Amazing results of LeelaKnightOdds-Dev

[Marcus91]

another game of the master when he could win but it seems that his rook had some problem to make a long move and played the losing Rf2 instead of Rxf1

https://lichess.org/fT7FvKG3

Good evening Mr. Larry Kaufman and Mr. Uriblass.
My hardware will need to be repaired in the shop. Defects have been detected and I will go into repair. I want to thank you both very especially for the dialogue around the posts started last September, for your patience and interest in the topic. If my Programmer from Heaven wants me back in battle, count on me tomorrow. Excuse me when I have been rude and apparently a bit vain, I just wanted to be faithful to the truth. I also want to thank the team behind the chess and computer hardware and software. No one at all is beyond the reach of my gratitude. Thanks again!! I send a big hug to the street players, to the players to the coffee players and to the Masters. Since my beginnings as a chess player, I have always believed that we are all good players. There will simply always be those who see a little more, a little less, self-esteem should never be lost, respect for other players is the very essence of combat, that is, the dignity that war and peace must entail for the opponent.
Sincerely,
Pablo Ignacio Restrepo

It seems that the LeelaQueenOdds bot has attained your goal of performing at LiChess 2500+ level in 3'0" chess. While there haven't been many games at that precise time control with Leela playing White as is customary in odds play (mostly by Respectful_Dave who broke even in ten games and is rated 2440 blitz but over 2800 bullet), even at 3'2" results are close to 2500 level, and at 1'1" close to 2700 (blitz rating) level, so I think it's clear that it would keep a rating over 2500 at 3'0". But the dropoff with longer time limits, in the Rapid range, is more severe than we expected. Players below master level can beat the bot in Rapid, but not typically in blitz. However the queen for knight bot seems to require something like IM level to be competitive even in Rapid.

I can add that maybe you are also too optimistic about 2500 level for the knight bot at 90+30.
If you underestimate the difference with odds between rapid and blitz then maybe you also underestimate the difference with odds between rapid and 90+30 time control.

Yes, I believe so too, however estimating how much is not easy, we will have a clearer vision after a match at 90+30 2500 level and 15+10 2700 level, should both be close matches though

[Uri Blass]

another game of the master when he could win but it seems that his rook had some problem to make a long move and played the losing Rf2 instead of Rxf1

https://lichess.org/fT7FvKG3

Good evening Mr. Larry Kaufman and Mr. Uriblass.
My hardware will need to be repaired in the shop. Defects have been detected and I will go into repair. I want to thank you both very especially for the dialogue around the posts started last September, for your patience and interest in the topic. If my Programmer from Heaven wants me back in battle, count on me tomorrow. Excuse me when I have been rude and apparently a bit vain, I just wanted to be faithful to the truth. I also want to thank the team behind the chess and computer hardware and software. No one at all is beyond the reach of my gratitude. Thanks again!! I send a big hug to the street players, to the players to the coffee players and to the Masters. Since my beginnings as a chess player, I have always believed that we are all good players. There will simply always be those who see a little more, a little less, self-esteem should never be lost, respect for other players is the very essence of combat, that is, the dignity that war and peace must entail for the opponent.
Sincerely,
Pablo Ignacio Restrepo

It seems that the LeelaQueenOdds bot has attained your goal of performing at LiChess 2500+ level in 3'0" chess. While there haven't been many games at that precise time control with Leela playing White as is customary in odds play (mostly by Respectful_Dave who broke even in ten games and is rated 2440 blitz but over 2800 bullet), even at 3'2" results are close to 2500 level, and at 1'1" close to 2700 (blitz rating) level, so I think it's clear that it would keep a rating over 2500 at 3'0". But the dropoff with longer time limits, in the Rapid range, is more severe than we expected. Players below master level can beat the bot in Rapid, but not typically in blitz. However the queen for knight bot seems to require something like IM level to be competitive even in Rapid.

fide master Ori Taichman could beat it 6-2 at 2+3 after winning 7.5-3.5 earlier.

practically he used almost only the 3 seconds increment and never got into time touble and it is clear he could do better.
I guess he could get more than 50% against the bot at 3+0

[Uri Blass]

