Wednesday, February 28, 2018

Season 11 premier division, after RR4

Crosstable after RR4



Stockfish is in the lead, already 4 points from 3rd place and well on its way to the superfinal. Houdini is second, now two points behind Stockfish after only 4/7 points in RR4. Komodo in 3rd is losing sight of the leaders after a disappointing 7 draws in RR4.
At the bottom Ginkgo continues on its sure path to relegation. Fizbo is still putting up a fight with 3.5/7 in the RR including 2 wins, one against 6th placed Andscacs.

A few interesting games

Komodo - Houdini: Komodo had an eval advantage which increased to over 1, but only got a RB vs RB ending with a pawn up, all pawns on the king side. Houdini held the position, Komodo could only get a draw, recall it lost this opening in RR3.

Fire - Ginkgo: Fire kept a small eval advantage for more than 45 moves. Evals started to increase after a series of exchanges reduced to a RB vs RN position, Fire with a pawn up and a queen side passer. Fire created a second passer on the king side and Ginkgo was forced to lose the knight to stop it, leading to a won RB vs R ending.

Andscacs - Fizbo: Andscacs traded a rook for a bishop and pawn early in the game. Fizbo's eval jumped to 1 and gradually increased as the engines exchanged pieces. The game reached a BB vs RN position, for a while the evals did not move. Fizbo then found a way to exchange a knight for a bishop and to get a winning 6-man position. Andscacs realized it was losing 20 moves later. An important win for Fizbo in its fight against relegation.

Stockfish - Ginkgo: There were many early exchanges with only RBN vs RBN left on move 25. Ginkgo had two weak pawns on the queen side and Stockfish's eval was over 1. After capturing a pawn Stockfish traded down to a winning N vs B ending.

Fire - Andscacs: The game reached an unusual material imbalance of RR vs BBN, evals remained low and the engines traded down to a drawn pawns vs B ending.

Fire - Chiron: The game started balanced in a closed position with a small eval advantage for Fire. When Chiron gave a rook for a knight and pawn the evals jumped over 1. After exchanging most pieces the game reached a R vs N ending, and the game was over after Fire created an advanced passer.

Houdini - Fizbo: Fizbo managed to keep the evals low for about 30 moves in a closed position. Houdini's eval went over 1 after it created a central passer, and evals continued to increase as the position opened up. Houdini slowly pushed the pawn forward, keeping it well protected, Fizbo could not stop it. Fizbo was about to lose a piece when the game was adjudicated.

Stockfish - Andscacs: Material was balanced in an open position, yet Stockfish's eval was over 1. Andscacs had a central passer, on the other hand its king was less protected. Stockfish pressed with its major pieces and Andscacs had to block attempts to infiltrate the back ranks. Eventually Andscacs traded down to a RN vs RN position but let go of its passer and Stockfish was a pawn up. Evals were high, the black pawns were all isolated. After Stockfish captured another pawn and created a passer the game was over.

Fizbo - Ginkgo: Fizbo's eval was over 1.5 in what looked like a closed position with both engines shuffling. However, when Fizbo opened the king side Ginkgo had to give a rook for a knight to get its king out of trouble. After exchanging queens only RBB vs BBN were left. Fizbo captured a pawn and reduced to an opposite color bishop ending, but with a pair of connected passers that were enough for a win.

Chiron - Fizbo: When the position opened up Chiron had two rooks controlling an open file on the queen side and a pawn up. Both kings were exposed on the king side, Fizbo had to defend against mating threats. With less pieces on the board the pawn advantage became more significant, the game ended in a QB vs QN position with Chiron about to capture a second pawn.

Stockfish - Komodo: Stockfish had an eval advantage of about 1 from the start. After the queen side opened the engines exchanged pieces and by move 38 only QBN vs QBN remained. Stockfish's eval peaked at 1.5, Komodo built a fortress it couldn't penetrate, more than 50 moves later the game ended in a draw.


Featured game: Chiron - Stockfish
Premier division, round 22
Link to game on TCEC

The game started in a closed position with evals close to 0. Stockfish had a pawn majority on the queen side while Chiron had an advanced central passer.


After the position opened up a little the engines exchanged almost all the pieces until the game reached an opposite color bishops ending with Chiron a pawn up. This is usually a drawish ending but Chiron's bishop was very weak and was blocked by its own pawns.


Stockfish captured the passer and then gave its bishop to create a connected pair of passers that Chiron couldn't stop. Stockfish won both sides of this opening.


Sunday, February 25, 2018

Season 11 premier division, after RR3

Crosstable after RR3



Stockfish and Houdini lead the table. Komodo in 3rd had a weak RR with 3.5/7 points including a loss to Houdini. Houdini is the only engine without loss.
Ginkgo closes the table with no wins. Fizbo is in 7th after a solid 3/7 points in the RR, including an unexpected win against higher placed Chiron. The gap to Andscacs in 6th place increased to 2.5 points.

A few interesting games

Stockfish - Chiron: Stockfish had an eval advantage after the opening, material was equal but Chiron had two weak pawns that were targets for attack. Evals increased over 1.5 after Chiron gave a pawn, trying to activate its pieces. In a QBN vs QBN position Chiron tried to build a fortress. After a series of checks Stockfish captured a second pawn and that was enough for a win. This win was important for Stockfish fans, in season 10 stage 2 Chiron drew its 4 games with Stockfish and that was part of the reason Stockfish did not qualify for the superfinal.

