Comparing top players and Niemann

[Ferdy]

Gather some games with classical time control of Niemann and other top players and analyze the games from move 12 to the end. Positions are analyzed by Stockfish 15 at one minute/pos on two threads and 512mb Hash in i7-2600k processor (3.4ghz, 4cores/8threads). Player's are compared from some performance metrics such as blunder, mistake, dubious moves, evaluation errors and others.

The table is like this.


Chrome: right click, open image in new tab.

Every id number is a single game. The names under column name are our subject. They won this game as indicated in the point column. Blunder/mistake/dubious/equal counts are indicated by b/m/d/e. These are further subdivided into 1, 2 and others.

Let's check id 0. Magnus has zero b, zero m, zero d and 1 e4. e4 is an equal position from [+50, +99] centipawn (cp) range. Equal position has an evaluation of [-49, +49] cp. So in this case, Magnus had played a suboptimal move from a slightly better position to an equal position.

This is the position. The score side POV.



merr is the mean or average error, where error is just engine_bestmove_eval less game_move_eval. Also called the average Centipawn Loss. There are 12 positions that are not yet winning (>= 300cp) and not yet losing (<= -300cp). The merr is 6 cp, and this is calculated when
the position is not yet winning/losing.

t1%, t2% and t3% are the percentage of top1, top2 and top3 moves from the engine that matches the game move played by Magnus. This is also calculated when the position is not yet winning/losing. So Magnus got 66.7% top1 move match from the engine. That would be 12x0.667 or 8 out of 12 positions.

werr is the mean or average error when the position is already winning. There are 8 won positions for him as indicated by wpos or winning positions. wt1% is the percentage of top1 engine move that matched his move. wfail% (0.0) is the percentage of winning position that he failed to maintain. In this case he never failed to find the best continuation.

In id 1, Niemann defeated Gretarsson. He got 75% top1 when position is not yet winning/losing, and 66.7% when he was already winning.

Let's have a look on some data. Rows highlighted with green color are good performances. Here we can see some random games from Niemann and how it compared with some top players.



More data with Carlsen - Niemann game highlighted in yellow.


Chrome: right click, open image in new tab. The table is sorted by date in ascending order.

Niemann had probably performed the best in this game so far. He got a high top1 of 83.3% when the position is not yet winning/lossing and 66.7 top1 when the position is already winning for him. Out of the 15 winning positions, he failed 6.7% or 15x0.067 or 1 position, to maintain the winning position.

This is the position. The score side POV.



It is not a blunder, not a mistake, not a dubious, and not an equal move, but a suboptimal move that is still better for him. According to Stockfish, the best is Rc2 with a score of 4.07 pawn unit searched to a depth of 31.

The annotated game.

[pgn][Event "9th Sinquefield Cup 2022"]
[Site "Saint Louis USA"]
[Date "2022.09.04"]
[Round "3.1"]
[White "Carlsen, Magnus"]
[Black "Niemann, Hans Moke"]
[Result "0-1"]
[BlackElo "2688"]
[BlackFideId "2093596"]
[BlackTitle "GM"]
[ECO "E20"]
[EventDate "2022.09.02"]
[Opening "Nimzo-Indian"]
[Variation "Romanishin-Kasparov (Steiner) system"]
[WhiteElo "2861"]
[WhiteFideId "1503014"]
[WhiteTitle "GM"]
[Annotator "Stockfish 15 @60s per pos, eval is SPOV"]

1.d4 Nf6 2.c4 e6 3.Nc3 Bb4 4.g3 O-O 5.Bg2 d5 6.a3 Bxc3+ 7.bxc3 dxc4 8.Nf3 c5 9.O-O cxd4 10.Qxd4 Nc6 11.Qxc4 e5 12.Bg5 { 0.00/31 } ( 12.a4 { 0.00/31 } ) 12...h6 { 0.00/31 } ( 12...Be6 { 0.09/31 } ) 13.Rfd1 { -0.24/34 } ( 13.Bxf6 { 0.00/34 } ) 13...Be6 { 0.37/32 } 14.Rxd8 { -0.30/34 } ( 14.Qc5 { -0.28/34 } ) 14...Bxc4 $7 { 0.31/31 } 15.Rxa8 { -0.29/31 } 15...Rxa8 $7 { 0.39/31 } 16.Bxf6 { -0.25/33 } ( 16.Nd2 { -0.19/33 } ) 16...gxf6 { 0.22/35 } 17.Kf1 { -0.30/40 } ( 17.Nd2 { -0.23/34 } ) 17...Rd8 { 0.45/39 } 18.Ke1 { -0.29/34 } 18...Na5 { 0.25/37 } 19.Rd1 { -0.22/36 } ( 19.Bh3 { -0.19/36 } ) 19...Rc8 { 0.28/39 } 20.Nd2 { -0.19/34 } ( 20.Bh3 { -0.15/34 } ) 20...Be6 { 0.28/38 } 21.c4 { -0.45/31 } ( 21.Ne4 { -0.21/32 } ) 21...Bxc4 { 0.23/39 } ( 21...b6 { 0.41/33 } ) 22.Nxc4 { -0.15/36 } ( 22.Ne4 { -0.13/36 } ) 22...Rxc4 { 0.20/29 } 23.Rd8+ { -0.12/35 } 23...Kg7 { 0.18/37 } 24.Bd5 { -0.17/36 } 24...Rc7 { 0.25/34 } 25.Ra8 { -0.39/41 } ( 25.f3 { -0.12/37 } ) 25...a6 { 0.37/37 } 26.Rb8 { -0.30/38 } ( 26.a4 { -0.25/38 } ) 26...f5 { 0.38/35 } 27.Re8 { -0.47/33 } ( 27.f3 { -0.18/33 } ) 27...e4 { 0.61/35 } 28.g4 $2 { -1.09/35 } ( 28.Rd8 { -0.48/33 } ) 28...Rc5 { 1.11/33 } 29.Ba2 { -1.95/30 } ( 29.Rd8 { -1.25/30 } ) 29...Nc4 { 1.69/32 } ( 29...fxg4 { 1.93/32 } ) 30.a4 { -2.25/32 } ( 30.Bxc4 { -1.77/33 } ) 30...Nd6 { 2.25/33 } 31.Re7 $4 { -4.12/32 } ( 31.Rd8 { -2.20/32 } ) 31...fxg4 $5 { 2.14/31 } ( 31...Rc2 { 4.07/31 } ) 32.Rd7 { -1.76/29 } 32...e3 { 1.89/31 } 33.fxe3 { -1.87/28 } 33...Ne4 { 1.84/33 } 34.Kf1 { -2.33/29 } 34...Rc1+ { 0.79/30 } ( 34...Rf5+ { 2.13/31 } ) 35.Kg2 { -1.04/30 } 35...Rc2 { 1.01/32 } 36.Bxf7 { -0.85/29 } 36...Rxe2+ { 0.69/31 } 37.Kg1 { -0.69/30 } 37...Re1+ { 0.94/32 } 38.Kg2 $7 { -0.66/37 } 38...Re2+ { 0.70/35 } 39.Kg1 { -0.69/30 } 39...Kf6 { 0.83/32 } ( 39...Re1+ { 0.94/32 } ) 40.Bd5 $2 { -1.27/34 } ( 40.Rxb7 { -0.51/34 } ) 40...Rd2 { 1.49/35 } 41.Rf7+ { -1.49/27 } 41...Kg6 { 1.50/30 } 42.Rd7 $4 { -5.77/29 } ( 42.Rf4 { -1.35/30 } ) 42...Ng5 { 5.73/30 } 43.Bf7+ { -7.18/27 } ( 43.Kf1 { -5.89/27 } ) 43...Kf5 { 4.65/29 } ( 43...Kf6 { 7.08/30 } ) 44.Rxd2 { -4.79/26 } 44...Nf3+ $7 { 2.66/25 } 45.Kg2 { -5.61/28 } 45...Nxd2 $7 { 5.11/24 } 46.a5 { -5.95/27 } 46...Ke5 { 6.67/32 } 47.Kg3 { -7.62/29 } ( 47.Kf2 { -7.45/29 } ) 47...Nf1+ { 7.76/39 } 48.Kf2 { -10.19/27 } ( 48.Kxg4 { -8.20/28 } ) 48...Nxh2 { 10.39/31 } 49.e4 { -13.52/31 } ( 49.Be8 { -11.54/29 } ) 49...Kxe4 { 13.65/30 } 50.Be6 { -15.80/30 } ( 50.Kg3 { -13.66/25 } ) 50...Kf4 { 15.83/28 } ( 50...h5 { 15.93/29 } ) 51.Bc8 { -21.66/25 } ( 51.Bf7 { -15.94/25 } ) 51...Nf3 { 22.79/25 } ( 51...h5 { 28.50/25 } ) 52.Bxb7 { -55.47/30 } 52...Ne5 { 55.47/33 } ( 52...Nd2 { 55.47/32 } ) 53.Bxa6 { -152.65/33 } 53...Nc6 { 152.65/37 } 54.Bb7 { -319.54/37 } ( 54.Bd3 { -152.65/31 } ) 54...Nxa5 { 319.77/45 } 55.Bd5 { -319.79/50 } 55...h5 { 319.79/48 } 56.Bf7 { -319.81/62 } 56...h4 { 56.94/29 } 57.Bd5 { -319.55/29 } 0-1[/pgn]