LeelaQueenOdds as white 5+3 (Reference) 2291.4
1 Master_Unknown 2266.6 38
LeelaQueenOdds as black 5+3 (Reference) 2241.4
2 Omnivor 2223.6 22
3 nn1133 2223.3 6167
LeelaQueenOdds as white 10+5 (Reference) 2210.9
4 Ajktulm 2197.7 181
5 uriblass 2180.9 134
LeelaQueenOdds as black 10+5 (Reference) 2160.9
LeelaQueenOdds as white 15+10 (Reference) 2155.2
6 Catecan 2123.0 272
7 Brunetticus 2110.3 404
LeelaQueenOdds as black 15+10 (Reference) 2105.2
8 Orichess88 2093.8 19
9 Respectful_Dave 2050.2 51
10 Hissha 2047.1 54
11 KentGoulding 2011.9 256
12 TheLlamaLord 2010.3 22

[lkaufman]

another game of the master when he could win but it seems that his rook had some problem to make a long move and played the losing Rf2 instead of Rxf1

https://lichess.org/fT7FvKG3

Good evening Mr. Larry Kaufman and Mr. Uriblass.
My hardware will need to be repaired in the shop. Defects have been detected and I will go into repair. I want to thank you both very especially for the dialogue around the posts started last September, for your patience and interest in the topic. If my Programmer from Heaven wants me back in battle, count on me tomorrow. Excuse me when I have been rude and apparently a bit vain, I just wanted to be faithful to the truth. I also want to thank the team behind the chess and computer hardware and software. No one at all is beyond the reach of my gratitude. Thanks again!! I send a big hug to the street players, to the players to the coffee players and to the Masters. Since my beginnings as a chess player, I have always believed that we are all good players. There will simply always be those who see a little more, a little less, self-esteem should never be lost, respect for other players is the very essence of combat, that is, the dignity that war and peace must entail for the opponent.
Sincerely,
Pablo Ignacio Restrepo

It seems that the LeelaQueenOdds bot has attained your goal of performing at LiChess 2500+ level in 3'0" chess. While there haven't been many games at that precise time control with Leela playing White as is customary in odds play (mostly by Respectful_Dave who broke even in ten games and is rated 2440 blitz but over 2800 bullet), even at 3'2" results are close to 2500 level, and at 1'1" close to 2700 (blitz rating) level, so I think it's clear that it would keep a rating over 2500 at 3'0". But the dropoff with longer time limits, in the Rapid range, is more severe than we expected. Players below master level can beat the bot in Rapid, but not typically in blitz. However the queen for knight bot seems to require something like IM level to be competitive even in Rapid.

fide master Ori Taichman could beat it 6-2 at 2+3 after winning 7.5-3.5 earlier.

practically he used almost only the 3 seconds increment and never got into time touble and it is clear he could do better.
I guess he could get more than 50% against the bot at 3+0

Yes, but it seems he switched from LiChess to Chess.com for his normal chess a couple years ago, so his LiChess blitz rating is way out of date. He is 2685 on chess.com blitz, which converts to 2625 on LiChess blitz per the conversion website. So even if he could score say 60% at 3 + 0, that would put the bot at 2555 at that TC. To answer another question, no the bots do not think on the humans' time; they can be set to do so, but one computer is running many games at once on one GPU so this isn't practical, and anyway there is no real benefit since they all have low, strict node limits. It might reduce the time to move on average from say one second to 3/4 second, not meaningful. Yes, I think you are right about underestimating the difference between Rapid and Classical time limit play, but the bots are doing so well in Rapid vs. very strong players that I think we have a decent "cushion" for error here. Seeing Leela win some games from strong IMs even at queen odds in blitz, even 5'3", gives me confidence regarding standard games at knight odds against just slightly stronger players. But we should stay close to 2500 for our first classical opponent, I think we would lose if we go much above that.

[Uri Blass]

LeelaQueenOdds as white 5+3 (Reference) 2291.4
1 Master_Unknown 2266.6 38
LeelaQueenOdds as black 5+3 (Reference) 2241.4
2 Omnivor 2223.6 22
3 nn1133 2223.3 6167
LeelaQueenOdds as white 10+5 (Reference) 2210.9
4 Ajktulm 2197.7 181
5 uriblass 2180.9 134
LeelaQueenOdds as black 10+5 (Reference) 2160.9
LeelaQueenOdds as white 15+10 (Reference) 2155.2
6 Catecan 2123.0 272
7 Brunetticus 2110.3 404
LeelaQueenOdds as black 15+10 (Reference) 2105.2
8 Orichess88 2093.8 19
9 Respectful_Dave 2050.2 51
10 Hissha 2047.1 54
11 KentGoulding 2011.9 256
12 TheLlamaLord 2010.3 22

It seems that the rating of LeelaQueenOdds changed and the result is also changes in the rating of the players like Orichess88 without changing the number of games(I do not understand why my rating improved to 2217.1 with only one additional game relative to previous list.