Fire - Fizbo: The game was balanced with a small eval advantage for Fire. Evals jumped over 1 when Fire gave a rook for a bishop and created a passer in the center. In a QBB vs QRN position the passer was too strong. Fizbo tried to counter with a desperate attack on the white king, the game was adjudicated before it lost more material.

Ginkgo - Stockfish: Stockfish took over the game with a pawn push in the center that resulted in a R for BN trade and a black pawn on the 2nd rank. Ginkgo gave a knight for the passer and the game reached a RB vs RR endgame, Ginkgo up 2 pawns. The endgame was difficult but Stockfish had no real problem. It brought the king forward and exchanged a pair of rooks. Ginkgo seemed ahead on material when the game was adjudicated but the white king side pawns were about to be eliminated.

Komodo - Andscacs: Andscacs left its king exposed facing Komodo's rooks in open files. Komodo chased the black king while winning material, eventually it was a rook up when the game was adjudicated.

Stockfish - Fizbo: After exchanging queens Stockfish blocked the king side with its knights, evals rising over 1. Fizbo countered on the queen side as Stockfish was a pawn up in a RRN vs RRB position. Fizbo lost more material and the game was over.

Houdini - Ginkgo: Houdini's eval was close to 1 out of book, Ginkgo's position was completely blocked and Houdini patiently prepared its king side attack. When Houdini was ready evals were already close to winning and the game was adjudicated before the captures began.

Ginkgo - Chiron: Ginkgo had a small eval advantage and was a pawn up. The center was cleared and Ginkgo's king exposed, Chiron's pieces started to target the white king and the eval changed to negative. Ginkgo pushed a passer to the 7th rank but couldn't go further. After Chiron captured the passer and the engines exchanged most pieces the game reached a BN vs BB ending with Chiron a pawn up. The white pieces were passive and the bishop pair was decisive for Chiron. This is the third loss in a row for Ginkgo, two playing white.

Stockfish - Fire: Stockfish started with an eval close to 1. Fire thought it was relatively safe for a while, material remained equal but Stockfish's gradually increased. When Stockfish pushed a central passer to the 7th rank Fire started to feel the pressure. The game reached a QRB vs QRB position, Fire got rid of the white passer and material was still balanced. However, through checks and pins Stockfish captured a pawn and traded down to a winning rook ending.

Fizbo - Chiron: Fizbo started with an eval close to 1 which Chiron didn't believe. As the game progressed and pieces were exchanged it became clear that Chiron had two weak pawns that Fizbo could target, and Chiron had less space for its defending pieces. In a QRB vs QRB position the black pieces were too busy and gradually Fizbo forced Chiron to let go of a pawn. The game was adjudicated before Fizbo won more material. A surprising win for Fizbo against a better opponent.

Andscacs - Ginkgo: Andscacs started with an eval advantage which gradually increased. Eventually Ginkgo lost a piece without real compensation, the game reached a QR vs QB position won for white.

Fire - Houdini: Fire was slightly ahead for the first 30 moves, evals were close to 0 after it gave a rook for a bishop and an advanced passer. Houdini exchanged pieces and captured the passer, evals became slightly negative though Fire still had queen side passers in a QBN vs QRN position. The evals jumped when Houdini found a way to make mate threats, forcing exchanges and reaching a won B vs R ending, despite being 3 pawns down.

Featured game: Houdini - Komodo
Premier division, round 15
Link to game on TCEC

Houdini had an eval advantage after the opening. The engines castled in opposite directions and both kings were slightly exposed, though not directly threatened. Houdini had a bishop pair and a better pawn structure, Komodo had weak c and h pawns.
 

Evals slowly increased and were above 1 when Houdini captured the black h pawn. Komodo didn't try to protect the pawn, instead it countered in the center with a forward rook. Houdini was a pawn up with a king side passer that looked like a potential problem for black.



The engines exchanges bishops and started to shuffle. For a long time evals remained constant, Houdini was reluctant to push its passer while Komodo took over the 2nd rank and was threatening the white king. The evals started to climb again after Houdini exchanged rooks and finally moved the passer forward.



Houdini still had to keep its king safe but after exchanging queens the threat was over and Komodo had to scramble to stop the advancing passer. Komodo stopped the passer only to reach a losing rook endgame. This was Komodo's first loss in the disivion. Was it always winning for white? Could Komodo have defended better?


Tuesday, February 20, 2018

Season 11 premier division, after RR2

Crosstable after RR2



Draw rate for the bookless part of the premier division is 34/56 (60.7%). This is quite low, the reason is that the bottom 2 engines lose many games. If we restrict to games in the top 5 the draw rate goes up to 17/20 (85%).

Stockfish, Komodo and Houdini are in the lead, Stockfish performed better against the weaker engines and it doesn't have a clear lead because of its loss to Komodo in RR1. At the bottom Fizbo and Ginkgo look like sure candidates for relegation. Andscacs is also losing games, but its wins over both Ginkgo and Fizbo give it a safety margin of 2 points above 7th place. Chiron and Fire are closer to the top with one loss each.

A few interesting games

Fire - Ginkgo: Ginkgo won a pawn after the opening, while Fire was ahead on eval. Fire temporarily gave a rook for a bishop defender, allowing it to attack the black king and later regain the material and be a up a pawn. Evals were at 1.5 when the game reached a rook ending with pawns on both flanks. The extra pawn was decisive and the engines' evals quickly jumped as they realized Fire was winning.