Code: Select all

[Event "9th Sinquefield Cup 2022"]
[Site "Saint Louis USA"]
[Date "2022.09.04"]
[Round "3.1"]
[White "Carlsen, Magnus"]
[Black "Niemann, Hans Moke"]
[Result "0-1"]
[BlackElo "2688"]
[BlackFideId "2093596"]
[BlackTitle "GM"]
[ECO "E20"]
[EventDate "2022.09.02"]
[Opening "Nimzo-Indian"]
[Variation "Romanishin-Kasparov (Steiner) system"]
[WhiteElo "2861"]
[WhiteFideId "1503014"]
[WhiteTitle "GM"]
[Annotator "Stockfish 15 @60s per pos, eval is SPOV"]

1.d4 Nf6 2.c4 e6 3.Nc3 Bb4 4.g3 O-O 5.Bg2 d5 6.a3 Bxc3+ 7.bxc3 dxc4 8.Nf3 c5 9.O-O cxd4 10.Qxd4 Nc6 11.Qxc4 e5 12.Bg5 { 0.00/31 } ( 12.a4 { 0.00/31 } ) 12...h6 { 0.00/31 } ( 12...Be6 { 0.09/31 } ) 13.Rfd1 { -0.24/34 } ( 13.Bxf6 { 0.00/34 } ) 13...Be6 { 0.37/32 } 14.Rxd8 { -0.30/34 } ( 14.Qc5 { -0.28/34 } ) 14...Bxc4 $7 { 0.31/31 } 15.Rxa8 { -0.29/31 } 15...Rxa8 $7 { 0.39/31 } 16.Bxf6 { -0.25/33 } ( 16.Nd2 { -0.19/33 } ) 16...gxf6 { 0.22/35 } 17.Kf1 { -0.30/40 } ( 17.Nd2 { -0.23/34 } ) 17...Rd8 { 0.45/39 } 18.Ke1 { -0.29/34 } 18...Na5 { 0.25/37 } 19.Rd1 { -0.22/36 } ( 19.Bh3 { -0.19/36 } ) 19...Rc8 { 0.28/39 } 20.Nd2 { -0.19/34 } ( 20.Bh3 { -0.15/34 } ) 20...Be6 { 0.28/38 } 21.c4 { -0.45/31 } ( 21.Ne4 { -0.21/32 } ) 21...Bxc4 { 0.23/39 } ( 21...b6 { 0.41/33 } ) 22.Nxc4 { -0.15/36 } ( 22.Ne4 { -0.13/36 } ) 22...Rxc4 { 0.20/29 } 23.Rd8+ { -0.12/35 } 23...Kg7 { 0.18/37 } 24.Bd5 { -0.17/36 } 24...Rc7 { 0.25/34 } 25.Ra8 { -0.39/41 } ( 25.f3 { -0.12/37 } ) 25...a6 { 0.37/37 } 26.Rb8 { -0.30/38 } ( 26.a4 { -0.25/38 } ) 26...f5 { 0.38/35 } 27.Re8 { -0.47/33 } ( 27.f3 { -0.18/33 } ) 27...e4 { 0.61/35 } 28.g4 $2 { -1.09/35 } ( 28.Rd8 { -0.48/33 } ) 28...Rc5 { 1.11/33 } 29.Ba2 { -1.95/30 } ( 29.Rd8 { -1.25/30 } ) 29...Nc4 { 1.69/32 } ( 29...fxg4 { 1.93/32 } ) 30.a4 { -2.25/32 } ( 30.Bxc4 { -1.77/33 } ) 30...Nd6 { 2.25/33 } 31.Re7 $4 { -4.12/32 } ( 31.Rd8 { -2.20/32 } ) 31...fxg4 $5 { 2.14/31 } ( 31...Rc2 { 4.07/31 } ) 32.Rd7 { -1.76/29 } 32...e3 { 1.89/31 } 33.fxe3 { -1.87/28 } 33...Ne4 { 1.84/33 } 34.Kf1 { -2.33/29 } 34...Rc1+ { 0.79/30 } ( 34...Rf5+ { 2.13/31 } ) 35.Kg2 { -1.04/30 } 35...Rc2 { 1.01/32 } 36.Bxf7 { -0.85/29 } 36...Rxe2+ { 0.69/31 } 37.Kg1 { -0.69/30 } 37...Re1+ { 0.94/32 } 38.Kg2 $7 { -0.66/37 } 38...Re2+ { 0.70/35 } 39.Kg1 { -0.69/30 } 39...Kf6 { 0.83/32 } ( 39...Re1+ { 0.94/32 } ) 40.Bd5 $2 { -1.27/34 } ( 40.Rxb7 { -0.51/34 } ) 40...Rd2 { 1.49/35 } 41.Rf7+ { -1.49/27 } 41...Kg6 { 1.50/30 } 42.Rd7 $4 { -5.77/29 } ( 42.Rf4 { -1.35/30 } ) 42...Ng5 { 5.73/30 } 43.Bf7+ { -7.18/27 } ( 43.Kf1 { -5.89/27 } ) 43...Kf5 { 4.65/29 } ( 43...Kf6 { 7.08/30 } ) 44.Rxd2 { -4.79/26 } 44...Nf3+ $7 { 2.66/25 } 45.Kg2 { -5.61/28 } 45...Nxd2 $7 { 5.11/24 } 46.a5 { -5.95/27 } 46...Ke5 { 6.67/32 } 47.Kg3 { -7.62/29 } ( 47.Kf2 { -7.45/29 } ) 47...Nf1+ { 7.76/39 } 48.Kf2 { -10.19/27 } ( 48.Kxg4 { -8.20/28 } ) 48...Nxh2 { 10.39/31 } 49.e4 { -13.52/31 } ( 49.Be8 { -11.54/29 } ) 49...Kxe4 { 13.65/30 } 50.Be6 { -15.80/30 } ( 50.Kg3 { -13.66/25 } ) 50...Kf4 { 15.83/28 } ( 50...h5 { 15.93/29 } ) 51.Bc8 { -21.66/25 } ( 51.Bf7 { -15.94/25 } ) 51...Nf3 { 22.79/25 } ( 51...h5 { 28.50/25 } ) 52.Bxb7 { -55.47/30 } 52...Ne5 { 55.47/33 } ( 52...Nd2 { 55.47/32 } ) 53.Bxa6 { -152.65/33 } 53...Nc6 { 152.65/37 } 54.Bb7 { -319.54/37 } ( 54.Bd3 { -152.65/31 } ) 54...Nxa5 { 319.77/45 } 55.Bd5 { -319.79/50 } 55...h5 { 319.79/48 } 56.Bf7 { -319.81/62 } 56...h4 { 56.94/29 } 57.Bd5 { -319.55/29 } 0-1

Now, I can understand why Magnus did this. Hans showed some accurate plays prior to this game, but this game is on top.

[Ferdy]

Position analysis from the player under the name column can be found in the api at https://yzfg9p.deta.dev/docs.

There is an endpoint at https://yzfg9p.deta.dev/position

You can query with epd/fen.

Code: Select all

https://yzfg9p.deta.dev/position?epd=2r3k1/pp3p2/4bp1p/n3p3/8/P1P3P1/3NPPBP/3RK3 w - - 8 21

Typical output:

Code: Select all

[{"epd":"2r3k1/pp3p2/4bp1p/n3p3/8/P1P3P1/3NPPBP/3RK3 w - -","move":"d2e4","score":-21,"depth":32,"engine":"Stockfish 15"},{"epd":"2r3k1/pp3p2/4bp1p/n3p3/8/P1P3P1/3NPPBP/3RK3 w - -","move":"c3c4","score":-45,"depth":31,"engine":"Stockfish 15"},{"epd":"2r3k1/pp3p2/4bp1p/n3p3/8/P1P3P1/3NPPBP/3RK3 w - -","move":"d1c1","score":-60,"depth":31,"engine":"Stockfish 15"}]

Or epd/fen with move.

Code: Select all

https://yzfg9p.deta.dev/position?epd=2r3k1/pp3p2/4bp1p/n3p3/8/P1P3P1/3NPPBP/3RK3 w - - 8 21&move=c3c4

[peter]

Gather some games with classical time control of Niemann and other top players and analyze the games from move 12 to the end. Positions are analyzed by Stockfish 15 at one minute/pos on two threads and 512mb Hash in i7-2600k processor (3.4ghz, 4cores/8threads). Player's are compared from some performance metrics such as blunder, mistake, dubious moves, evaluation errors and others.

The table is like this.


Chrome: right click, open image in new tab.

Every id number is a single game. The names under column name are our subject. They won this game as indicated in the point column. Blunder/mistake/dubious/equal counts are indicated by b/m/d/e. These are further subdivided into 1, 2 and others.

Let's check id 0. Magnus has zero b, zero m, zero d and 1 e4. e4 is an equal position from [+50, +99] centipawn (cp) range. Equal position has an evaluation of [-49, +49] cp. So in this case, Magnus had played a suboptimal move from a slightly better position to an equal position.