My performance is certainly not 100 elo better than the last rating of LeelaQueenOdds as white

1 Master_Unknown 2448.3 38
2 Omnivor 2349.5 25
LeelaQueenOdds as black 3+2 (Reference) 2349.4
LeelaQueenOdds as white 5+3 (Reference) 2316.4(earlier 2291.4)
3 Orichess88 2263.8 19
4 Respectful_Dave 2248.6 62
5 Ajktulm 2228.1 195
6 uriblass 2217.1 135
LeelaQueenOdds as black 5+3 (Reference) 2216.4(earlier 2241.4)
7 Brunetticus 2202.5 421
8 TheLlamaLord 2194.6 22
9 nn1133 2175.4 6167
10 Jagen 2115.2 55
11 Catecan 2113.8 272
LeelaQueenOdds as white 10+5 (Reference) 2110.2(earlier 2210.9)
12 Forgive1219 2098.9 64
13 Punypawn315 2079.5 19
14 jzm 2047.6 15
15 Hissha 2047.1 54
16 KentGoulding 2026.0 258
LeelaQueenOdds as black 10+5 (Reference) 2010.2(earlier 2160.9)
17 susiedent 2003.1 14
18 desperado64 1990.1 41
19 SilverMitt 1975.5 13
20 SerbianPigeon 1971.9 17
21 junglegame 1958.3 9
22 OjaiJoao 1949.7 4
LeelaQueenOdds as white 15+10 (Reference) 1948.2(earlier 2155.2)
LeelaQueenOdds as black 15+10 (Reference) 1848.2(earlier 2105.2)

[Father]

LeelaQueenOdds as white 5+3 (Reference) 2291.4
1 Master_Unknown 2266.6 38
LeelaQueenOdds as black 5+3 (Reference) 2241.4
2 Omnivor 2223.6 22
3 nn1133 2223.3 6167
LeelaQueenOdds as white 10+5 (Reference) 2210.9
4 Ajktulm 2197.7 181
5 uriblass 2180.9 134
LeelaQueenOdds as black 10+5 (Reference) 2160.9
LeelaQueenOdds as white 15+10 (Reference) 2155.2
6 Catecan 2123.0 272
7 Brunetticus 2110.3 404
LeelaQueenOdds as black 15+10 (Reference) 2105.2
8 Orichess88 2093.8 19
9 Respectful_Dave 2050.2 51
10 Hissha 2047.1 54
11 KentGoulding 2011.9 256
12 TheLlamaLord 2010.3 22

It seems that the rating of LeelaQueenOdds changed and the result is also changes in the rating of the players like Orichess88 without changing the number of games(I do not understand why my rating improved to 2217.1 with only one additional game relative to previous list.

My performance is certainly not 100 elo better than the last rating of LeelaQueenOdds as white

1 Master_Unknown 2448.3 38
2 Omnivor 2349.5 25
LeelaQueenOdds as black 3+2 (Reference) 2349.4
LeelaQueenOdds as white 5+3 (Reference) 2316.4(earlier 2291.4)
3 Orichess88 2263.8 19
4 Respectful_Dave 2248.6 62
5 Ajktulm 2228.1 195
6 uriblass 2217.1 135
LeelaQueenOdds as black 5+3 (Reference) 2216.4(earlier 2241.4)
7 Brunetticus 2202.5 421
8 TheLlamaLord 2194.6 22
9 nn1133 2175.4 6167
10 Jagen 2115.2 55
11 Catecan 2113.8 272
LeelaQueenOdds as white 10+5 (Reference) 2110.2(earlier 2210.9)
12 Forgive1219 2098.9 64
13 Punypawn315 2079.5 19
14 jzm 2047.6 15
15 Hissha 2047.1 54
16 KentGoulding 2026.0 258
LeelaQueenOdds as black 10+5 (Reference) 2010.2(earlier 2160.9)
17 susiedent 2003.1 14
18 desperado64 1990.1 41
19 SilverMitt 1975.5 13
20 SerbianPigeon 1971.9 17
21 junglegame 1958.3 9
22 OjaiJoao 1949.7 4
LeelaQueenOdds as white 15+10 (Reference) 1948.2(earlier 2155.2)
LeelaQueenOdds as black 15+10 (Reference) 1848.2(earlier 2105.2)

The beast continues to rise: LiamCleryChess player 2900 elo bullet in Lichess result as of November 19, 2024 in Liichess against the beauty queen, the machine LeelaQueenOdds, 1 victory for the human and 17 losses for the human: 17 versus 1 in favor of the machine ... I hope to see here in "the ring and street arena" the street fighters and the big leagues!! The bets increase!! Who gives more? !! Who gives more? !! ...