Stockfish - Ginkgo: Stockfish had an eval advantage of about 0.8 after the opening, with a bishop pair against Ginkgo's two knights. With less pieces on the board the bishops were very active and the evals increased. Stockfish traded down to a rook ending where all Ginkgo's pawns were isolated, an easy win.

Fizbo - Stockfish: After the opening both engines thought they were better. Stockfish created a pawn majority on the queen side, Fizbo countered by sacrificing a bishop for two pawns and creating an advanced passer on the queen side and another passer in the center. Stockfish won a pawn and took over the center. Fizbo traded a rook for a bishop and was a rook down, its two advanced passers were not enough compensation. The game ended with 3 queens on the board, Stockfish's material advantage was winning.

Ginkgo - Houdini: Evals were close to 0 after the opening. Ginkgo made an error on move 26 and allowed Houdini to put a strong bishop in the center. Houdini slowly gathered its forces on the queen side and then gave a pawn to open a file for the rooks to attack. Ginkgo had to defend its king, Houdini captured two pawns and this was enough for a win in a QRN vs QRB position.

Chiron - Ginkgo: Ginkgo had an small eval advantage for more than 30 moves, after a B for R trade. The game reached a RB vs RR position, and then Chiron's king side pawns started to advance and the evals turned to Chiron's favor. Ginkgo gave back the material but in the rook ending Chiron's passers were better and gave Chiron the win.

Houdini - Fizbo: Fizbo's king remained in the center without the option of castling. Houdini had mating threats on f7 and on the d file, Fizbo's defense collpased and the game was over after 26 moves.

Stockfish - Andscacs: Andscacs gave a rook for a bishop early in the game, Stockfish had an eval advantage of about 1 and increasing. By the time the game reached a R vs B ending Stockfish's eval was winning. Andscacs took longer to realize, without tablebases it continued to play a 6-man lost position for 30 moves until adjudication.

Fizbo - Ginkgo: Fizbo had an eval of 1 for a long time with the engines shuffling in a closed position. Fizbo opened the center, giving a pawn but ruining Ginkgo's pawn structure. Evals started to increase when Fizbo's king came forward, both engines had passers but Fizbo's passer was better protected and more advanced. The game was adjudicated in a RN vs RB position before Ginkgo lost a piece. A first win for Fizbo

Chiron - Fizbo: The game reached a QBN vs QBN position on move 29. Chiron had an eval advantage of over 1 as it went a pawn up, Fizbo tried to counter by sacrificing a bishop for pawns and exposing the white king. Fizbo's QN attacked but had to give Chiron a move after every 3 checks. In cycles of 3 moves Chiron created an advanced passer that Fizbo couldn't stop.

Stockfish - Komodo: Stockfish lost to Komodo in RR1 and the fans were hoping it could beat Komodo in RR2. Stockfish had a small eval advantage, the queen side and center were locked and the engines shuffled for a while. Komodo opened up the center and created a passer while giving Stockfish a pawn majority on the king side. The evals jumped to around 1 but fluctuated a lot as Komodo had possible perpetual check draws and Stockfish tried hard to avoid these. Stockfish gave a piece and tried to push a passer but could only achieve a draw.

Featured game: Komodo - Chiron
Premier division, round 13
Link to game on TCEC

Komodo had an eval advantage after the opening. Chiron chose to give a pawn early and as compensation had more space and its pieces were better developed and active.

Komodo defended patiently while the eval increased, 'a pawn is a pawn' as the saying goes. After a while Komodo's pieces were as strong as Chiron's, both engines battling to control the center.



The pieces slowly came off the board and Komodo was still a pawn up and had better pawn structure. The game reached a BN vs NN ending, with evals over 2.5.

The ending was slow and technical, when the white king moved forward Komodo created a passers in the center and Chiron couldn't stop it without losing material. This was the first loss for Chiron in the premier division.

Season 11 premier division bookless opening summary

The games of first two RRs of the premier division are played without book. Below are a few statistics of the bookless part of the premier division. It is interesting to compare these with the bookless part of division 1.

First move
26/56 (46.4%) start with d4
26/56 46.4%) start with e4
4/56 (7.1%) start with Nf3

By engine:
Andscacs: 1 d4 5 e4 1 Nf3
Chiron: 2 d4 3 e4 2 Nf3
Fire: 7 e4
Fizbo: 4 d4 3 e4
Ginkgo: 5 d4 2 e4
Houdini: 6 d4 1 e4
Komodo: 5 d4 2 e4
Stockfish: 3 d4 3 e4 1 Nf3

Opening by first ECO letter

A - 3 (5.4%), 2 Indian
B - 6 (10.7%), 5 Caro-Kann
C - 20 (35.7%), 16 French variations, only 4 French Exchange.
D - 18 (32.1%), 11 QGD variations
E - 9 (16.1%), 7 Queen's Indian

Decisive games openings

A35 English: Symmetrical, Four Knights
A45 Indian (win in black)
A47 Neo-Queen's Indian (win in black)
B15 Caro-Kann: Tartakower (2 white wins)
C06 French: Tarrasch, Closed  
C07 French: Tarrasch, Open
C11 French: Steinitz, Boleslavsky Variation
C14 French: Classical, Steinitz
C70 Spanish (win in black)
C96 Spanish: Closed, Chigorin, Keres Variation
D02 Queen's Pawn (win in black)
D06 QGD
D24 QGA
D30 QGD
D31 QGD: Janowski Variation
D32 QGD Tarrasch: von Hennig-Schara Gambit
D35 QGD: Exchange 
D37 QGD: Classical (1 white, 1 black)
D44 Semi-Slav: Botvinnik, Lilienthal Variation
E11 Bogo-Indian

Sunday, February 18, 2018

Season 11 premier division, after RR1

Crosstable after RR1



The draw rate is only 57.1% so far. This is much lower than expected for bookless games between strong engines. The difference in strength between engines at the top and those at the bottom is proving to be high.