This is the position. The score side POV.



merr is the mean or average error, where error is just engine_bestmove_eval less game_move_eval. Also called the average Centipawn Loss. There are 12 positions that are not yet winning (>= 300cp) and not yet losing (<= -300cp). The merr is 6 cp, and this is calculated when
the position is not yet winning/losing.

t1%, t2% and t3% are the percentage of top1, top2 and top3 moves from the engine that matches the game move played by Magnus. This is also calculated when the position is not yet winning/losing. So Magnus got 66.7% top1 move match from the engine. That would be 12x0.667 or 8 out of 12 positions.

werr is the mean or average error when the position is already winning. There are 8 won positions for him as indicated by wpos or winning positions. wt1% is the percentage of top1 engine move that matched his move. wfail% (0.0) is the percentage of winning position that he failed to maintain. In this case he never failed to find the best continuation.

In id 1, Niemann defeated Gretarsson. He got 75% top1 when position is not yet winning/losing, and 66.7% when he was already winning.

Let's have a look on some data. Rows highlighted with green color are good performances. Here we can see some random games from Niemann and how it compared with some top players.



More data with Carlsen - Niemann game highlighted in yellow.


Chrome: right click, open image in new tab. The table is sorted by date in ascending order.

Niemann had probably performed the best in this game so far. He got a high top1 of 83.3% when the position is not yet winning/lossing and 66.7 top1 when the position is already winning for him. Out of the 15 winning positions, he failed 6.7% or 15x0.067 or 1 position, to maintain the winning position.

This is the position. The score side POV.



It is not a blunder, not a mistake, not a dubious, and not an equal move, but a suboptimal move that is still better for him. According to Stockfish, the best is Rc2 with a score of 4.07 pawn unit searched to a depth of 31.

The annotated game.

[pgn][Event "9th Sinquefield Cup 2022"]
[Site "Saint Louis USA"]
[Date "2022.09.04"]
[Round "3.1"]
[White "Carlsen, Magnus"]
[Black "Niemann, Hans Moke"]
[Result "0-1"]
[BlackElo "2688"]
[BlackFideId "2093596"]
[BlackTitle "GM"]
[ECO "E20"]
[EventDate "2022.09.02"]
[Opening "Nimzo-Indian"]
[Variation "Romanishin-Kasparov (Steiner) system"]
[WhiteElo "2861"]
[WhiteFideId "1503014"]
[WhiteTitle "GM"]
[Annotator "Stockfish 15 @60s per pos, eval is SPOV"]

1.d4 Nf6 2.c4 e6 3.Nc3 Bb4 4.g3 O-O 5.Bg2 d5 6.a3 Bxc3+ 7.bxc3 dxc4 8.Nf3 c5 9.O-O cxd4 10.Qxd4 Nc6 11.Qxc4 e5 12.Bg5 { 0.00/31 } ( 12.a4 { 0.00/31 } ) 12...h6 { 0.00/31 } ( 12...Be6 { 0.09/31 } ) 13.Rfd1 { -0.24/34 } ( 13.Bxf6 { 0.00/34 } ) 13...Be6 { 0.37/32 } 14.Rxd8 { -0.30/34 } ( 14.Qc5 { -0.28/34 } ) 14...Bxc4 $7 { 0.31/31 } 15.Rxa8 { -0.29/31 } 15...Rxa8 $7 { 0.39/31 } 16.Bxf6 { -0.25/33 } ( 16.Nd2 { -0.19/33 } ) 16...gxf6 { 0.22/35 } 17.Kf1 { -0.30/40 } ( 17.Nd2 { -0.23/34 } ) 17...Rd8 { 0.45/39 } 18.Ke1 { -0.29/34 } 18...Na5 { 0.25/37 } 19.Rd1 { -0.22/36 } ( 19.Bh3 { -0.19/36 } ) 19...Rc8 { 0.28/39 } 20.Nd2 { -0.19/34 } ( 20.Bh3 { -0.15/34 } ) 20...Be6 { 0.28/38 } 21.c4 { -0.45/31 } ( 21.Ne4 { -0.21/32 } ) 21...Bxc4 { 0.23/39 } ( 21...b6 { 0.41/33 } ) 22.Nxc4 { -0.15/36 } ( 22.Ne4 { -0.13/36 } ) 22...Rxc4 { 0.20/29 } 23.Rd8+ { -0.12/35 } 23...Kg7 { 0.18/37 } 24.Bd5 { -0.17/36 } 24...Rc7 { 0.25/34 } 25.Ra8 { -0.39/41 } ( 25.f3 { -0.12/37 } ) 25...a6 { 0.37/37 } 26.Rb8 { -0.30/38 } ( 26.a4 { -0.25/38 } ) 26...f5 { 0.38/35 } 27.Re8 { -0.47/33 } ( 27.f3 { -0.18/33 } ) 27...e4 { 0.61/35 } 28.g4 $2 { -1.09/35 } ( 28.Rd8 { -0.48/33 } ) 28...Rc5 { 1.11/33 } 29.Ba2 { -1.95/30 } ( 29.Rd8 { -1.25/30 } ) 29...Nc4 { 1.69/32 } ( 29...fxg4 { 1.93/32 } ) 30.a4 { -2.25/32 } ( 30.Bxc4 { -1.77/33 } ) 30...Nd6 { 2.25/33 } 31.Re7 $4 { -4.12/32 } ( 31.Rd8 { -2.20/32 } ) 31...fxg4 $5 { 2.14/31 } ( 31...Rc2 { 4.07/31 } ) 32.Rd7 { -1.76/29 } 32...e3 { 1.89/31 } 33.fxe3 { -1.87/28 } 33...Ne4 { 1.84/33 } 34.Kf1 { -2.33/29 } 34...Rc1+ { 0.79/30 } ( 34...Rf5+ { 2.13/31 } ) 35.Kg2 { -1.04/30 } 35...Rc2 { 1.01/32 } 36.Bxf7 { -0.85/29 } 36...Rxe2+ { 0.69/31 } 37.Kg1 { -0.69/30 } 37...Re1+ { 0.94/32 } 38.Kg2 $7 { -0.66/37 } 38...Re2+ { 0.70/35 } 39.Kg1 { -0.69/30 } 39...Kf6 { 0.83/32 } ( 39...Re1+ { 0.94/32 } ) 40.Bd5 $2 { -1.27/34 } ( 40.Rxb7 { -0.51/34 } ) 40...Rd2 { 1.49/35 } 41.Rf7+ { -1.49/27 } 41...Kg6 { 1.50/30 } 42.Rd7 $4 { -5.77/29 } ( 42.Rf4 { -1.35/30 } ) 42...Ng5 { 5.73/30 } 43.Bf7+ { -7.18/27 } ( 43.Kf1 { -5.89/27 } ) 43...Kf5 { 4.65/29 } ( 43...Kf6 { 7.08/30 } ) 44.Rxd2 { -4.79/26 } 44...Nf3+ $7 { 2.66/25 } 45.Kg2 { -5.61/28 } 45...Nxd2 $7 { 5.11/24 } 46.a5 { -5.95/27 } 46...Ke5 { 6.67/32 } 47.Kg3 { -7.62/29 } ( 47.Kf2 { -7.45/29 } ) 47...Nf1+ { 7.76/39 } 48.Kf2 { -10.19/27 } ( 48.Kxg4 { -8.20/28 } ) 48...Nxh2 { 10.39/31 } 49.e4 { -13.52/31 } ( 49.Be8 { -11.54/29 } ) 49...Kxe4 { 13.65/30 } 50.Be6 { -15.80/30 } ( 50.Kg3 { -13.66/25 } ) 50...Kf4 { 15.83/28 } ( 50...h5 { 15.93/29 } ) 51.Bc8 { -21.66/25 } ( 51.Bf7 { -15.94/25 } ) 51...Nf3 { 22.79/25 } ( 51...h5 { 28.50/25 } ) 52.Bxb7 { -55.47/30 } 52...Ne5 { 55.47/33 } ( 52...Nd2 { 55.47/32 } ) 53.Bxa6 { -152.65/33 } 53...Nc6 { 152.65/37 } 54.Bb7 { -319.54/37 } ( 54.Bd3 { -152.65/31 } ) 54...Nxa5 { 319.77/45 } 55.Bd5 { -319.79/50 } 55...h5 { 319.79/48 } 56.Bf7 { -319.81/62 } 56...h4 { 56.94/29 } 57.Bd5 { -319.55/29 } 0-1[/pgn]

Code: Select all

[Event "9th Sinquefield Cup 2022"]
[Site "Saint Louis USA"]
[Date "2022.09.04"]
[Round "3.1"]
[White "Carlsen, Magnus"]
[Black "Niemann, Hans Moke"]
[Result "0-1"]
[BlackElo "2688"]
[BlackFideId "2093596"]
[BlackTitle "GM"]
[ECO "E20"]
[EventDate "2022.09.02"]
[Opening "Nimzo-Indian"]
[Variation "Romanishin-Kasparov (Steiner) system"]
[WhiteElo "2861"]
[WhiteFideId "1503014"]
[WhiteTitle "GM"]
[Annotator "Stockfish 15 @60s per pos, eval is SPOV"]