[Father]

LeelaQueenOdds as white 5+3 (Reference) 2291.4
1 Master_Unknown 2266.6 38
LeelaQueenOdds as black 5+3 (Reference) 2241.4
2 Omnivor 2223.6 22
3 nn1133 2223.3 6167
LeelaQueenOdds as white 10+5 (Reference) 2210.9
4 Ajktulm 2197.7 181
5 uriblass 2180.9 134
LeelaQueenOdds as black 10+5 (Reference) 2160.9
LeelaQueenOdds as white 15+10 (Reference) 2155.2
6 Catecan 2123.0 272
7 Brunetticus 2110.3 404
LeelaQueenOdds as black 15+10 (Reference) 2105.2
8 Orichess88 2093.8 19
9 Respectful_Dave 2050.2 51
10 Hissha 2047.1 54
11 KentGoulding 2011.9 256
12 TheLlamaLord 2010.3 22

It seems that the rating of LeelaQueenOdds changed and the result is also changes in the rating of the players like Orichess88 without changing the number of games(I do not understand why my rating improved to 2217.1 with only one additional game relative to previous list.

My performance is certainly not 100 elo better than the last rating of LeelaQueenOdds as white

1 Master_Unknown 2448.3 38
2 Omnivor 2349.5 25
LeelaQueenOdds as black 3+2 (Reference) 2349.4
LeelaQueenOdds as white 5+3 (Reference) 2316.4(earlier 2291.4)
3 Orichess88 2263.8 19
4 Respectful_Dave 2248.6 62
5 Ajktulm 2228.1 195
6 uriblass 2217.1 135
LeelaQueenOdds as black 5+3 (Reference) 2216.4(earlier 2241.4)
7 Brunetticus 2202.5 421
8 TheLlamaLord 2194.6 22
9 nn1133 2175.4 6167
10 Jagen 2115.2 55
11 Catecan 2113.8 272
LeelaQueenOdds as white 10+5 (Reference) 2110.2(earlier 2210.9)
12 Forgive1219 2098.9 64
13 Punypawn315 2079.5 19
14 jzm 2047.6 15
15 Hissha 2047.1 54
16 KentGoulding 2026.0 258
LeelaQueenOdds as black 10+5 (Reference) 2010.2(earlier 2160.9)
17 susiedent 2003.1 14
18 desperado64 1990.1 41
19 SilverMitt 1975.5 13
20 SerbianPigeon 1971.9 17
21 junglegame 1958.3 9
22 OjaiJoao 1949.7 4
LeelaQueenOdds as white 15+10 (Reference) 1948.2(earlier 2155.2)
LeelaQueenOdds as black 15+10 (Reference) 1848.2(earlier 2105.2)

The beast continues to rise: LiamCleryChess player 2900 elo bullet in Lichess result as of November 19, 2024 in Liichess against the beauty queen, the machine LeelaQueenOdds, 1 victory for the human and 17 losses for the human: 17 versus 1 in favor of the machine ... I hope to see here in "the ring and street arena" the street fighters and the big leagues!! The bets increase!! Who gives more? !! Who gives more? !! ...

** 17.5 versos 0.5. No victories.

[lkaufman]

LeelaQueenOdds as white 5+3 (Reference) 2291.4
1 Master_Unknown 2266.6 38
LeelaQueenOdds as black 5+3 (Reference) 2241.4
2 Omnivor 2223.6 22
3 nn1133 2223.3 6167
LeelaQueenOdds as white 10+5 (Reference) 2210.9
4 Ajktulm 2197.7 181
5 uriblass 2180.9 134
LeelaQueenOdds as black 10+5 (Reference) 2160.9
LeelaQueenOdds as white 15+10 (Reference) 2155.2
6 Catecan 2123.0 272
7 Brunetticus 2110.3 404
LeelaQueenOdds as black 15+10 (Reference) 2105.2
8 Orichess88 2093.8 19
9 Respectful_Dave 2050.2 51
10 Hissha 2047.1 54
11 KentGoulding 2011.9 256
12 TheLlamaLord 2010.3 22

It seems that the rating of LeelaQueenOdds changed and the result is also changes in the rating of the players like Orichess88 without changing the number of games(I do not understand why my rating improved to 2217.1 with only one additional game relative to previous list.