The top 3 engines Stockfish, Houdini and Komodo are already leading the table. Komodo beat Stockfish in the last game of the RR, quite an upset - again, these are bookless games.
Fizbo is 8th so far, losing 4 games. Ginkgo and Andscacs lost 3 games each, Andscacs had an advantage because it beat both Fizbo and Ginkgo.

In some of the games the engines lower their evals when they realize the game is heading for a draw. As a result the draw rule is applied earlier making the games shorter. This effect appeared in previous seasons in the late stages when strong engines played each other. 

A few interesting games

Fizbo - Andscacs: Andscacs equalized and had a small eval advantage starting from move 20. Both engines had active pieces, Fizbo had a pawn majority on the queen side and a strong knight outpost, while Andscacs pushed pawns on the king side. On a probable Fizbo blunder Andscacs' eval jumped over 2, suddenly Andscacs had a dangerous attack on the white king. Fizbo gave a rook to stop the immediate threat, it still had a few assets but eventually Andscacs traded down to a Q vs QN position and a win. Andscacs is first to draw blood in the premier division, a win in black in a bookless game.

Fire - Fizbo: Evals were low with a small advantage for Fire, they remained stable for many moves. The game reached a same color bishops ending on move 62, Fizbo was a pawn up but had a vulnerable backward pawn on the flank that had to be watched by the king. Fire used this (together with another Fizbo blunder?)  to create a passer on the queen side, trading for the black bishop and winning the game with its final pawn.

Komodo - Ginkgo: Ginkgo sacrificed a pawn early, Komodo had a small eval advantage in a RRN vs RRB position on move 20. The evals slowly increased as Komodo outplayed Ginkgo in the endgame. The knight was stronger than the bishop, and the white king was more active, allowing Komodo to capture a second pawn and eventually win the game.

Houdini - Chiron: Houdini had an eval advantage close to 1, when the position opened up Chiron traded RR for Q which later became a Q vs RN imbalance. Evals peaked over 1.5, Houdini couldn't break Chiron's defense and the engines agreed on a draw after a long game.

Chiron - Fire: a miniature draw after 18 moves. The exact move sequence appeared in Fire's PV of move 6.

Komodo - Andscacs: Andscacs let its bishop be trapped on move 19 and was forced to trade it for two pawns. Komodo's eval jumped over 2 and continued to increase. The game was adjudicated when Andscacs lost another piece. A surprising blunder at this level of play.

Stockfish - Fizbo: Stockfish started with an eval advantage that kept growing. By move 29 it was a pawn up with a pair of connected passers in the center. Eventually Fizbo had to give a rook to get rid of the passers, an easy win for Stockfish.

Houdini - Ginkgo: The game was balanced until Ginkgo sacrificed a pawn on move 20, in an attempt to weaken the white center and open a file for its rook. Houdini's eval immediately started to increase, slowly at first. Houdini had a pawn majority on the king side. After a few exchanges Houdini attacked through the queen side with a rook that Ginkgo wouldn't take for fear of creating a passer on the 6th rank. Houdini's eval was winning and the game was over after several more moves.

Andscacs - Stockfish: Stockfish advanced pawns on the queen side, Andscacs ignored the danger until it was too late. Stockfish gave a pawn to create a queen side passer on the 2nd rank, and Andscacs eventually lost a piece after getting rid of it. The game ended in a R vs RB position, Stockfish winning in black.

Stockfish - Fire: All knights were off early, Stockfish with a small eval advantage. When Stockfish started to push pawns on the king side with many backing pieces the evals jumped over 1.5. Fire traded down to a RRB vs RRB position, the black pawns were weak while white had supported passers. Stockfish started to win material, with quick adjudication. An important win for Stockfish, Fire's first loss.

Houdini - Andscacs: Houdini had an eval advantage that went over 1 when Andscacs traded a rook for a bishop and a pawn. Andscacs thought it was relatively safe as it traded down to a R vs B ending, but this was only because it did not have tablebase support. Houdini knew it was a 6-man win.

Fizbo - Chiron: Fizbo chose to trade R for BN, after exchanging queens Chiron was two pawns up with eval over 2. Chiron's pawns started to advance and the black king came forward, an easy win.

Andscacs - Ginkgo: Andscacs played the opening better, its eval constantly increasing. Andscacs was a pawn up on move 28, and after several exchanges the game reached a RR vs RN position. Andscacs slowly improved its position in the endgame, it took many moves but the win was never in doubt. 


Featured game: Komodo - Stockfish
Premier division, round 7
Link to game on TCEC

The engines played an opening with an early R vs BN imbalance. Evals were low, Komodo had a small eval advantage with a potential king side passer. Stockfish had a bishop pair but the LSB was limited by the black pawns.



The black king was exposed and had to be defended by pieces. Komodo used the open file on the king side to bring the rooks forward. Stockfish managed to free its LSB but at the cost of a ruined pawn structrure. Stockfish's eval indicated it was in trouble.

Komodo created a connected pair of passers in the center and attacked with its rooks. After Komodo won a piece Stockfish made a weak attempt to counter with an attack on the white king.

