Season 24 superfinal statistics

A summary statistics table of previous stages and seasons. 

Draw rate, wins

Final draw rate was 64%.

Game termination

The three most common game termination causes were:

55% - TCEC draw rule
19% - SyzygyTB
18% - Mate

There were no crashes in the stage.

Moves per game

Median= 72
Average= 77.5

There were 12 games longer than 100 moves, the longest was 275 moves (Stockfish - Leela, game 68, draw).

Time per game (hours) 

Median= 4:17
Average= 4:07

Openings

There were variable length book openings in this stage chosen by Jeroen Noomen and GM Matthew Sadler. The first letter of the ECO codes was distributed as follows:

The engines had almost no freedom to choose the opening variant, all the game pairs repeated the same ECO and the same opening variant twice.

Reverse pairs, wins 

Reverse pairs, same moves 

Pairs of reverse games diverged slower than usual, 18% diverged immediately out of book, 58% of the pairs diverged at most after 1 move. The longest repeated sequence of moves was 25 plys, in two game pairs (Leela - Stockfish games 67 and 68, Sicilian Velimirovic attack, two draws ; games 83 and 84, Dutch 2.Bg5 variation, two white wins).

Season 24 superfinal games 91-100, season summary

Stockfish is the winner of the season 24 superfinal. The final score is 20-16 with 64 draws. In the final 10 games Leela won 3 times and each time Stockfish won the reverse. The other 2 game pairs ended in two draws, as a result the gap between Stockfish and Leela did not change. 

This season the superfinal was exciting and close, with 4 lead changes. The final gap of 4 wins was lower than all superfinals since season 14. Leela won the first game pair, but in the next 8 game pairs Stockfish won 3 times and led by 2 wins. In game pairs 10-21 Leela managed to win 3 times, again it was leading by a win. Stockfish won twice in game pairs 22-23 and at the half way point Stockfish led by a win. The second half started with many drawn pairs, but Stockfish managed to win a few and extend its lead to 5 wins on game pair 43. Leela won game pair 44, after 24 straight pairs without a win, reducing the lead to 4 wins. After that all game pairs were drawn, Leela won 3 games but each time Stockfish won the reverse. 

The most memorable game for me was game 46. The engines played a queen ending for 60 moves, Stockfish was a pawn up and its eval increased from 1.7 to ~200 (predicted tablebase win) and Leela's eval was still under 1. I haven't seen such an eval discrepancy in a superfinal in a very long time.

In the bookmaker competition of which lines led to a decisive game pair the final score was a draw 7-7. Interestingly, In the drawn game pairs category GM Matthew Sadler had more double draws (score 16-9) and Jeroen Noomen had more double wins (score 2-9).

In season 24 the qualification leagues was replaced by the Swiss event (compulsory format change?). There were 42 engines in the Swiss event, 22 were already assigned to leagues based on their placement in season 23. The other 20 engines were qualification candidates, 5 of which were new to TCEC. The 4 candidates with the best performance in the Swiss event qualified to the regular league 2. As expected there was a lot of luck involved, two of the four qualifiers were tied with 6 more engines. Also, based on the game results I'm not sure the strongest candidates were chosen. As usual, TCEC tried to entertain and not to determine engine strength, so I'm not complaining. Viewers were much more interested in Leela's performance in the Swiss event. An unexpected result of its contempt configuration was that it lost two games in the first rounds against much weaker engines. Its opponents in the next rounds were stronger and Leela came back to its senses, in the end finishing in 3rd place.

Before league 2 started it was announced that the Igel developer has decided to withdraw and give its spot to Booot, which was ready to play after missing two TCEC seasons (both engines are developed by Ukrainians). In the first few rounds Booot crashed twice, then TCEC announced that Booot's results would be ignored and the Igel will replace it after all. When the league was over Igel played against all the engines, using the book openings of Booot's games. In total Booot crashed 4 times in 44 games, so I'm not sure it was necessary to replace it. Igel qualified easily, finishing first with a huge margin. The newcomer Uralochka almost succeeded to promote a second time, it was close but didn't make it in the end. 

In league 1 Koivisto won convincingly with a margin of 3 points. The race for second place was close with 5 candidates within one point in the final rounds, in the end Berserk qualified. Igel finished in joint 3rd place, the other qualifiers from league 2 were relegated back.