1.d4 Nf6 2.c4 e6 3.Nc3 Bb4 4.g3 O-O 5.Bg2 d5 6.a3 Bxc3+ 7.bxc3 dxc4 8.Nf3 c5 9.O-O cxd4 10.Qxd4 Nc6 11.Qxc4 e5 12.Bg5 { 0.00/31 } ( 12.a4 { 0.00/31 } ) 12...h6 { 0.00/31 } ( 12...Be6 { 0.09/31 } ) 13.Rfd1 { -0.24/34 } ( 13.Bxf6 { 0.00/34 } ) 13...Be6 { 0.37/32 } 14.Rxd8 { -0.30/34 } ( 14.Qc5 { -0.28/34 } ) 14...Bxc4 $7 { 0.31/31 } 15.Rxa8 { -0.29/31 } 15...Rxa8 $7 { 0.39/31 } 16.Bxf6 { -0.25/33 } ( 16.Nd2 { -0.19/33 } ) 16...gxf6 { 0.22/35 } 17.Kf1 { -0.30/40 } ( 17.Nd2 { -0.23/34 } ) 17...Rd8 { 0.45/39 } 18.Ke1 { -0.29/34 } 18...Na5 { 0.25/37 } 19.Rd1 { -0.22/36 } ( 19.Bh3 { -0.19/36 } ) 19...Rc8 { 0.28/39 } 20.Nd2 { -0.19/34 } ( 20.Bh3 { -0.15/34 } ) 20...Be6 { 0.28/38 } 21.c4 { -0.45/31 } ( 21.Ne4 { -0.21/32 } ) 21...Bxc4 { 0.23/39 } ( 21...b6 { 0.41/33 } ) 22.Nxc4 { -0.15/36 } ( 22.Ne4 { -0.13/36 } ) 22...Rxc4 { 0.20/29 } 23.Rd8+ { -0.12/35 } 23...Kg7 { 0.18/37 } 24.Bd5 { -0.17/36 } 24...Rc7 { 0.25/34 } 25.Ra8 { -0.39/41 } ( 25.f3 { -0.12/37 } ) 25...a6 { 0.37/37 } 26.Rb8 { -0.30/38 } ( 26.a4 { -0.25/38 } ) 26...f5 { 0.38/35 } 27.Re8 { -0.47/33 } ( 27.f3 { -0.18/33 } ) 27...e4 { 0.61/35 } 28.g4 $2 { -1.09/35 } ( 28.Rd8 { -0.48/33 } ) 28...Rc5 { 1.11/33 } 29.Ba2 { -1.95/30 } ( 29.Rd8 { -1.25/30 } ) 29...Nc4 { 1.69/32 } ( 29...fxg4 { 1.93/32 } ) 30.a4 { -2.25/32 } ( 30.Bxc4 { -1.77/33 } ) 30...Nd6 { 2.25/33 } 31.Re7 $4 { -4.12/32 } ( 31.Rd8 { -2.20/32 } ) 31...fxg4 $5 { 2.14/31 } ( 31...Rc2 { 4.07/31 } ) 32.Rd7 { -1.76/29 } 32...e3 { 1.89/31 } 33.fxe3 { -1.87/28 } 33...Ne4 { 1.84/33 } 34.Kf1 { -2.33/29 } 34...Rc1+ { 0.79/30 } ( 34...Rf5+ { 2.13/31 } ) 35.Kg2 { -1.04/30 } 35...Rc2 { 1.01/32 } 36.Bxf7 { -0.85/29 } 36...Rxe2+ { 0.69/31 } 37.Kg1 { -0.69/30 } 37...Re1+ { 0.94/32 } 38.Kg2 $7 { -0.66/37 } 38...Re2+ { 0.70/35 } 39.Kg1 { -0.69/30 } 39...Kf6 { 0.83/32 } ( 39...Re1+ { 0.94/32 } ) 40.Bd5 $2 { -1.27/34 } ( 40.Rxb7 { -0.51/34 } ) 40...Rd2 { 1.49/35 } 41.Rf7+ { -1.49/27 } 41...Kg6 { 1.50/30 } 42.Rd7 $4 { -5.77/29 } ( 42.Rf4 { -1.35/30 } ) 42...Ng5 { 5.73/30 } 43.Bf7+ { -7.18/27 } ( 43.Kf1 { -5.89/27 } ) 43...Kf5 { 4.65/29 } ( 43...Kf6 { 7.08/30 } ) 44.Rxd2 { -4.79/26 } 44...Nf3+ $7 { 2.66/25 } 45.Kg2 { -5.61/28 } 45...Nxd2 $7 { 5.11/24 } 46.a5 { -5.95/27 } 46...Ke5 { 6.67/32 } 47.Kg3 { -7.62/29 } ( 47.Kf2 { -7.45/29 } ) 47...Nf1+ { 7.76/39 } 48.Kf2 { -10.19/27 } ( 48.Kxg4 { -8.20/28 } ) 48...Nxh2 { 10.39/31 } 49.e4 { -13.52/31 } ( 49.Be8 { -11.54/29 } ) 49...Kxe4 { 13.65/30 } 50.Be6 { -15.80/30 } ( 50.Kg3 { -13.66/25 } ) 50...Kf4 { 15.83/28 } ( 50...h5 { 15.93/29 } ) 51.Bc8 { -21.66/25 } ( 51.Bf7 { -15.94/25 } ) 51...Nf3 { 22.79/25 } ( 51...h5 { 28.50/25 } ) 52.Bxb7 { -55.47/30 } 52...Ne5 { 55.47/33 } ( 52...Nd2 { 55.47/32 } ) 53.Bxa6 { -152.65/33 } 53...Nc6 { 152.65/37 } 54.Bb7 { -319.54/37 } ( 54.Bd3 { -152.65/31 } ) 54...Nxa5 { 319.77/45 } 55.Bd5 { -319.79/50 } 55...h5 { 319.79/48 } 56.Bf7 { -319.81/62 } 56...h4 { 56.94/29 } 57.Bd5 { -319.55/29 } 0-1

Now, I can understand why Magnus did this. Hans showed some accurate plays prior to this game, but this game is on top.

Many thanks for the work, Ferdy, regards

[chrisw]

Gather some games with classical time control of Niemann and other top players and analyze the games from move 12 to the end. Positions are analyzed by Stockfish 15 at one minute/pos on two threads and 512mb Hash in i7-2600k processor (3.4ghz, 4cores/8threads). Player's are compared from some performance metrics such as blunder, mistake, dubious moves, evaluation errors and others.

The table is like this.


Chrome: right click, open image in new tab.

Every id number is a single game. The names under column name are our subject. They won this game as indicated in the point column. Blunder/mistake/dubious/equal counts are indicated by b/m/d/e. These are further subdivided into 1, 2 and others.

Let's check id 0. Magnus has zero b, zero m, zero d and 1 e4. e4 is an equal position from [+50, +99] centipawn (cp) range. Equal position has an evaluation of [-49, +49] cp. So in this case, Magnus had played a suboptimal move from a slightly better position to an equal position.

This is the position. The score side POV.



merr is the mean or average error, where error is just engine_bestmove_eval less game_move_eval. Also called the average Centipawn Loss. There are 12 positions that are not yet winning (>= 300cp) and not yet losing (<= -300cp). The merr is 6 cp, and this is calculated when
the position is not yet winning/losing.

t1%, t2% and t3% are the percentage of top1, top2 and top3 moves from the engine that matches the game move played by Magnus. This is also calculated when the position is not yet winning/losing. So Magnus got 66.7% top1 move match from the engine. That would be 12x0.667 or 8 out of 12 positions.

werr is the mean or average error when the position is already winning. There are 8 won positions for him as indicated by wpos or winning positions. wt1% is the percentage of top1 engine move that matched his move. wfail% (0.0) is the percentage of winning position that he failed to maintain. In this case he never failed to find the best continuation.

In id 1, Niemann defeated Gretarsson. He got 75% top1 when position is not yet winning/losing, and 66.7% when he was already winning.

Let's have a look on some data. Rows highlighted with green color are good performances. Here we can see some random games from Niemann and how it compared with some top players.



More data with Carlsen - Niemann game highlighted in yellow.


Chrome: right click, open image in new tab. The table is sorted by date in ascending order.

Niemann had probably performed the best in this game so far. He got a high top1 of 83.3% when the position is not yet winning/lossing and 66.7 top1 when the position is already winning for him. Out of the 15 winning positions, he failed 6.7% or 15x0.067 or 1 position, to maintain the winning position.

This is the position. The score side POV.



It is not a blunder, not a mistake, not a dubious, and not an equal move, but a suboptimal move that is still better for him. According to Stockfish, the best is Rc2 with a score of 4.07 pawn unit searched to a depth of 31.

The annotated game.