Komodo defended easily, it traded its passers for the black passer and targeted the remaining black pawns to get the win. It was a surprise for me to see Komodo beating Stockfish in a bookless game, the Stockfish fans must be in shock.

Saturday, February 17, 2018

Season 11 premier division, openings after RR1

The games of first two RRs of the premier division are played without book. Below are a few statistics after RR1 is over. It is interesting to compare these with the bookless part of division 1.

First move
12/28 (42.9%) start with d4
14/28 (50%) start with e4
2/28 (7.1%) start with Nf3

By engine:
Andscacs: 2 e4 1 Nf3
Chiron: 1 d4 1 e4 1 Nf3
Fire: 3 e4
Fizbo: 2 d4 2 e4
Ginkgo: 2 d4 2 e4
Houdini: 3 d4 1 e4
Komodo: 2 d4 1 e4
Stockfish: 2 d4 2 e4

Fire is the only engine to only play one opening move, all other vary their openings. There were two Nf3 first moves, both transposed into a d4 opening.

Opening by first ECO letter

A - 1 (3.6%),
B - 3 (10.7%), 2 Caro-Kann
C - 11 (39.3%), 9 French variations, only 2 French Exchange.
D - 7 (25%), 4 QGD variations
E - 6 (21.4%), 4 Queen's Indian

Decisive games openings

A47 Neo-Queen's Indian (win in black)
B15 Caro-Kann: Tartakower
C07 French: Tarrasch, Open
C11 French: Steinitz, Boleslavsky Variation
C14 French: Classical, Steinitz
C70 Spanish (win in black)
C96 Spanish: Closed, Chigorin, Keres Variation
D02 Queen's Pawn (win in black)
D30 QGD 
D37 QGD: Classical
D44 Semi-Slav: Botvinnik, Lilienthal Variation
E11 Bogo-Indian


Monday, February 12, 2018

Season 11 division 1, final results

Final crosstable after 8RR



The final draw rate is 141/224 (62.9%).

Andscacs was the clear winner of the division. Fizbo finished 2nd but it was a close race that lasted until the last round when Fizbo held a draw against Booot.

Nirvana and Hannibal finished last and will be relegated to division 2 next season. Hannibal had a chance to catch up to Laser in 6th place, but after losing 4 times to Laser it was nearly impossible. Nirvana and Hannibal also crashed twice each in the early rounds of the division, but remained stable after being limited to 16 threads, and avoided disqualification.

The premier division is the next event of season 11, with Stockfish, Houdini, Komodo, Fire, Ginkgo and Chiron, joined by Andscacs and Fizbo. The time control will be 90+10, so we can expect 3 hour games. There will be 6 double RRs, one bookless and 5 with Cato openings. This is going to be a long event with exciting chess by the best engines available. My prediction is that Stockfish (new version) and Houdini will qualify for the superfinal, and that Fizbo will finish last. Andscacs will fight to stay in the premier division.

A few interesting games

Laser - Fizbo: The game reached a RRB vs RBN position on move 20, Fizbo had a pair of connected passers on the queen side, evals close to 0. Fizbo's passers advanced to the 2nd rank before Laser gave a bishop to take them out, Fizbo's eval over 1. The position looked like a draw, but Fizbo outplayed Laser in the endgame. After exchanging a pair of rooks it used its knight to block the white passer, and the king and bishop to target the weak pawns, resulting in a winning position.

Laser - Andscacs: Laser's king remained uncastled and the king side rook was passive in the corner. Andscacs' eval constantly increased as Laser tried to block the black pieces without success. The game was adjudicated a win for Andscacs after Laser lost a bishop and was about to lose more material. Three straight losses for Laser, two in black. Hannibal closed the score gap though it is still behind due to the crash tiebreak.

Booot - Laser: The game reached a bishop ending with same color bishops. Booot was a pawn up and all its pawns were invisible to the bishops, while Laser had one vulnerable pawn. The ending looked tricky to me, the moves did not make sense, but Booot found a way to drive the black king away so it could capture a pawn, enough for a win. Laser's 4th loss and Hannibal is ahead by 0.5 points.

Andscacs - Fizbo: The engines created a R vs BB imbalance early in the game, and by move 36 these were the remaining pieces on the board. Andscacs was 3 pawns up with connected passers on the queen side. When the passers advanced the black bishops were forced to defend, and the pawns majority on the king side was enough for a win. This is Andscacs' first win against Fizbo in the division.

Laser - Hannibal: The game reached a QRN vs QRB position on move 24 with evals close to 0. Gradually Laser improved its position, its knight more effective than the black bishop. Evals increased to over 1.5 when Hannibal traded down to a queen ending, but this was a winning position for Laser. Once again Laser beats Hannibal, it is again ahead by a point.


After RR7 is over Andscacs is 5.5 points ahead of Fizbo in 2nd and 6.5 points ahead of Booot in 3rd. With 7 games left Andscacs is surely going to be the winner of division 1. The race for 2nd place between Fizbo and Booot is still close. It may be decided in the last round when they play each other, Booot in white. At the bottom of the table Nirvana is sure to finish last. Laser had 4 losses in this RR, but two wins and especially a win against Hannibal mean that it is still a point ahead of Hannibal in 7th. This was Laser's 4th win against Hannibal in 7 games without loss, this unusual result is keeping Laser in the division.