In the premier division the top3 engines (Stockfish, Leela and KomodoDragon) repeated their dominance of the last few seasons. Both Stockfish and Leela beat KomodoDragon, their head to head score was a draw with one game pair win each. Stockfish won the league due to better results against KomodoDragon and the remaining engines. Surprisingly Berserk performed much better than Koivisto, despite the fact that in league 1 it appeared Koivisto was much stronger. Berserk finished 4th tied with Ethereal, while Koivisto was relegated back to league 1. This season there was a playoff played with the 4 engines in places 1-4 in the premier division. There was no change in the standings after the playoff, Stockfish and Leela qualified for the superfinal. Unlike the premier division, in the playoff Stockfish beat Leela 3 times without loss.

This season the superfinal coincided with the human world chess championship between Ian Nepomniachtchi and Ding Liren (I don't really think this was a coincidence). The match was full of decisive games and blunders under pressure. Nepomniachtchi led 3 times and each time Ding managed to equalize. In the end Ding won in the last game of the rapid tiebreak games. Compared to TCEC in human games the excitement is all about human properties: opening preparation, surprise, missing a crucial winning move, playing under time pressure. Engines can lose games but they rarely blunder the way humans do.

As for things outside of computer chess: Covid is almost out of the news. In my season 23 summary I was concerned about the winter - nothing happened, and also about China. Covid did hit China, though there were very few official reports. With travel restrictions gone there probably was a sizeable portion of the population infected, and I assume many many fatalities in vulnerable parts of the population (elderly, unvaccinated, immune deficient). However, here we are a few months later, China is still going strong and no one is talking about Covid. Me and my family were also infected for the first time, it took a few weeks for everything to get back to normal. I hope we don't have to go back to this subject in the next seasons. 

Messi and Argentina won the soccer (football) world cup in Qatar. Seems so long ago now. The war in Ukraine is still being fought with no end in sight after over a year. It is estimated that there were tens of thousands dead soldiers on both sides just in the last few months. Why? What for? This needs to stop. In more local news (for me) there has been a lot of civil unrest in Israel. The political deadlock is finally over with a stable elected government, but this government concentrated its efforts in the first few months on a "judicial reform". The opponents of this reform claim it is an attempt to change the balance between government and the justice system, so that it would be much harder for the supreme court to say a law is illegal, and that the government would have more power in choosing judges. After weeks of protests the government agreed to put this reform on hold for the moment, but they may change their minds any time. Add to that the fact that Israel was hit twice by rocket strikes in the last few weeks, once from Lebanon and once from Gaza, and you understand my mood is not great.

But, as Monty Python sang, always look on the bright side of life. So I'm hoping for a better world next TCEC season.

Go TCEC!!

================================

Games 91-92 started with a 26-ply sideline of the KID Orthodox variation. All pieces and pawns were still on the board and there was a diagonal pawn wall in the center. In game 91 there was one pawn exchange on the queen side that opened the c file, and one pawn exchange on the king side. The engines played behind their pawn lines, Stockfish created a passer on the king side, Leela's eval remained just under 1. Stockfish pushed pawns on the queen side and opened the b file, the first piece was exchanged on move 32. All rooks moved to the queen side, evals came down after a pair of rooks was exchanged. Leela reduced to a QRN vs QRN position and went a pawn up. Evals were close to 0, the game reached a queens ending and the game was adjudicated. In game 92 evals dropped down early. There were a few pawn exchanges on the queen side, files stayed closed but Stockfish created connected passers and Leela also had a passer there. Leela captured one of the white passers, the first piece was exchanged on move 32 (again). In a series of exchanges the engines reduced to a QBN vs QRB position, evals were 0 despite the advanced passers, after exchanging minors the game was adjudicated.

Games 93-94 started with a rare sideline of the Gruenfeld, exchange variation, played in high level human chess. In game 93 Leela didn't castle its king, it had strong centl pawns while Stockfish had a backwards pawn on e7 that restricted the black pieces. In a long PV agreement the engines exchanged a few minor pieces in the center, evals stayed stable around 1. Leela pushed the h pawn and moved pieces to the king side, evals started to increase. On move 30 the queens were exchanged, the black pieces were passive, mainly protecting weak pawns.

Leela slowly increased the pressure, the white king moved forward on the king side. Leela exchanged bishop for the black knight, the most active black piece. Stockfish moved its bishop and allowed the white king to grab two pawns. In a RRB vs RRB position Stockfish managed to activate pieces, but too late to save the game.

Leela pushed pawns on both sides of the board, it traded a rook for a bishop and created an advanced passer supported by its bishop. Stockfish couldn't stop the white pawns on the king side, Leela queened twice and mated.

