by not playing
7...Qxe2+

I think that this may be a good test position for chess programs.
Maybe some different move is better than Qxe2+ but at least at small depths Qxe2+ should be the only move that the program suggest.

[d]r1b1k1nr/pppp1ppp/8/4P3/1b2q3/2N5/PP1KQ1PP/R1B2B1R b kq - 0 7 bm Qxe2+

[Laskos]

Finally, I got better results giving knight odds to Nebula 2 both without contempt and with contempt 100 than with 150. So probably we'll reduce the contempt setting for this weekend. I think 150 makes Komodo play too passively for fear of exchanges.

You also said that contempt 50 for two pawns (c2 and f2) handicap is probably better than 75. I now made some more precision measure at ultra-fast, 1,000 games each match (self-play), but I am not sure how ultra-fast influences the contempt and how is that relevant against humans.

Handicapped by two pawns Komodo at longer time control at different Contempt factors against a shorter time control Komodo at Contempt=0 and Contempt=50:

Contempt=0
Contempt=0
+25 ELO points

Contempt=25
Contempt=0
+40 ELO points
Difference: 15 ELO points

Contempt=50
Contempt=0
+70 ELO points
Difference: 45 ELO points

Contempt=75
Contempt=0
+63 ELO points
Difference: 38 ELO points

Contempt=100
Contempt=0
+50 ELO points
Difference: 25 ELO points

The best against Contempt=0 of the non-handicapped side seems to be Contempt=50 of Komodo with 45 ELO points improvement.

But a human would possibly have a negative contempt, I set the non-handicapped side now to Contempt=-50:

Contempt=0
Contempt=-50
-25 ELO points

Contempt=25
Contempt=-50
-8 ELO points
Difference: 17 ELO points

Contempt=50
Contempt=-50
+24 ELO points
Difference: 49 ELO points

Contempt=75
Contempt=-50
+45 ELO points
Difference: 70 ELO points

Contempt=100
Contempt=-50
+26 ELO points
Difference: 51 ELO points

The best now against Contempt=-50 of the non-handicapped side seems to be Contempt=75 of Komodo with 70 ELO points improvement.

You said that a too large contempt makes Komodo play passively for the fear of exchanges. I think this is a very general issue with strong engines tuned against strong engines, they play best moves in some general conditions, not the most anti-human. I guess even a weaker engine specifically playing complex tactically moves could perform better against a human. And to beat Fernando in less than 30 moves, as he wishes for from engines .