Gull - Nirvana: Gull had an eval advantage from the opening, the center was blocked, almost all pieces remained on the board and the engines shuffled for a while. Nirvana advanced pawns on the king side and completed a pawn wall on the king side, and the white pieces could hardly move. Gull realized it was in trouble when Nirvana captured a pawn and created an opening on the queen side. A bishop for pawns sacrifice opened a second hole on the king side, Nirvana exposed the white king and advanced a pair of connected passers. Gull couldn't stop the black pieces on the queen side, the game was adjudicated before it lost more material. An unexpected win for Nirvana and in black.

After 54 rounds Hannibal is 1.5 points behind Laser, and because of the crash tiebreak Hannibal needs 2 wins in the next 2 rounds, together with two Laser losses to avoid relegation. Fizbo is only 0.5 points ahead of Booot, very close and Booot will play white against Fizbo in the last round.

Fizbo - Andscacs: Queens were off the board early in the game, Fizbo was a pawn up with an eval of about 1. Andscacs traded down to a RRN vs RRN position and captured back the pawn, Fizbo's rooks remained active but Andscacs defended well and held the draw. Fizbo and Booot are still 0.5 points apart.

Hannibal - Laser: Hannibal needed a win to have a small chance of surviving the division. It had an eval advantage from the opening but Laser equalized without too much effort. The game ended in a 3-fold repetition on move 30. Hannibal will be relegated and Laser survived.

Booot - Fizbo: Booot had an eval advantage for a while, after Fizbo traded down to a RN vs RN position evals came down to 0 with a drawn result. Fizbo is promoted to play in the premier division, joining Andscacs. 

Featured game: Jonny - Andscacs
Division 1, round 49
Link to game on TCEC

Jonny had an eval advantage of over 1 after the opening, much higher than Andscacs' eval but Jonny is known to exaggerate by a factor of 2.


Here Jonny played b4, forcing Andscacs to move the c5 knight while further exposing its king. The white knights went on the attack and Jonny managed to win a rook for a knight. Despite Jonny's material advantage evals were in favor of Andscacs.


Jonny's king was exposed and under some pressure, and its king side was not fully developed. Andscacs used the pinned c4 pawn and Jonny's reluctance to open files on the queen side to create a connected pair of passers in the center. Evals were over 2 in favor of Andscacs after the engines exchanged queens.


Andscacs was still a trade down but after a few exchanges the game reached a RB vs BB position and the passers were worth at least a rook, giving the win to Andscacs.


Season 11 division 1 statistics

Draw rate, wins



Final draw rate was 62.9%.


Game termination

The three most common game termination causes were:
33.9% - TCEC win rule
32.1% - TCEC draw rule
18.8% - 3-fold repetition

There were 4 crashes in the division, two for Hannibal and two for Nirvana. After they were llimited to 16 threads they became stable and were not disqualified with a third crash. Both were relegated to division 2 in any case.

Moves per game



Median= 62
Average= 69

The game lengths were similar to those in division 2.

Time per game (hours)



Median= 2:05
Average=2:03

Openings

The first double RR was bookless, and a summary of the bookless openings can be found in a previous post. Here I summarize the results of the remaining games, with book sequences by guest Nikolaos Konstantakis. The first letter of the ECO codes was distributed as follows:


If we use the opening 'family name' (using format FAMILY_NAME: VARIANT....) the top 3 are:
Caro-Kann - 16 times
Sicillian - 15 times
English, King's Indian, Nimzo-Indian, Slav - 10 times

The opening variant was mostly determined by the book sequence. 90.5% of the game pairs repeated the same ECO code twice, and 85.7% repeated the same opening variant. In only 2.4% of the game pairs the ECO first letter was not repeated twice.

Reverse pairs, wins

Again I'm restricting to the games with book sequences.



There were 7 game pairs with a 1-1 biased result.

Reverse pairs, same moves 

Here I'm also restricting to the games with book sequences.



Pairs of reverse games diverged very quickly, 45.2% of the pairs diverged immediately after book, 84.5% repeated at most 2 plys.

Wednesday, February 7, 2018

Season 11 division 1, after 6RR

Crosstable after 6RR



The current draw rate is 102/168 (60.1%). There were more draws in RR5-6 compared to RR3-4.

Andscacs extended its lead to 3 points ahead of Fizbo. Booot and Jonny are still in the race, both have lost 0.5 points to Fizbo in RR5-6. Nirvana and Hannibal close the table. Hannibal reduced the score difference to Laser and Gull by 0.5 point and still has a reasonable probability to avoid relegation. It is interesting to note that Hannibal's results are better than Laser's against most opponents, and the reason Laser is ahead is that it beat Hannibal 3 times in their head to head games.

There were no crashes in RR5-6, Nirvana and Hannibal are still 1 crash away from disqualification.

A few interesting games

Laser - Fizbo: Queens were off early and Laser had an advanced passer on the 6th rank. After exchanging all rooks Fizbo's eval started to increase, reaching a peak of over 2. In a BNN vs BNN position the black king was very active as an extra piece. Laser held on to its passer as long as it could but lost it eventually, however by then evals were dropping and there were not enough pawns on the board, Laser held the draw.

Hannibal - Gull: The game reached a RB vs RB position on move 29. Hannibal was up a pawn but evals stayed constant at around 1 for a long while. Somehow Hannibal stumbled on a path to improve, exchanging the rooks and using zugzwang to capture two pawns and win.

Hannibal - Jonny: Jonny pushed pawns on the king side, driving off defenders and threatening the white king. Hannibal exchanged queens to stop the attack, Jonny's eval was over 1 in a RRN vs RRB position. Hannibal still felt safe after Jonny traded a rook for a knight and pawn, but Jonny's pawn majority on the king side was quickly turned into an advanced passer, and a win for Jonny.