In game 94 there were no exchanges after the start, both engines castled short and the engines played behind their pawn lines. Evals increased very slowly, Stockfish had more space to move its pieces but the engines seemed to be shuffling. On move 33 the first pawns were exchanged on the queen side. 

The engines continued to shuffle, in fact Stockfish was preparing for an attack and waited for the right moment. On move 52 it started with exchanges of minor pieces in the center, removing both the black knights. The engines opened the queen side and all rooks moved to the open files there. Leela blocked Stockfish's attempts to attack from the queen side, it gave a pawn on the king side to give its queen some more room to move. On move 69 the engines exchanged a pair of rooks. 

Stockfish gave a pawn and its e pawn became a passer. After exchanging queens Stockfish pushed the passer to the 7th rank, Leela blocked it but started to lose material. Stockfish underpromoted to a rook, expecting Leela to trade it for a bishop. Leela refused to take but the material was enough for a quick tablebase win. There were two white wins in this game pair, Stockfish leads 18-14.

Games 95-96 started with a rare sideline of the Giuoco Piano, played in human chess but not recently. In game 95 there was one minor piece exchange after the start, Leela pushed the black pieces back and most of the time the most advanced black piece was the queen. On move 17 there was a pawn exchange on the king side, then in a long PV agreement the engines exchanged pieces, Leela went a pawn up and the game reached a RBN vs RBN position on move 29. 

Stockfish had two isolated pawns on the king side. After stabilizing the queen side Leela move its rook to the h file. Stockfish exchanged its f pawn, creating a pawn majority for Leela on the king side. The white king moved forward and attacked the black h pawn. Stockfish managed to capture the g pawn but lost its bishop. 

After a pawn trade the white d pawn became a passer, Leela pushed it forward and the white king moved forward to assist. Stockfish couldn't stop the passer, it exchanged pieces and in a N vs pawns position Leela queened, game over. 

In game 96 the engines repeated the reverse for 10 plies, but later they transposed back and the games really diverged after 21 plies. There were a few pawn exchanges on the queen side and Stockfish created a pawn majority there. Stockfish's eval started to increase slowly, it pushed pawns on both sides while Leela waited. On move 31 Stockfish's eval was over 2, Leela's eval remained mostly stable above 1 though its PV agreed with Stockfish's and saw what was coming.

Evals increased quickly as the engines played out the PV agreement, exchanges opened files in the center and Stockfish moved a rook to the 7th rank. The game reached a QRB vs QRB position with white a pawn up, both engines had a passer in the center. Stockfish pushed its passer to the 7th rank and reduced to a rook ending. 

Stockfish saw a tablebase win in its PV, the engines captured the central passers and traded pawns on the queen side, leaving a white passer. Stockfish pushed its passer, the black king was too far to help and Leela lost its rook, game over. There were two white wins in this game pair, Stockfish leads 19-15.

Games 97-98 started with the QGD Botvinnik variation, played at the highest level of human chess, black was a pawn up at the start. In game 97 the engines followed human games for more than 20 plies. Leela gave a knight for two pawns and opened the king side, then captured a pinned knight. Leela pushed a passer on the king side, Stockfish created connected passers on the queen side. Evals came down as all passers were blocked, Stockfish captured a pawn to equalize material. The kings moved forward, Stockfish chased the white king for a while. A series of exchanges reduced to an opposite color bishops ending and the game was adjudicated. In game 98 the engines chose a different line from human chess. Exchanges opened the king side and Stockfish regained the pawn. Both kings walked to the king side, evals came down rapidly and the engines shuffled for a while. After move 29 the engines exchanged pieces, the game was adjudicated in a rook ending. 

Games 99-100 started with an 18-ply sideline of the Bogo-Indian defense, Wade-Smyslov variation not found in human chess. In game 99 Leela pushed pawns on the king side and castled long. Stockfish pushed pawns on the queen side and went a pawn up, yet evals increased steadily. Stockfish captured two more pawns, this gave the white major pieces more open files to use. Leela captured a rook for a knight using a fork, then captured two pawns on the queen side. On move 31 the engines exchanged queens and the game reached a RRB vs RBN position. 

Leela had a pawn majority in the queen side, it moved its pieces there. Stockfish avoided exchanging rooks, Leela moved a rook to the 7th rank and captured the h pawn. Then it captured the a pawn, creating connected passers. Leela slowly pushed the passers forward and Stockfish started to lose material. Leela queened twice and mated.