My performance is certainly not 100 elo better than the last rating of LeelaQueenOdds as white

1 Master_Unknown 2448.3 38
2 Omnivor 2349.5 25
LeelaQueenOdds as black 3+2 (Reference) 2349.4
LeelaQueenOdds as white 5+3 (Reference) 2316.4(earlier 2291.4)
3 Orichess88 2263.8 19
4 Respectful_Dave 2248.6 62
5 Ajktulm 2228.1 195
6 uriblass 2217.1 135
LeelaQueenOdds as black 5+3 (Reference) 2216.4(earlier 2241.4)
7 Brunetticus 2202.5 421
8 TheLlamaLord 2194.6 22
9 nn1133 2175.4 6167
10 Jagen 2115.2 55
11 Catecan 2113.8 272
LeelaQueenOdds as white 10+5 (Reference) 2110.2(earlier 2210.9)
12 Forgive1219 2098.9 64
13 Punypawn315 2079.5 19
14 jzm 2047.6 15
15 Hissha 2047.1 54
16 KentGoulding 2026.0 258
LeelaQueenOdds as black 10+5 (Reference) 2010.2(earlier 2160.9)
17 susiedent 2003.1 14
18 desperado64 1990.1 41
19 SilverMitt 1975.5 13
20 SerbianPigeon 1971.9 17
21 junglegame 1958.3 9
22 OjaiJoao 1949.7 4
LeelaQueenOdds as white 15+10 (Reference) 1948.2(earlier 2155.2)
LeelaQueenOdds as black 15+10 (Reference) 1848.2(earlier 2105.2)

The beast continues to rise: LiamCleryChess player 2900 elo bullet in Lichess result as of November 19, 2024 in Liichess against the beauty queen, the machine LeelaQueenOdds, 1 victory for the human and 17 losses for the human: 17 versus 1 in favor of the machine ... I hope to see here in "the ring and street arena" the street fighters and the big leagues!! The bets increase!! Who gives more? !! Who gives more? !! ...

** 17.5 versos 0.5. No victories.

This is of course 1'0" bullet chess. The total score for LeelaQueenOdds since the Nov.12 upgrade to the queen odds net at this time control, against all players rated over 2700 bullet, is as follows: Leela playing White (true queen odds) +14 =3 -0. Leela playing Black (queen and move odds): +26=1-2. If we call the average rating 2800, which looks about right, the performance is a bit over 3200 with both colors. Not quite the very top of LiChess bullet, but it suggests that only roughly the top six LiChess bullet players could win a 1+0 match from LeelaQueenOdds, even with alternating colors. But I even doubt that they could do so. Leela only needs to hold out until flag fall. But 1'1" is another story entirely, here the result should be decided on the board. Even then, the results are above 2800 bullet rating average combining both colors.

[lkaufman]

LeelaQueenOdds as white 5+3 (Reference) 2291.4
1 Master_Unknown 2266.6 38
LeelaQueenOdds as black 5+3 (Reference) 2241.4
2 Omnivor 2223.6 22
3 nn1133 2223.3 6167
LeelaQueenOdds as white 10+5 (Reference) 2210.9
4 Ajktulm 2197.7 181
5 uriblass 2180.9 134
LeelaQueenOdds as black 10+5 (Reference) 2160.9
LeelaQueenOdds as white 15+10 (Reference) 2155.2
6 Catecan 2123.0 272
7 Brunetticus 2110.3 404
LeelaQueenOdds as black 15+10 (Reference) 2105.2
8 Orichess88 2093.8 19
9 Respectful_Dave 2050.2 51
10 Hissha 2047.1 54
11 KentGoulding 2011.9 256
12 TheLlamaLord 2010.3 22

It seems that the rating of LeelaQueenOdds changed and the result is also changes in the rating of the players like Orichess88 without changing the number of games(I do not understand why my rating improved to 2217.1 with only one additional game relative to previous list.