[pgn][Event "9th Sinquefield Cup 2022"]
[Site "Saint Louis USA"]
[Date "2022.09.04"]
[Round "3.1"]
[White "Carlsen, Magnus"]
[Black "Niemann, Hans Moke"]
[Result "0-1"]
[BlackElo "2688"]
[BlackFideId "2093596"]
[BlackTitle "GM"]
[ECO "E20"]
[EventDate "2022.09.02"]
[Opening "Nimzo-Indian"]
[Variation "Romanishin-Kasparov (Steiner) system"]
[WhiteElo "2861"]
[WhiteFideId "1503014"]
[WhiteTitle "GM"]
[Annotator "Stockfish 15 @60s per pos, eval is SPOV"]

1.d4 Nf6 2.c4 e6 3.Nc3 Bb4 4.g3 O-O 5.Bg2 d5 6.a3 Bxc3+ 7.bxc3 dxc4 8.Nf3 c5 9.O-O cxd4 10.Qxd4 Nc6 11.Qxc4 e5 12.Bg5 { 0.00/31 } ( 12.a4 { 0.00/31 } ) 12...h6 { 0.00/31 } ( 12...Be6 { 0.09/31 } ) 13.Rfd1 { -0.24/34 } ( 13.Bxf6 { 0.00/34 } ) 13...Be6 { 0.37/32 } 14.Rxd8 { -0.30/34 } ( 14.Qc5 { -0.28/34 } ) 14...Bxc4 $7 { 0.31/31 } 15.Rxa8 { -0.29/31 } 15...Rxa8 $7 { 0.39/31 } 16.Bxf6 { -0.25/33 } ( 16.Nd2 { -0.19/33 } ) 16...gxf6 { 0.22/35 } 17.Kf1 { -0.30/40 } ( 17.Nd2 { -0.23/34 } ) 17...Rd8 { 0.45/39 } 18.Ke1 { -0.29/34 } 18...Na5 { 0.25/37 } 19.Rd1 { -0.22/36 } ( 19.Bh3 { -0.19/36 } ) 19...Rc8 { 0.28/39 } 20.Nd2 { -0.19/34 } ( 20.Bh3 { -0.15/34 } ) 20...Be6 { 0.28/38 } 21.c4 { -0.45/31 } ( 21.Ne4 { -0.21/32 } ) 21...Bxc4 { 0.23/39 } ( 21...b6 { 0.41/33 } ) 22.Nxc4 { -0.15/36 } ( 22.Ne4 { -0.13/36 } ) 22...Rxc4 { 0.20/29 } 23.Rd8+ { -0.12/35 } 23...Kg7 { 0.18/37 } 24.Bd5 { -0.17/36 } 24...Rc7 { 0.25/34 } 25.Ra8 { -0.39/41 } ( 25.f3 { -0.12/37 } ) 25...a6 { 0.37/37 } 26.Rb8 { -0.30/38 } ( 26.a4 { -0.25/38 } ) 26...f5 { 0.38/35 } 27.Re8 { -0.47/33 } ( 27.f3 { -0.18/33 } ) 27...e4 { 0.61/35 } 28.g4 $2 { -1.09/35 } ( 28.Rd8 { -0.48/33 } ) 28...Rc5 { 1.11/33 } 29.Ba2 { -1.95/30 } ( 29.Rd8 { -1.25/30 } ) 29...Nc4 { 1.69/32 } ( 29...fxg4 { 1.93/32 } ) 30.a4 { -2.25/32 } ( 30.Bxc4 { -1.77/33 } ) 30...Nd6 { 2.25/33 } 31.Re7 $4 { -4.12/32 } ( 31.Rd8 { -2.20/32 } ) 31...fxg4 $5 { 2.14/31 } ( 31...Rc2 { 4.07/31 } ) 32.Rd7 { -1.76/29 } 32...e3 { 1.89/31 } 33.fxe3 { -1.87/28 } 33...Ne4 { 1.84/33 } 34.Kf1 { -2.33/29 } 34...Rc1+ { 0.79/30 } ( 34...Rf5+ { 2.13/31 } ) 35.Kg2 { -1.04/30 } 35...Rc2 { 1.01/32 } 36.Bxf7 { -0.85/29 } 36...Rxe2+ { 0.69/31 } 37.Kg1 { -0.69/30 } 37...Re1+ { 0.94/32 } 38.Kg2 $7 { -0.66/37 } 38...Re2+ { 0.70/35 } 39.Kg1 { -0.69/30 } 39...Kf6 { 0.83/32 } ( 39...Re1+ { 0.94/32 } ) 40.Bd5 $2 { -1.27/34 } ( 40.Rxb7 { -0.51/34 } ) 40...Rd2 { 1.49/35 } 41.Rf7+ { -1.49/27 } 41...Kg6 { 1.50/30 } 42.Rd7 $4 { -5.77/29 } ( 42.Rf4 { -1.35/30 } ) 42...Ng5 { 5.73/30 } 43.Bf7+ { -7.18/27 } ( 43.Kf1 { -5.89/27 } ) 43...Kf5 { 4.65/29 } ( 43...Kf6 { 7.08/30 } ) 44.Rxd2 { -4.79/26 } 44...Nf3+ $7 { 2.66/25 } 45.Kg2 { -5.61/28 } 45...Nxd2 $7 { 5.11/24 } 46.a5 { -5.95/27 } 46...Ke5 { 6.67/32 } 47.Kg3 { -7.62/29 } ( 47.Kf2 { -7.45/29 } ) 47...Nf1+ { 7.76/39 } 48.Kf2 { -10.19/27 } ( 48.Kxg4 { -8.20/28 } ) 48...Nxh2 { 10.39/31 } 49.e4 { -13.52/31 } ( 49.Be8 { -11.54/29 } ) 49...Kxe4 { 13.65/30 } 50.Be6 { -15.80/30 } ( 50.Kg3 { -13.66/25 } ) 50...Kf4 { 15.83/28 } ( 50...h5 { 15.93/29 } ) 51.Bc8 { -21.66/25 } ( 51.Bf7 { -15.94/25 } ) 51...Nf3 { 22.79/25 } ( 51...h5 { 28.50/25 } ) 52.Bxb7 { -55.47/30 } 52...Ne5 { 55.47/33 } ( 52...Nd2 { 55.47/32 } ) 53.Bxa6 { -152.65/33 } 53...Nc6 { 152.65/37 } 54.Bb7 { -319.54/37 } ( 54.Bd3 { -152.65/31 } ) 54...Nxa5 { 319.77/45 } 55.Bd5 { -319.79/50 } 55...h5 { 319.79/48 } 56.Bf7 { -319.81/62 } 56...h4 { 56.94/29 } 57.Bd5 { -319.55/29 } 0-1[/pgn]

Code: Select all

[Event "9th Sinquefield Cup 2022"]
[Site "Saint Louis USA"]
[Date "2022.09.04"]
[Round "3.1"]
[White "Carlsen, Magnus"]
[Black "Niemann, Hans Moke"]
[Result "0-1"]
[BlackElo "2688"]
[BlackFideId "2093596"]
[BlackTitle "GM"]
[ECO "E20"]
[EventDate "2022.09.02"]
[Opening "Nimzo-Indian"]
[Variation "Romanishin-Kasparov (Steiner) system"]
[WhiteElo "2861"]
[WhiteFideId "1503014"]
[WhiteTitle "GM"]
[Annotator "Stockfish 15 @60s per pos, eval is SPOV"]