In game 100 Leela gave a pawn on the queen side and Stockfish created a protected passer on c6 before castling long. Stockfish gave a rook for a bishop and pushed pawns on the king side. Leela couldn't get rid of the white passer, king safety became another issue after Stockfish opened the g file. 

Stockfish maneuvered its knights so it could attack the black knight on c5, Leela's eval jumped over 2 after a few moves. Leela captured a pawn and the a pawn became a passer, Stockfish hid its king behind the passer. On move 34 the engines exchanged knights and Leela had to block the white passer with a rook. Stockfish added a threat on the black king, forcing a queen exchange. Stockfish captured a pawn and connected a second passer facing the black rooks.

The rooks couldn't stop the white passers with the black king so far away. Leela tried to delay exchanges to avoid a tablebase position, eventually it lost both its rooks for the white passers, game over. There were two white wins in this game pair, Stockfish leads 20-16.

Season 24 superfinal games 81-90

After 90 games Stockfish leads 17-13 with 60 draws. Stockfish increased the lead to 5 wins, and then Leela won a game pair and reduced the lead back to 4 wins. The Leela game pair win came after 24 game pairs with 6 game pair wins for Stockfish and none for Leela. There are only 5 game pairs left to play, it is safe to assume Stockfish will win the match, but Leela may yet reduce the gap .

Games 81-82 started with a rare sideline of the KID Saemisch, Yates defense. In game 81 there were a few pawn exchanges after the start, the engines opened the a file and evals came down. The engines exchanged a pair of rooks, then Leela moved a rook to the 7th rank and the engines cleared the b file pawns as well. The exchanges gradually continued, evals were low and the game was adjudicated in a RB vs RN position. In game 82 the engines played a slightly different line, they delayed exchanging pawns a little but again the a file was opened. Leela moved a rook forward but Stockfish didn't try to exchange rooks. Nevertheless evals came down while the engines exchanged minor pieces. Stockfish pushed pawns on the king side, Leela gave a pawn and both engines created a passer in the center. After exchanging queens evals were close to 0, Leela gave a rook for a bishop and captured all but one white pawn. The game was adjudicated in a RR vs RB position.

Games 83-84 started with a rare sideline of the Dutch, Hopton attack. In game 83 a pawn exchange opened the center, Leela castled long, and queens were off early. Evals stayed stable around 1, Stockfish castled long and Leela went a pawn up with a passer on the king side. On move 29 the game reached a RBN vs RBN position, the kings moved to the center. Both engines pushed pawns on the     queen side, Leela captured another pawn and had doubled pawns on the queen side. Stockfish's eval started to fluctuate, on move 49 the engines exchanged bishops and evals started to increase.

All the white pawns were isolated, Leela had to be careful and patient. Stockfish captured the d pawn and created a passer. Stockfish pushed its pawns forward and they became targets. Leela protected its h pawn, but managed to capture all the black pawns, keeping two white pawns. 

Progress was still slow, Leela pushed both pawns until they were on the 6th rank. Stockfish blocked one pawn but eventually lost its knight for the second, the game ended in a tablebase win.

In game 84 the engines followed the reverse game for 25 plies, with Stockfish castling long and the queens off. Leela castled long, the engines seemed to be shuffling but Stockfish's eval slowly increased. After move 29 Leela's eval increased as well, in a series of exchanges the engines reduced to a RBB vs RNN position with white a pawn up and doubled passers on the king side. There were a few pawn exchanges, Leela regained the pawn, Stockfish created a passer on the queen side and Leela had doubled passers in the center. 

Stockfish pushed the g pawn so it was protected by its rook, then turned to attack the doubled black pawns. Leela tried to protect both pawns with a knight, Stockfish traded bishop for knight and captured both pawns. Leela couldn't block both the white passers, Stockfish pushed the c pawn forward, queened and mated. There were two white wins in this game pair, Stockfish leads 16-12.

Games 85-86 started with a rare sideline of the Sicilian main line, with 5.. Qc7 (instead of the Najdorf 5.. a6). In game 85 after a few development moves there was a series of exchanges including the queens. Evals came down, Leela castled long and Stockfish kept its king uncastled. The game reached a RRN vs RRN position, not a lot happened until the engines reduced to a rook ending and the game was adjudicated.