Gull - Andscacs: Gull started with a significant eval advantage, and the evals stayed high for over 40 moves, reaching over 1.5. Andscacs sacrificed a pawn and a bishop, drawing the white queen away. Then with its queen, a rook and two advanced pawns Andscacs created a strong enough attack that forced a draw.

Laser - Hannibal: Laser had an opportunity to move further away from the relegation zone. It was a pawn ahead with an eval advantage that peaked over 1.5. Hannibal traded down to a drawn rook ending, it is still fighting to stay in the division.

Jonny - Andscacs: Jonny was up a pawn with evals over 1 and rising. The engines traded most pieces through a queen pin along a major diagonal, reaching a QB vs QB position on move 30. Evals continued to climb and after exchanging queens Jonny's eval was over 4, Andscacs' eval over 2. Jonny had 2 pawns and Andscacs only one, Jonny gave a high score to this 7-man position using 6-man tablebases, and yet Andscacs held the draw. Did Jonny miss the winning path, or was it a draw to begin with?

RR5 is over. There were many draws in RR5, 19/28. This is closer to the draw rate of the bookless openings in RR1+2. As a result there are only a few changes in the crosstable. Andscacs is in the lead, 1 point ahead of Fizbo. Booot and Jonny are still in the race. Hannibal and Nirvana close the table. Nirvana looks a sure bet for relegation. Hannibal still has hopes, closing the gap to Laser by 0.5 points and to Gull by 1 point in RR5. There were no further crashes in RR5, Nirvana and Hannibal are still one strike away from disqualification.

Andscacs - Nirvana: Andscacs advanced pawns on the queen side facing the white king. Nirvana ignored this and attacked the uncastled black king with a QN combination. Nirvana gave a knight for pawns and exposed the black king, but its attack was not effective. Evals were over 2 in favor of Andscacs with the white king close to being mated. Nirvana tried to defend but gradually lost pawns, the extra piece was enough for a win for Andscacs. This is the 2nd double win for Andscacs against Nirvana.

Hannibal - Andscacs: For the first 30 moves the game was balanced with many pieces on the board and evals close to 0. Andscacs' eval climbed over 1 as more pieces were exchanged and Andscacs won a pawn. The game reached a rook endgame, for a while it seemed that the pawn advantage was only good for a draw. Evals jumped when Hannibal allowed the black king to advance. This apparent blunder gave Andscacs the win.

Laser - Booot: Booot equalized and took the initiative. It had a pawn majority on the king side, and Booot converted that to a passer with evals over 1. Then Booot traded down to a B vs N ending, Laser had to stop the passer allowing Booot's king to get to the queen side first for the win.

Jonny - Fizbo: Jonny was up two pawns with a connected pair of passers, its eval over 2.5. However, Fizbo had its own connected pair of passers which was more advanced, and even after being a rook down Fizbo was able to force a draw. A missed opportunity for Jonny to get closer to 2nd place.

Hannibal - Laser: The engines started to shuffle on move 26, both reluctant to advance their passers, Hannibal in the center and Laser on the queen side. Evals were close to 0 for 50 moves, Laser's pawns crept forward slowly. Laser's eval started to increase as it opened the king side, and all of a sudden it had passers on all fronts. After exchanging down to a RB vs RN position Hannibal couldn't handle all the pawn threats. An important win for Laser, after Hannibal almost closed the gap between them.

Featured game: Hannibal - Fizbo 
Division 1, round 37
Link to game on TCEC

Hannibal had an eval advantage from the opening. On move 17 Fizbo gave a pawn on the king side. The white pawn on e5 was hanging but Fizbo preferred to leave it for a few moves. Taking with the queen leads to a skewer attack by the DS bishop, and taking with the knight opens a file with an attack on the king.



However, a few moves later Hannibal was already two pawns ahead and its DS bishop found a different skewer attack, this time on the queen side. Fizbo couldn't take the bishop with the knight since both the rook on d8 and the bishop on f8 were hanging.



Fizbo was forced to lose material. A few moves later after more pieces were exchanged Hannibal was up a piece and a pawn in a winning position.



The game was adjudicated a win for Hannibal not long afterwards. This opening had a 1-1 biased result, but it was surprising how fast Fizbo's position crumbled. This was a good result for Hannibal, it still had hope of staying in the division.





Friday, February 2, 2018

Season 11 division 1, after 4RR

Crosstable after 4RR



The current draw rate is 67/112 (59.8%). The bookless part of the division is over, and in the first two RRs with book there have been only 23/56 (41.1%) draws. There are many more decisive games, a draw is like a win and a win in black (against the book bias) is very significant. There have been 4 openings with a biased 1-1 result (two white wins) in RRs 3-4.

Andscacs is leading the table, 1.5 points ahead of Fizbo in second. Andscacs had a score of 9.5/14 in the last two RRs, including 3 wins in black. Fizbo led after the first 2 RRs, but scored only 7.5/14 in the next two, including 4 losses. Booot (8.5/14) and Jonny (9/14) are not far behind.

Hannibal and Nirvana close the table. Both have crashed in these two RRs, and both are a crash away from disqualification. They have been limited to 16 threads in order to improve their stability, this may have also contributed to their worse scores. 