1.d4 Nf6 2.c4 e6 3.Nc3 Bb4 4.g3 O-O 5.Bg2 d5 6.a3 Bxc3+ 7.bxc3 dxc4 8.Nf3 c5 9.O-O cxd4 10.Qxd4 Nc6 11.Qxc4 e5 12.Bg5 { 0.00/31 } ( 12.a4 { 0.00/31 } ) 12...h6 { 0.00/31 } ( 12...Be6 { 0.09/31 } ) 13.Rfd1 { -0.24/34 } ( 13.Bxf6 { 0.00/34 } ) 13...Be6 { 0.37/32 } 14.Rxd8 { -0.30/34 } ( 14.Qc5 { -0.28/34 } ) 14...Bxc4 $7 { 0.31/31 } 15.Rxa8 { -0.29/31 } 15...Rxa8 $7 { 0.39/31 } 16.Bxf6 { -0.25/33 } ( 16.Nd2 { -0.19/33 } ) 16...gxf6 { 0.22/35 } 17.Kf1 { -0.30/40 } ( 17.Nd2 { -0.23/34 } ) 17...Rd8 { 0.45/39 } 18.Ke1 { -0.29/34 } 18...Na5 { 0.25/37 } 19.Rd1 { -0.22/36 } ( 19.Bh3 { -0.19/36 } ) 19...Rc8 { 0.28/39 } 20.Nd2 { -0.19/34 } ( 20.Bh3 { -0.15/34 } ) 20...Be6 { 0.28/38 } 21.c4 { -0.45/31 } ( 21.Ne4 { -0.21/32 } ) 21...Bxc4 { 0.23/39 } ( 21...b6 { 0.41/33 } ) 22.Nxc4 { -0.15/36 } ( 22.Ne4 { -0.13/36 } ) 22...Rxc4 { 0.20/29 } 23.Rd8+ { -0.12/35 } 23...Kg7 { 0.18/37 } 24.Bd5 { -0.17/36 } 24...Rc7 { 0.25/34 } 25.Ra8 { -0.39/41 } ( 25.f3 { -0.12/37 } ) 25...a6 { 0.37/37 } 26.Rb8 { -0.30/38 } ( 26.a4 { -0.25/38 } ) 26...f5 { 0.38/35 } 27.Re8 { -0.47/33 } ( 27.f3 { -0.18/33 } ) 27...e4 { 0.61/35 } 28.g4 $2 { -1.09/35 } ( 28.Rd8 { -0.48/33 } ) 28...Rc5 { 1.11/33 } 29.Ba2 { -1.95/30 } ( 29.Rd8 { -1.25/30 } ) 29...Nc4 { 1.69/32 } ( 29...fxg4 { 1.93/32 } ) 30.a4 { -2.25/32 } ( 30.Bxc4 { -1.77/33 } ) 30...Nd6 { 2.25/33 } 31.Re7 $4 { -4.12/32 } ( 31.Rd8 { -2.20/32 } ) 31...fxg4 $5 { 2.14/31 } ( 31...Rc2 { 4.07/31 } ) 32.Rd7 { -1.76/29 } 32...e3 { 1.89/31 } 33.fxe3 { -1.87/28 } 33...Ne4 { 1.84/33 } 34.Kf1 { -2.33/29 } 34...Rc1+ { 0.79/30 } ( 34...Rf5+ { 2.13/31 } ) 35.Kg2 { -1.04/30 } 35...Rc2 { 1.01/32 } 36.Bxf7 { -0.85/29 } 36...Rxe2+ { 0.69/31 } 37.Kg1 { -0.69/30 } 37...Re1+ { 0.94/32 } 38.Kg2 $7 { -0.66/37 } 38...Re2+ { 0.70/35 } 39.Kg1 { -0.69/30 } 39...Kf6 { 0.83/32 } ( 39...Re1+ { 0.94/32 } ) 40.Bd5 $2 { -1.27/34 } ( 40.Rxb7 { -0.51/34 } ) 40...Rd2 { 1.49/35 } 41.Rf7+ { -1.49/27 } 41...Kg6 { 1.50/30 } 42.Rd7 $4 { -5.77/29 } ( 42.Rf4 { -1.35/30 } ) 42...Ng5 { 5.73/30 } 43.Bf7+ { -7.18/27 } ( 43.Kf1 { -5.89/27 } ) 43...Kf5 { 4.65/29 } ( 43...Kf6 { 7.08/30 } ) 44.Rxd2 { -4.79/26 } 44...Nf3+ $7 { 2.66/25 } 45.Kg2 { -5.61/28 } 45...Nxd2 $7 { 5.11/24 } 46.a5 { -5.95/27 } 46...Ke5 { 6.67/32 } 47.Kg3 { -7.62/29 } ( 47.Kf2 { -7.45/29 } ) 47...Nf1+ { 7.76/39 } 48.Kf2 { -10.19/27 } ( 48.Kxg4 { -8.20/28 } ) 48...Nxh2 { 10.39/31 } 49.e4 { -13.52/31 } ( 49.Be8 { -11.54/29 } ) 49...Kxe4 { 13.65/30 } 50.Be6 { -15.80/30 } ( 50.Kg3 { -13.66/25 } ) 50...Kf4 { 15.83/28 } ( 50...h5 { 15.93/29 } ) 51.Bc8 { -21.66/25 } ( 51.Bf7 { -15.94/25 } ) 51...Nf3 { 22.79/25 } ( 51...h5 { 28.50/25 } ) 52.Bxb7 { -55.47/30 } 52...Ne5 { 55.47/33 } ( 52...Nd2 { 55.47/32 } ) 53.Bxa6 { -152.65/33 } 53...Nc6 { 152.65/37 } 54.Bb7 { -319.54/37 } ( 54.Bd3 { -152.65/31 } ) 54...Nxa5 { 319.77/45 } 55.Bd5 { -319.79/50 } 55...h5 { 319.79/48 } 56.Bf7 { -319.81/62 } 56...h4 { 56.94/29 } 57.Bd5 { -319.55/29 } 0-1

Now, I can understand why Magnus did this. Hans showed some accurate plays prior to this game, but this game is on top.

Hans’s profile is also consistent with being a genius new paradigm player and potential world champion. He’s new paradigm because he’s from the generation that developed entirely under the influence of AZ, Leela and SF NNUE and has been able to internalise imbalance, fortress and activity/paralysis concepts from them. It’s not so much that the concepts are new, it’s more a case of understanding that despite material deficit some structures are sufficient for a win.
Magnus was right when he said Hans played moves “above his grade” and which he couldn’t understand, but wrong when he said this must be from cheating. The old dinosaurs never can understand the new.

[M ANSARI]

Well ... MC the old "dinosaur" seems to have done well and easily beat Arjun Erigaisi ... one of the top new prodigies in chess and much higher rated than Hans Nieman. Only difference is that Hans Nieman is an admitted serial cheater while Arjun has never been accused of cheating. I really doubt that chess is so simple that studying moves from SF NNUE and LC0 can make you play like those engines. For sure it can add depth to your game and you can use some methods that NNUE engines have shown to be extremely strong ... like pushing and embedding a pawn deep ... but MC seems to be the one that has most gained from that.

You could be right that Hans Nieman is just misunderstood young kid who did mistakes and that he is the strongest human chess player ever. But another simple explanation could be that he is cheating. Time will tell which one it is.

[Uri Blass]

Well ... MC the old "dinosaur" seems to have done well and easily beat Arjun Erigaisi ... one of the top new prodigies in chess and much higher rated than Hans Nieman. Only difference is that Hans Nieman is an admitted serial cheater while Arjun has never been accused of cheating. I really doubt that chess is so simple that studying moves from SF NNUE and LC0 can make you play like those engines. For sure it can add depth to your game and you can use some methods that NNUE engines have shown to be extremely strong ... like pushing and embedding a pawn deep ... but MC seems to be the one that has most gained from that.

You could be right that Hans Nieman is just misunderstood young kid who did mistakes and that he is the strongest human chess player ever. But another simple explanation could be that he is cheating. Time will tell which one it is.

If studying moves from SF NNUE and LC0 can make you play like those engines in some types of positions then you can achieve 100% correlation in some games and not be the strongest chess player because not in every game you get the positions that you know to play.

Hans is certainly not the strongest human chess player based on his results but even if in some games(and not in every game) he can play 100% like an engine it does not prove that he is cheating.

It may be interesting to make a competition between humans in predicting different engine moves without using an engine to see if humans really can learn the style of chess engines.

Take 2 strong engines A and B and some human who can prepare by learning both A and B.
You can use A=Stockfish15 B=Dragon3.1 but you can also use something different.

Take many positions and anlayze the position by both A and B and choose from them 1000 positions when the choice of A is different than the choice of B.

What is the percentage correct that the human may guess for A and B out of 1000 positions with no engine help?

Something like 40% for A and 38% for B but if you replace the choice between A and B only 18% for A and 19% for B suggest that the human is able to learn the engine style.

Note that there may be cases when the best strategy of human may be to choose the same move for both A and B if the target of the human is to get the most correct when you get a point for every correct and you can get at most 1 out of 2 by guessing the same move for both engines(because you know in advance that the moves are different) but you have a risk of getting 0 out of 2 if you guess different moves for both engines.

[Chessqueen]

Well ... MC the old "dinosaur" seems to have done well and easily beat Arjun Erigaisi ... one of the top new prodigies in chess and much higher rated than Hans Nieman. Only difference is that Hans Nieman is an admitted serial cheater while Arjun has never been accused of cheating. I really doubt that chess is so simple that studying moves from SF NNUE and LC0 can make you play like those engines. For sure it can add depth to your game and you can use some methods that NNUE engines have shown to be extremely strong ... like pushing and embedding a pawn deep ... but MC seems to be the one that has most gained from that.

You could be right that Hans Nieman is just misunderstood young kid who did mistakes and that he is the strongest human chess player ever. But another simple explanation could be that he is cheating. Time will tell which one it is.

If studying moves from SF NNUE and LC0 can make you play like those engines in some types of positions then you can achieve 100% correlation in some games and not be the strongest chess player because not in every game you get the positions that you know to play.

Hans is certainly not the strongest human chess player based on his results but even if in some games(and not in every game) he can play 100% like an engine it does not prove that he is cheating.

It may be interesting to make a competition between humans in predicting different engine moves without using an engine to see if humans really can learn the style of chess engines.

Take 2 strong engines A and B and some human who can prepare by learning both A and B.
You can use A=Stockfish15 B=Dragon3.1 but you can also use something different.

Take many positions and anlayze the position by both A and B and choose from them 1000 positions when the choice of A is different than the choice of B.

What is the percentage correct that the human may guess for A and B out of 1000 positions with no engine help?

Something like 40% for A and 38% for B but if you replace the choice between A and B only 18% for A and 19% for B suggest that the human is able to learn the engine style.

Note that there may be cases when the best strategy of human may be to choose the same move for both A and B if the target of the human is to get the most correct when you get a point for every correct and you can get at most 1 out of 2 by guessing the same move for both engines(because you know in advance that the moves are different) but you have a risk of getting 0 out of 2 if you guess different moves for both engines.