In game 86 most pieces stayed on the board after the start. There was a hole in the pawn support on the king side and Leela chose to castle long. Stockfish chose to abandon its queen side pawns and castled short. Leela captured a pawn with its queen, evals started to increase. The white pieces were mostly on the king side, but the main issue was the safety of the black king on the other side. The white knights dominated the center, Stockfish moved its rooks to the half open b file, the black queen was stuck in a defensive role in front of the pawns. By move 34 Stockfish was ready for attack.

In a long PV agreement Stockfish sacrificed a knight and opened the a file, then it gave a rook for a knight and a bishop. Leela used a rook to hide its king, its eval drifted down while Stockfish's eval increased. In a series of exchanges the engines reduced to a Q vs RR ending with white two pawns up. 

Stockfish gave the c pawn and pushed the f pawn forward to the 7th rank. Leela finally realized its was losing, it couldn't take the f pawn because it would lose a rook in a check fork. Stockfish created a second passer and pushed it forward supported by the king. Leela wouldn't give material to stop the queening, the game quickly ended with two white queens. Stockfish wins the game pair, it leads 17-12.

Games 87-88 started with a rare sideline of the Hungarian (Benko) symmetrical variation. In game 87 Leela's eval increased from the start, it pushed the h pawn forward until it reached h6. In a very long PV agreement the engines opened the king side, they exchanged minor pieces and queens, Leela's eval drifted back. Leela pushed pawns on the king side, Stockfish captured a pawn on the queen side but evals slowly increased again. Leela focused on the king side, it placed a knight on e6 and Stockfish felt the pressure.

Leela exchanged a pair of pawns, creating a pawn majority on the king side, then it placed its knight on f6. Stockfish had no counter, when it was ready Leela attacked the black king and captured the h7 pawn. The engines reduced to a double rook ending, Leela with advanced connected passers on the king side. The black king had to run to get out of the way. 

Stockfish delayed with checks, but eventually Leela queened the h pawn. Stockfish lost material fast, Leela queened the g pawn as well and mated.

In game 88 Stockfish again pushed the h pawn, but after two long thinks and a 2-fold repetition it decided to exchange it and to open the h file. Stockfish's eval dropped and a few moves later so did Leela's eval. Stockfish walked with its king to the king side, the engines shuffled and exchanged a pair of knights, Stockfish couldn't find a way to attack the black king. For a few moves Leela's eval jumped over 1, it wasn't happy with its PV. Then after thinking for 8 minutes Leela changed its move to Stockfish's PV and its eval dropped back down. In a long PV agreement the engines reduced to a QR vs QBB, Stockfish gave checks and the engines traded pawns. Evals were close to 0, after Stockfish gave a rook for a bishop the game was adjudicated. Leela wins the game pair, Stockfish leads 17-13.

Games 89-90 started with a 20-ply rare sideline of the Ruy Lopez, Chigorin defense, with all pieces and pawns still on the board. In game 89 evals came down from the start, the engines exchanged most pawns on the queen side and a few minor pieces. Stockfish had a passer on the queen side, it managed to get as far as the 3rd rank in a QRB vs QRB position. The engines started to shuffle with evals close to 0, the game ended in a 3-fold repetition. In game 90 the engines locked pawns on the queen side at first, then in a long PV agreement cleared most of the pawns on the queen side. Evals came slowly down, the first piece was exchanged on move 26, in another long PV agreement the engines reduced to a Q vs RB position. Stockfish was reluctant to lower its eval, the game was adjudicated after 30 more moves.

Season 24 superfinal games 71-80

After 80 games Stockfish leads 15-11 with 54 draws. Stockfish increased its lead after winning two game pairs. This is the biggest lead of the match so far, and with 10 game pairs more to play a gap of 4 wins seems decisive. The last Leela game pair win was in games 37-38, so 21 consecutive game pairs without a win for Leela.

Games 71-72 started in the KGA Mason-Keres gambit, with a bias for black. In game 71 Stockfish attacked the black king and let Leela capture a rook with a knight. Leela's eval started very low but jumped over 1 after the rook sacrifice. The white knight was trapped in the corner but Stockfish was focused on the attack. Leela exchanged queens, Stockfish gave a knight but its remaining pieces all targeted the black king. Leela gave a rook, the position finally stabilized and Stockfish captured the trapped knight. The game reached a RB vs RBN position, Leela was two pawns up with a passer on the 7th rank. Evals came down, Leela captured another pawn and placed a second passer on the 7th rank. Both passers could not advance, after Stockfish made sure all its pieces were secure it captured both passers, and Leela created a passer on the king side. The engines started to shuffle on move 41 and evals came down. Pawn moves extended the shuffle until finally Stockfish gave the knight for a pawn and the game was adjudicated, on move 188. In game 72 Leela prevented the white knight fork that would capture its rook. Stockfish's evals immediately dropped, the attack on the white king was over quickly when the engines exchanged queens. Both evals were low, the game was adjudicated on move 36 after enough pieces were exchanged.