A few interesting games

Booot - Jonny: Booot had an eval advantage out of the opening but Jonny equalized quickly. Most pawns remained on the board and the engines shuffled for a while, Jonny had more space and a small eval advantage. The engines opened a file on the king side and exchanged all rooks, and several exchanges later only QN vs QB remained. Evals were at 0 when Booot blundered the endgame. It allowed the black king to move forward, creating mating threats and eventually an advanced passed pawn in a queen endgame. A rare black win in the biased openings part of the division.

Andscacs - Booot: Andscacs started with an eval advantage that gradually increased. There were many early exchanges and by move 24 only R vs R remained with Andscacs a pawn up. A few moves later the pawn position was: on the king side 3 against 3 pawns and on the queen side an isolated white passer. Evals increased over 2 as Andscacs pushed its passer, but without support from its king the passer couldn't get past the 7th rank. It took 50 more moves but Booot held the draw. Did Andscacs miss a win?

Laser - Nirvana: Nirvana took the pawn on b2 early in the game, but it turned out to be a poisoned pawn. Laser was better developed, and after a series of exchanges it won a piece and reached a QN vs Q position. The knight was sufficient for a win after the queens were off.

Gull - Andscacs: Gull had an eval advantage after the opening. All pieces were on the board when Gull sacrificed a knight for pawns on move 18 to open the queen side. Gull's pieces remained passive as Andscacs started to press on the king side. Gull realized too late that it was in danger when Andscacs made mating threats, the game was quickly over.

Nirvana - Jonny: Nirvana had an eval advantage after the opening, Jonny equalized without too much trouble. The game seemed to be heading for a draw with evals at 0. Nirvana was a pawn down but had a queen side passer, Jonny's pieces built a fortress on the king side and didn't move. When Nirvana moved the passer forward it started to look dangerous. Jonny sent a knight to defend and all of a sudden the king side collapsed. Jonny lost material and its king was exposed to attack, the game was soon adjudicated a win for Nirvana. Did Jonny blunder? A rare win for Nirvana.

Fizbo - Booot: Fizbo had an eval advantage after the opening, it concentrated its pieces on the king side which had an open file facing the black king. Booot simply marched its king out of danger, while clearing the center and capturing a pawn. After Booot connected its rooks the white king started to look exposed, evals jumped over 2 in favor of Booot. Fizbo could not stop the black pieces from getting to the back ranks, and then Booot had too many threats on the king and the defending pieces, the game was adjudicated.

After RR 3 Fizbo and Andscacs are in the lead with a margin of 1.5 points between 2nd and 3rd place in the table. Nirvana is 2.5 points below the rest, it had a very poor RR with only 1.5/7 points. The engines in places 3-7 are all within 0.5 points. Hannibal and Nirvana crashed again, both are one crash away from disqualification, both are limited to 16 threads for better stability.

Nirvana - Andscacs: Nirvana had an eval advantage after the opening, with more space while all pawns were on the board. After a few exchanges Andscacs equalized, and with a queen side passer it was a little better. Andscacs offered a pawn, and when Nirvana took it the evals jumped over 1. Nirvana traded a bishop and pawns for a rook but now the black queen and rook moved behind the white pawn line and the white king was exposed to attacks in a QB vs QR position. The endgame was not straightforward but eventually the game reached a queen ending won for Andscacs. Andscacs won both sides of this opening.

Laser - Jonny: Laser started with an eval advantage, after many exchanges Jonny equalized and only RRN vs RRN were left by move 25. Laser had a weak c pawn which forced its rooks to passively protect it, and Jonny took the initiative. After exchanging a pair of rooks Jonny captured a pawn and traded down to a won knight ending. A double win for Jonny in this opening.

Hannibal - Andscacs: Andscacs took over the game with a pawn storm on the king side, despite its castled king. After opening a file and exchanging all rooks only QN vs QB remained. After a file was opened on the queen side black's queen crossed Hannibal's pawn line and captured a pawn. Evals increased rapidly and the game was soon over. Another win in black for Andscacs.

Nirvana - Laser: The game reached a QRB vs QRN position on move 30, evals remained close to 0. Nirvana then traded a rook for a knight and pawns, evals stayed 0 even after it had to give the bishop to stop a passer. It seemed that Nirvana's king side passers were strong enough to counter Laser's extra rook, but Nirvana blundered the endgame and gave the win to Laser.

Gull - Booot: Gull had an eval advantage after the opening but Booot had more space on the king side. Gull's defending pieces on the king side could hardly move and Booot took advantage of this and won a rook for a bishop with a skewer. Gull continued to struggle and lose material until the game was adjudicated.

Hannibal - Laser: Evals were close to 0 in a closed position with almost all pieces on the board. After opening the queen side both engines had a passer there, Laser's passer had better support. Laser captured the white passer, advanced its passer to the 2nd rank and traded down to a winning rook ending. Laser won both sides of this opening, and together with two wins against Nirvana it escapes the relegation zone.


Featured game:  Andscacs - Fizbo
Division 1, round 20
Link to game on TCEC

For a while Andscacs thought it had an advantage, while Fizbo disagreed. Fizbo had an open file on the king side facing the white king.


Andscacs tried to counter on the queen side, in a series of exchanges it traded a bishop for pawns to clear the black queen side. Fizbo ignored this and put all its heavy pieces on the g file, its evals jumping over 2.


Andscacs saw the danger too late, and after a second series of exchanges the evals were over 3 in a RB vs Q position.



The queen was too strong against the exposed white king and scattered pawns, a win for Fizbo. First loss for Andscacs in the division, the win in black gives Fizbo the lead.