Another, point that I would like to mention is that in the past we have posted several times that even if you take a top player rated around 2680 like Niemann and let him play 1 move 2 moves or three moves from any top engine during his game versus player of the caliber of Carlsen , GM Ding Liren, or GM Caruana, he would not be able to beat them, since during those games he would eventually make a few inferior moves that Carlsen, or GM Ding Liren would immediately punish and turn the game in their favor. So the only explanation is that GM Hans has either reached a very high level in chess or he has to be cheating with more than 33% of the moves.==>https://talkchess.com/forum3/viewtopic. ... en#p928964

[swami]

Code: Select all

[Event "9th Sinquefield Cup 2022"]
[Site "Saint Louis USA"]
[Date "2022.09.04"]
[Round "3.1"]
[White "Carlsen, Magnus"]
[Black "Niemann, Hans Moke"]
[Result "0-1"]
[BlackElo "2688"]
[BlackFideId "2093596"]
[BlackTitle "GM"]
[ECO "E20"]
[EventDate "2022.09.02"]
[Opening "Nimzo-Indian"]
[Variation "Romanishin-Kasparov (Steiner) system"]
[WhiteElo "2861"]
[WhiteFideId "1503014"]
[WhiteTitle "GM"]
[Annotator "Stockfish 15 @60s per pos, eval is SPOV"]

1.d4 Nf6 2.c4 e6 3.Nc3 Bb4 4.g3 O-O 5.Bg2 d5 6.a3 Bxc3+ 7.bxc3 dxc4 8.Nf3 c5 9.O-O cxd4 10.Qxd4 Nc6 11.Qxc4 e5 12.Bg5 { 0.00/31 } ( 12.a4 { 0.00/31 } ) 12...h6 { 0.00/31 } ( 12...Be6 { 0.09/31 } ) 13.Rfd1 { -0.24/34 } ( 13.Bxf6 { 0.00/34 } ) 13...Be6 { 0.37/32 } 14.Rxd8 { -0.30/34 } ( 14.Qc5 { -0.28/34 } ) 14...Bxc4 $7 { 0.31/31 } 15.Rxa8 { -0.29/31 } 15...Rxa8 $7 { 0.39/31 } 16.Bxf6 { -0.25/33 } ( 16.Nd2 { -0.19/33 } ) 16...gxf6 { 0.22/35 } 17.Kf1 { -0.30/40 } ( 17.Nd2 { -0.23/34 } ) 17...Rd8 { 0.45/39 } 18.Ke1 { -0.29/34 } 18...Na5 { 0.25/37 } 19.Rd1 { -0.22/36 } ( 19.Bh3 { -0.19/36 } ) 19...Rc8 { 0.28/39 } 20.Nd2 { -0.19/34 } ( 20.Bh3 { -0.15/34 } ) 20...Be6 { 0.28/38 } 21.c4 { -0.45/31 } ( 21.Ne4 { -0.21/32 } ) 21...Bxc4 { 0.23/39 } ( 21...b6 { 0.41/33 } ) 22.Nxc4 { -0.15/36 } ( 22.Ne4 { -0.13/36 } ) 22...Rxc4 { 0.20/29 } 23.Rd8+ { -0.12/35 } 23...Kg7 { 0.18/37 } 24.Bd5 { -0.17/36 } 24...Rc7 { 0.25/34 } 25.Ra8 { -0.39/41 } ( 25.f3 { -0.12/37 } ) 25...a6 { 0.37/37 } 26.Rb8 { -0.30/38 } ( 26.a4 { -0.25/38 } ) 26...f5 { 0.38/35 } 27.Re8 { -0.47/33 } ( 27.f3 { -0.18/33 } ) 27...e4 { 0.61/35 } 28.g4 $2 { -1.09/35 } ( 28.Rd8 { -0.48/33 } ) 28...Rc5 { 1.11/33 } 29.Ba2 { -1.95/30 } ( 29.Rd8 { -1.25/30 } ) 29...Nc4 { 1.69/32 } ( 29...fxg4 { 1.93/32 } ) 30.a4 { -2.25/32 } ( 30.Bxc4 { -1.77/33 } ) 30...Nd6 { 2.25/33 } 31.Re7 $4 { -4.12/32 } ( 31.Rd8 { -2.20/32 } ) 31...fxg4 $5 { 2.14/31 } ( 31...Rc2 { 4.07/31 } ) 32.Rd7 { -1.76/29 } 32...e3 { 1.89/31 } 33.fxe3 { -1.87/28 } 33...Ne4 { 1.84/33 } 34.Kf1 { -2.33/29 } 34...Rc1+ { 0.79/30 } ( 34...Rf5+ { 2.13/31 } ) 35.Kg2 { -1.04/30 } 35...Rc2 { 1.01/32 } 36.Bxf7 { -0.85/29 } 36...Rxe2+ { 0.69/31 } 37.Kg1 { -0.69/30 } 37...Re1+ { 0.94/32 } 38.Kg2 $7 { -0.66/37 } 38...Re2+ { 0.70/35 } 39.Kg1 { -0.69/30 } 39...Kf6 { 0.83/32 } ( 39...Re1+ { 0.94/32 } ) 40.Bd5 $2 { -1.27/34 } ( 40.Rxb7 { -0.51/34 } ) 40...Rd2 { 1.49/35 } 41.Rf7+ { -1.49/27 } 41...Kg6 { 1.50/30 } 42.Rd7 $4 { -5.77/29 } ( 42.Rf4 { -1.35/30 } ) 42...Ng5 { 5.73/30 } 43.Bf7+ { -7.18/27 } ( 43.Kf1 { -5.89/27 } ) 43...Kf5 { 4.65/29 } ( 43...Kf6 { 7.08/30 } ) 44.Rxd2 { -4.79/26 } 44...Nf3+ $7 { 2.66/25 } 45.Kg2 { -5.61/28 } 45...Nxd2 $7 { 5.11/24 } 46.a5 { -5.95/27 } 46...Ke5 { 6.67/32 } 47.Kg3 { -7.62/29 } ( 47.Kf2 { -7.45/29 } ) 47...Nf1+ { 7.76/39 } 48.Kf2 { -10.19/27 } ( 48.Kxg4 { -8.20/28 } ) 48...Nxh2 { 10.39/31 } 49.e4 { -13.52/31 } ( 49.Be8 { -11.54/29 } ) 49...Kxe4 { 13.65/30 } 50.Be6 { -15.80/30 } ( 50.Kg3 { -13.66/25 } ) 50...Kf4 { 15.83/28 } ( 50...h5 { 15.93/29 } ) 51.Bc8 { -21.66/25 } ( 51.Bf7 { -15.94/25 } ) 51...Nf3 { 22.79/25 } ( 51...h5 { 28.50/25 } ) 52.Bxb7 { -55.47/30 } 52...Ne5 { 55.47/33 } ( 52...Nd2 { 55.47/32 } ) 53.Bxa6 { -152.65/33 } 53...Nc6 { 152.65/37 } 54.Bb7 { -319.54/37 } ( 54.Bd3 { -152.65/31 } ) 54...Nxa5 { 319.77/45 } 55.Bd5 { -319.79/50 } 55...h5 { 319.79/48 } 56.Bf7 { -319.81/62 } 56...h4 { 56.94/29 } 57.Bd5 { -319.55/29 } 0-1

Now, I can understand why Magnus did this. Hans showed some accurate plays prior to this game, but this game is on top.

In that famous Carlsen vs Niemann game, ENDGAME started from move 14 onwards (so early) and it continued till the end of the game (move 57)

I am reminded of the interview from GM Jacob Aagaard (about Hans Niemann):

Our training session was a week. It was meant to be a camp, but no one else could make it. Hans was difficult to train. I tried to do calculation and endgame training with him (he had requested endgame training). At first, I showed exercises from recent games (last 18 months) that I really liked. He knew them ALL. I was astonished by his memory. I was astonished by his intuition. Both were off the charts for what I have seen training Shankland, Gelfand, and other 2600+ and a few 2700s.

There were obvious big holes in his chess, but to be honest, I see big holes in the game of Giri, Aronian, Mamedyarov, Firouzja, and other top players. When I get a 2650 student, I usually try to find out what part of their game is at a much lower level. There is always some area of chess where they are just blank. Maybe they cannot really visualise. They don’t know how to make simple decisions. They cannot calculate a line till the end. All three examples of real 2650 players I have worked with.

Hans’s confidence in his own intuition and his surprise when it was wrong was a recurring theme of the week he was here. Another was that whenever I came to his room, he was looking at chess. Playing through ALL games from all tournaments on Follow chess.

I have seen nothing out of the ordinary in the last two days. Hans playing reasonably well against opponents that are not playing that well. His big confidence. His awkwardness in front of the camera. His highly intuitive way of thinking. His lack of accuracy in variations. Him blundering when suggesting things, he thinks he might have looked at.

I also did not see anything out of the ordinary from Carlsen. Entitlement. Lack of responsibility. Lack of accountability. A Norwegian troll army ready to defame a man who only 400 days ago was a minor. Carlsen has acted badly in many situations after losing in the past. In that way, he reminds me of Federer, who was a badly behaved teenager. Become the best player in the world and behaved excellently. Then started losing to Djokovic and needed a period to adjust to reality.

People say that Carlsen does not behave badly when he is losing in his Meltwater Tour to Praggnanandhaa. It is partly because it is like Federer losing a set. It is partly because Praggnanandhaa is deferential to Magnus. Hans is not. Hans wants to kill the king. Wants to take the throne. He has no remorse over this at all.

Endgame may be his forte, and that could be the reason he was playing so accurately in that game with Carlsen, with Queens off the board as early as move 13!