Games 73-74 started with the Benoni defense, a line played at the highest level of human chess. In game 73 there were no exchanges after the start. Starting from move 16 the engines played out a long PV agreement, they exchanged minor pieces and opened files on the queen side. Evals came down, Stockfish had a passer on the queen side and Leela created a passer in the center. In another PV agreement Stockfish gave a bishop, both passers were captured and the queens were exchanged. Leela gave a rook for a knight and a pawn, the game reached a RBN vs RR position with one pawn left for each engine. The engines started to shuffle on move 43, evals were close to 0 and the game was adjudicated on move 93 after the engines exchanged a pair of rooks. In game 74 the engines followed theory for a while, again there were no exchanges after the start. After move 19 the engines opened the queen side and Stockfish went a pawn up, though evals came down. Leela regained the pawn and created a passer on the queen side. Then in a long PV agreement there was a series of exchanges that reduced to a same color bishops ending, evals were low enough for the draw rule.

Games 75-76 started with a rare sideline of the Gruenfeld defense, Zaitsev gambit. In game 75 Leela went a pawn up and opened the queen side after the start. The engines continued to develop pieces without exchanges, both queens move forward. After move 22 the engines started to exchange pieces and evals came down. The game reached a RN vs RB position, the pawns stabilized and the game was adjudicated when Leela's eval was low enough.

In game 76 Stockfish let Leela regain the pawn, its eval increased as it pushed pawns in the center. In a series of exchanges Stockfish created a passer in the center and captured a rook for a bishop. The white king was exposed and it walked to the king side. Leela's eval drifted down in a PV agreement, it captured a pawn and created a passer on the queen side. Stockfish moved its pieces to the king side, it expected Leela to capture another pawn on the queen side, but Leela had other ideas. Almost all black pieces were on the queen side, Leela felt safe and Stockfish followed Leela's PV. Stockfish's eval jumped quickly, on move 26 the action started with a rook sacrifice, still in PV agreement.

The white queen moved forward and attacked the exposed black king. Leela's eval jumped many moves too late, the black pieces were too far away and not coordinated. Stockfish gave a rook but captured the black queen, reducing to a Q vs RB position on move 39.

The endgame was long but Stockfish never had a doubt about the outcome. It captured two pawns on the queen side, then slowly pushed the a pawn forward. Leela managed to capture the passer but lost its rook. Stockfish then mated using only dark squares, invisible to the black bishop. Stockfish wins the game pair, it leads 13-10.

Games 77-78 started with a rare sideline of the Sicilian four knights variation. In game 77 there was a long PV agreement early, the center opened and the queens were exchanged, the engines traded rook and pawn for two knights. Evals started to come down and the engines mostly shuffled. The game ended in a 3-fold repetition which seemed to surprise Leela, it expected to play on.

In game 78 there was a different early PV agreement, there were no exchanges after the start, all pieces remained on the board and Stockfish castled long. Leela kept its king uncastled. Stockfish's eval increased slowly while Leela's eval drifted down. After move 20 there were a few exchanges and the king side opened. Leela moved its king without castling and on move 25 Leela's eval jumped, it had tripled pawns on the f file and the two adjacent files were open.

In a long PV agreement Stockfish gave a pawn on the queen side and moved its queen forward on the king side to join the attack. Leela attacked first, it gave a bishop for two pawns and cleared the white queen side pawns. The exposed white king ran forward to the center, Leela pushed a passer to the 2nd rank but Stockfish had the back rank covered with two rooks. The white queen moved forward, after exchanging queens the game reached a RRB vs RR position with black 3 pawns up.

Leela chased the white king but was not strong enough. Stockfish pushed the h passer forward, after exchanging a pair of rooks Leela captured the white passer but lost its advanced a passer. Stockfish protected its last pawn and slowly captured black pawns. The white king moved forward and soon the black king was in a mating net. The game ended in a tablebase win just before mate. Stockfish wins the game pair, it leads 14-10.

Games 79-80 started with a rare sideline of the Caro-Kann, two knights attack. In game 79 Leela moved its queen forward early, it captured a pawn on the queen side. Stockfish pushed pawns across the board, it opened the c file and exchanged queens. Evals started to increase, on move 25 the game reached a RBB vs RBN position with white a pawn up. 