My performance is certainly not 100 elo better than the last rating of LeelaQueenOdds as white

1 Master_Unknown 2448.3 38
2 Omnivor 2349.5 25
LeelaQueenOdds as black 3+2 (Reference) 2349.4
LeelaQueenOdds as white 5+3 (Reference) 2316.4(earlier 2291.4)
3 Orichess88 2263.8 19
4 Respectful_Dave 2248.6 62
5 Ajktulm 2228.1 195
6 uriblass 2217.1 135
LeelaQueenOdds as black 5+3 (Reference) 2216.4(earlier 2241.4)
7 Brunetticus 2202.5 421
8 TheLlamaLord 2194.6 22
9 nn1133 2175.4 6167
10 Jagen 2115.2 55
11 Catecan 2113.8 272
LeelaQueenOdds as white 10+5 (Reference) 2110.2(earlier 2210.9)
12 Forgive1219 2098.9 64
13 Punypawn315 2079.5 19
14 jzm 2047.6 15
15 Hissha 2047.1 54
16 KentGoulding 2026.0 258
LeelaQueenOdds as black 10+5 (Reference) 2010.2(earlier 2160.9)
17 susiedent 2003.1 14
18 desperado64 1990.1 41
19 SilverMitt 1975.5 13
20 SerbianPigeon 1971.9 17
21 junglegame 1958.3 9
22 OjaiJoao 1949.7 4
LeelaQueenOdds as white 15+10 (Reference) 1948.2(earlier 2155.2)
LeelaQueenOdds as black 15+10 (Reference) 1848.2(earlier 2105.2)

The ratings of the bot at various time controls have been changed to reflect experience since the queen odds net was applied to the bot. The effect of both time and color is much larger than originally assumed, so the ratings of the bot in the rapid range are lower now and the ratings in the bullet and fast blitz range are higher now, and White is now valued at 100 above Black. Most of the ratings look much more reasonable now, but I don't quite know why your own rating is so high now. Lately you have been doing well at 10'5", but not well enough for your rating it seems to me. I'll let you know if we figure out the reason. If you play some more games, we can see if the adjustment is correct.

[Marcus91]

LeelaQueenOdds as white 5+3 (Reference) 2291.4
1 Master_Unknown 2266.6 38
LeelaQueenOdds as black 5+3 (Reference) 2241.4
2 Omnivor 2223.6 22
3 nn1133 2223.3 6167
LeelaQueenOdds as white 10+5 (Reference) 2210.9
4 Ajktulm 2197.7 181
5 uriblass 2180.9 134
LeelaQueenOdds as black 10+5 (Reference) 2160.9
LeelaQueenOdds as white 15+10 (Reference) 2155.2
6 Catecan 2123.0 272
7 Brunetticus 2110.3 404
LeelaQueenOdds as black 15+10 (Reference) 2105.2
8 Orichess88 2093.8 19
9 Respectful_Dave 2050.2 51
10 Hissha 2047.1 54
11 KentGoulding 2011.9 256
12 TheLlamaLord 2010.3 22

It seems that the rating of LeelaQueenOdds changed and the result is also changes in the rating of the players like Orichess88 without changing the number of games(I do not understand why my rating improved to 2217.1 with only one additional game relative to previous list.

My performance is certainly not 100 elo better than the last rating of LeelaQueenOdds as white

1 Master_Unknown 2448.3 38
2 Omnivor 2349.5 25
LeelaQueenOdds as black 3+2 (Reference) 2349.4
LeelaQueenOdds as white 5+3 (Reference) 2316.4(earlier 2291.4)
3 Orichess88 2263.8 19
4 Respectful_Dave 2248.6 62
5 Ajktulm 2228.1 195
6 uriblass 2217.1 135
LeelaQueenOdds as black 5+3 (Reference) 2216.4(earlier 2241.4)
7 Brunetticus 2202.5 421
8 TheLlamaLord 2194.6 22
9 nn1133 2175.4 6167
10 Jagen 2115.2 55
11 Catecan 2113.8 272
LeelaQueenOdds as white 10+5 (Reference) 2110.2(earlier 2210.9)
12 Forgive1219 2098.9 64
13 Punypawn315 2079.5 19
14 jzm 2047.6 15
15 Hissha 2047.1 54
16 KentGoulding 2026.0 258
LeelaQueenOdds as black 10+5 (Reference) 2010.2(earlier 2160.9)
17 susiedent 2003.1 14
18 desperado64 1990.1 41
19 SilverMitt 1975.5 13
20 SerbianPigeon 1971.9 17
21 junglegame 1958.3 9
22 OjaiJoao 1949.7 4
LeelaQueenOdds as white 15+10 (Reference) 1948.2(earlier 2155.2)
LeelaQueenOdds as black 15+10 (Reference) 1848.2(earlier 2105.2)

There was a bug, thanks for reporting!