[Uri Blass]

Well ... MC the old "dinosaur" seems to have done well and easily beat Arjun Erigaisi ... one of the top new prodigies in chess and much higher rated than Hans Nieman. Only difference is that Hans Nieman is an admitted serial cheater while Arjun has never been accused of cheating. I really doubt that chess is so simple that studying moves from SF NNUE and LC0 can make you play like those engines. For sure it can add depth to your game and you can use some methods that NNUE engines have shown to be extremely strong ... like pushing and embedding a pawn deep ... but MC seems to be the one that has most gained from that.

You could be right that Hans Nieman is just misunderstood young kid who did mistakes and that he is the strongest human chess player ever. But another simple explanation could be that he is cheating. Time will tell which one it is.

If studying moves from SF NNUE and LC0 can make you play like those engines in some types of positions then you can achieve 100% correlation in some games and not be the strongest chess player because not in every game you get the positions that you know to play.

Hans is certainly not the strongest human chess player based on his results but even if in some games(and not in every game) he can play 100% like an engine it does not prove that he is cheating.

It may be interesting to make a competition between humans in predicting different engine moves without using an engine to see if humans really can learn the style of chess engines.

Take 2 strong engines A and B and some human who can prepare by learning both A and B.
You can use A=Stockfish15 B=Dragon3.1 but you can also use something different.

Take many positions and anlayze the position by both A and B and choose from them 1000 positions when the choice of A is different than the choice of B.

What is the percentage correct that the human may guess for A and B out of 1000 positions with no engine help?

Something like 40% for A and 38% for B but if you replace the choice between A and B only 18% for A and 19% for B suggest that the human is able to learn the engine style.

Note that there may be cases when the best strategy of human may be to choose the same move for both A and B if the target of the human is to get the most correct when you get a point for every correct and you can get at most 1 out of 2 by guessing the same move for both engines(because you know in advance that the moves are different) but you have a risk of getting 0 out of 2 if you guess different moves for both engines.

Another, point that I would like to mention is that in the past we have posted several times that even if you take a top player rated around 2680 like Niemann and let him play 1 move 2 moves or three moves from any top engine during his game versus player of the caliber of Carlsen , GM Ding Liren, or GM Caruana, he would not be able to beat them, since during those games he would eventually make a few inferior moves that Carlsen, or GM Ding Liren would immediately punish and turn the game in their favor. So the only explanation is that GM Hans has either reached a very high level in chess or he has to be cheating with more than 33% of the moves.==>https://talkchess.com/forum3/viewtopic. ... en#p928964

I think that it is possible that a player with rating 2680 beat carlsen in a single game with no cheating.
200 elo difference does not mean that the weaker player has no chance to win in a single game.

There were bigger surprises in chess.

[Chessqueen]

Well ... MC the old "dinosaur" seems to have done well and easily beat Arjun Erigaisi ... one of the top new prodigies in chess and much higher rated than Hans Nieman. Only difference is that Hans Nieman is an admitted serial cheater while Arjun has never been accused of cheating. I really doubt that chess is so simple that studying moves from SF NNUE and LC0 can make you play like those engines. For sure it can add depth to your game and you can use some methods that NNUE engines have shown to be extremely strong ... like pushing and embedding a pawn deep ... but MC seems to be the one that has most gained from that.

You could be right that Hans Nieman is just misunderstood young kid who did mistakes and that he is the strongest human chess player ever. But another simple explanation could be that he is cheating. Time will tell which one it is.

If studying moves from SF NNUE and LC0 can make you play like those engines in some types of positions then you can achieve 100% correlation in some games and not be the strongest chess player because not in every game you get the positions that you know to play.

Hans is certainly not the strongest human chess player based on his results but even if in some games(and not in every game) he can play 100% like an engine it does not prove that he is cheating.

It may be interesting to make a competition between humans in predicting different engine moves without using an engine to see if humans really can learn the style of chess engines.

Take 2 strong engines A and B and some human who can prepare by learning both A and B.
You can use A=Stockfish15 B=Dragon3.1 but you can also use something different.

Take many positions and anlayze the position by both A and B and choose from them 1000 positions when the choice of A is different than the choice of B.

What is the percentage correct that the human may guess for A and B out of 1000 positions with no engine help?

Something like 40% for A and 38% for B but if you replace the choice between A and B only 18% for A and 19% for B suggest that the human is able to learn the engine style.

Note that there may be cases when the best strategy of human may be to choose the same move for both A and B if the target of the human is to get the most correct when you get a point for every correct and you can get at most 1 out of 2 by guessing the same move for both engines(because you know in advance that the moves are different) but you have a risk of getting 0 out of 2 if you guess different moves for both engines.

Another, point that I would like to mention is that in the past we have posted several times that even if you take a top player rated around 2680 like Niemann and let him play 1 move 2 moves or three moves from any top engine during his game versus player of the caliber of Carlsen , GM Ding Liren, or GM Caruana, he would not be able to beat them, since during those games he would eventually make a few inferior moves that Carlsen, or GM Ding Liren would immediately punish and turn the game in their favor. So the only explanation is that GM Hans has either reached a very high level in chess or he has to be cheating with more than 33% of the moves.==>https://talkchess.com/forum3/viewtopic. ... en#p928964

I think that it is possible that a player with rating 2680 beat carlsen in a single game with no cheating.
200 elo difference does not mean that the weaker player has no chance to win in a single game.

There were bigger surprises in chess.

You are correct Mr. Uri ==>


LAXMAN, R.R.. 2378 - SWAYHAM, P DAS. 1902 0-1
1st Bikaner Open International Grandmast 2022.10.01
Shahid
1. d4 1 d5 9 2. ♘f3 5 ♘f6 8 3. c4 4 c6 6 4. e3 5 e6 25 5. b3 6 ♘bd7 11 6. ♗d3 6 ♗d6 54 7. ♗b2 10 b6 91 8. O-O 9 ♗b7 29 9. ♘c3 49 O-O 17 10. ♕e2 171 ♖e8 83 11. ♖fd1 37 ♖c8 148 12. ♖ac1 34 ♕e7 63 13. cxd5 66 ♘xd5 208 14. ♘e4 571 ♗b8 189 15. ♘e5 198 ♘f8 321 16. ♕h5 315 f6 155 17. ♘c4 66 ♘b4 1261 18. ♗b1 118 ♖ed8 382 19. a3 292 ♘a6 171 20. b4 24 ♖d5 415 21. ♕e2 137 ♖cd8 83 22. ♘cd2 32 ♘c7 104 23. ♘f3 231 ♖5d7 295 24. ♖e1 197 ♘d5 225 25. ♕c2 80 ♕f7 118 26. ♘g3 28 ♖c8 92 27. ♗a2 90 ♔h8 178 28. ♕e2 399 ♕e7 50 29. ♘e4 21 ♕f7 54 30. ♗b1 151 ♕e7 22 31. g3 22 ♔g8 77 32. ♘ed2 8 ♕d8 193 33. ♗a2 125 ♔h8 15 34. e4 7 ♘e7 75 35. ♘f1 23 ♘eg6 69 36. h4 9 ♕e8 109 37. ♘e3 80 c5 123 38. d5 exd5 136 39. ♘xd5 41
39. exd5
39... ♘e5 75 40. bxc5 125 ♖xc5 66 41. ♖xc5 2 bxc5 4 42. ♗xe5 45 ♗xe5 10 43. ♕c4 38 ♘e6 371 44. ♘xe5 92 fxe5 4 45. ♖b1 50 h6 87 46. a4 58 ♗c6 163 47. a5 408 ♖d8 8 48. ♕a6 515 ♕d7 46 49. ♘e3? 116
49. ♘e7 ♗xe4
49... ♕xe7 50. ♕xc6 ♘d4 51. ♕b7
50. ♖e1 ♘d4 51. ♖xe4 ♕xe7 52. ♔g2
49... ♘d4−+ 50 50. ♗d5 5 ♕h3 43 51. ♘g2 39 ♗xd5 49 52. ♕d3 29 ♗c4 12 53. ♕e3 5 ♘e2+ 4 0-1

SINGH, PRITAM. 1899 - NGUYEN, DUC HOA. 2360 1-0
1st Bikaner Open International Grandmast 2022.10.01
Shahid
1. d4 1 ♘f6 2. c4 g6 3. ♘c3 d5 4. ♗g5 ♘e4 5. ♘xe4 dxe4 6. e3 f6 7. ♗h4 ♘c6 8. a3 h5 9. h3 e5 10. dxe5 295 ♕xd1+ 32 11. ♖xd1 48 g5 9 12. ♗g3 34 h4 15 13. ♗h2 26 ♘xe5 31 14. ♗xe5 146 fxe5 6 15. ♘e2 178 ♖h6 228 16. ♘c3 56 ♗f5 27 17. ♗e2 140 ♖b6?? 278
17... ♖d8 18. O-O
18. ♘d5 c6
18... ♖xd1 19. ♖xd1 ♖d6
18. ♘d5+− 768 ♗d6 1183
18... ♖c6 19. ♗h5+ ♔d8 20. ♘b6++−
19. c5 247 ♗xc5 18 20. ♘xc7+ 10 ♔f8 30 21. ♘xa8 75 ♖xb2 5 22. ♗g4 91 1-0