Leela used its bishops to protect the center pawns and to keep the black rook away. Leela captured the black a pawn and created doubled passers. Stockfish gave up its bishop and captured the white b pawns, the game reached a RB vs R position. 

Leela gave the d pawn, Stockfish had weak pawns to defend and it needed to keep the white king back. Stockfish gave a pawn to weaken the white pawn chain, Leela used its bishop to protect its pawns. Leela pushed the h pawn forward and Stockfish started to lose material. After exchanging rooks Leela queened and mated.

In game 80 the engines repeated the reverse for 17 plies, Stockfish captured the pawn on the queen side with its queen. Leela drove the queen back and then opened the c file. Stockfish's eval started to increase as it opened the g file, a few moves later Leela's eval reacted after the game reached a RBB vs RBN position.

The engines exchanged bishop for knight, then cleared pawns on the queen side, Stockfish was two pawns up with a passer. Leela pushed the h pawn and tried to give its bishop to create a passer. Stockfish saw the trap, it let Leela capture the h pawn and pushed the d pawn until it was a passer too. Stockfish was first to promote a pawn, Leela lost its bishop, it captured the white a pawn but lost its h pawn. With patience Stockfish converted the RB vs R ending. There were two white wins in this game pair, Stockfish leads 15-11.

Season 24 superfinal games 61-70

After 70 games Stockfish leads 12-10 with 48 draws. Stockfish extended its lead to 2 wins, this is the biggest lead we've had in this match so far. Stockfish also had a 2 win lead after game 16, before Leela won a game pair of games 19-20. The decisive game pair came after 8 consecutive drawn game pairs, 2 of which were double white wins. This is the first decisive game pair in the second half of the match, there were 9 in the first half.

Games 61-62 started with a 20-ply rare sideline in the KID Orthodox, Korchnoi attack. All pieces and pawns were still on the board with the center blocked. In game 61 there were no exchanges after the start, the engines stabilized the queen side. Starting from move 23 there were a few exchanges, a pair of pawns on the king side, on the queen side a minor piece exchange and another pair of pawns. This opened the a file and the engines exchanged a pair of rooks. Leela created a passer in the center, then the engines mostly shuffled for a while. Evals came down and the engines reduced to a QBB vs QNN position. The engines shuffled until the queens were exchanged and the game was adjudicated. In game 62 again there were no exchanges after the start, after move 20 the engines exchanged minor pieces and opened the b file. Leela was a pawn up with a passer but evals came down. After exchanging a pair of knights Leela gave up its passer and moved its queen and a rook forward on the queen side. Stockfish countered by moving its queen forward too, after exchanging a pair of rooks Stockfish started to give checks. There could have been a quick draw but Stockfish kept delaying with another move. It even gave a rook, but eventually the game ended in a check repetition. 

Games 63-64 started in the French Winawer advance variation, played at the highest level of human chess. In game 63 the engines started to play a long PV agreement on move 9, the white king moved without castling and Leela created a pawn majority on the queen side. Leela's eval increased over 1 while Stockfish's eval came down. On move 21 Leela thought for 8 minutes and its eval dropped, despite the fact that Stockfish also moved its king and this trapped a black rook in the corner. There were more minor piece exchanges, Stockfish gave a pawn and created a passer on the queen side. Leela traded a rook for a knight and pawn, also creating a passer on the queen side. In a series of exchanges the engines reduced to a QN vs QR position, Leela was two pawns up but all passers were gone. Leela was not able to find a way to improve, its eval slowly came down and after 45 moves the game was adjudicated.

In game 64 Stockfish pushed pawns on both sides and kept its king uncastled. Leela castled short and Stockfish traded minor pieces and went a pawn up, opening a hole in the black king's pawn support. The engines exchanged a pair of rooks, Stockfish's eval increased slowly while Leela's eval was stable. Leela captured a pawn on the queen side and expected Stockfish to recapture. Stockfish wanted to insert a 2-fold repetition first, Leela decided not to go along and to capture another pawn. This created connected passers for Leela on the queen side, but also allowed Stockfish to capture a pawn and create a passer on the king side, with its queen facing the black king. Stockfish's eval jumped over 2. 

Stockfish kept playing 2-fold repetition checks with its queen, slowing the game progress. It gave a knight and captured the black e pawn, creating another passer. Then it captured the black knight, reducing to a QRN vs QRB position. Leela's eval started to increase, its passers looked intimidating but it was unable to push them forward. Stockfish captured one passer, the other reached the 2nd rank. Then Stockfish captured another pawn and the black bishop, while Leela connected a second passer on the queen side again. 

Stockfish was a piece up but still had to deal with the black passers. After exchanging queens Stockfish started to see a tablebase win in its PV. Leela captured the white advanced passers, eventually Stockfish gave its knight for the two black passers. The game ended in a winning rook ending with white two pawns up. Stockfish wins the game pair, it leads 12-10.

Games 65-66 started in the Sicilian Kalashnikov variation, played at the highest level of human chess. In game 65 the engines followed theory for a while, they opened the c file and Leela had a pawn majority on the queen side. Leela placed a knight on b6 but Stockfish exchanged it. Stockfish created a passer in the center after exchanging a pair of rooks, evals started to come down. The black queen moved forward, then the engines played out a long PV in which the center pawns were cleared and the game reached a QB vs QB position with white a pawn up. Stockfish gave checks and avoided exchanging queens, it took Leela 60 moves to lower its eval for the draw rule. In game 66 the engines diverged from the reverse game and then transposed back, the real divergence was after 26 plies. Leela was more patient and did not attack the center immediately. After some shuffling Leela opened the e file and Stockfish's eval came down. Stockfish created a passer on the queen side and moved a rook to the 7th rank, Leela avoided exchanges and the engines mainly shuffled with evals close to 0. There was a series of exchanges in a long PV agreement and the engines reduced to a QRB vs QRB position. Stockfish pushed its passer to the 7th rank, Leela threatened mate and Stockfish ended the game in a check repetition.

Games 67-68 started with a 20-ply in a rare line diverging from the Sicilian Velimirovic attack. The engines castled in opposite directions and all pieces were on the board. In game 67 there was a long PV agreement from the start, the engines exchanged minor pieces and pushed pawns across the board without exchanges. Both engines moved rooks to the center, then all rooks were exchanged and Leela went a pawn up. After exchanging queens the game reached a BN vs BB position, the pawns on the king side stabilized and the engines started to shuffle. Leela's eval came down slowly, the game was adjudicated after 45 moves. Game 68 repeated the reverse for 25 plies. Stockfish pushed the h pawn forward and the engines started to shuffle on move 27. The game continued in cycles of shuffles and then a pawn move or a capture that resets the counter. Evals came down very slowly but even when they were low enough there were too many pieces for the draw rule. The game ended in a 50-move draw on move 275.

Games 69-70 started in the Queen's pawn game, Stonewall attack, played in human chess and with a bias for black. In game 69 there were no exchanges after the start, except for a pair of bishops. The first pawn exchange was on move 20, (negative) evals came down. Not a lot happened in the game, in a series of exchanges starting on move 34 the engines reduced to a QN vs QB position. Evals were close to 0 and the game was adjudicated. In game 70 the engines played a different line, again a pair of bishops was exchanged and there were no pawn exchanges after the start. Both queens move forward to the queen side and were exchanged early. Stockfish's (negative) eval came down, the engines opened the g file and Leela kept its king in the center. Leela cleared the white pawns on the queen side, it had two passers there while Stockfish had a passer in the center. The game reached a RRN vs RRN position, Stockfish captured one black passer, then the engines mainly shuffled for a while. Leela's eval came down, after a pair of rooks was exchanged the game was adjudicated. 

Season 24 superfinal games 51-60

After 60 games Stockfish leads 11-10 with 39 draws. There were 4 consecutive double draws followed by a double win game pair, no change in Stockfish's lead. The tension is rising as the end of the match approaches, only 20 more game pairs to play and the engines are only 1 point apart.

Games 51-52 started with a 26-ply rare sideline of the Ruy Lopez, Chigorin defense variation. All pawns and pieces were on the board at the start. In game 51 Stockfish moved a knight forward to f4 and Leela exchanged it. The engines opened the b file and exchanged a pair of rooks, evals started to come down. The engines mainly shuffled for a while, then opened the a file. After some more shuffling the engines exchanged the remaining rooks and the white queen moved forward on the queen side. The engines traded pawns and Leela created a central passer, it got to the 6th rank and was blocked. Evals were close to 0, after exchanging queens he game was adjudicated. In game 52 there were no exchanges after the start, on move 26 the engines exchanged a pair of pawns and opened the b file. Leela doubled rooks on the open file, the engines shuffled for a while and Stockfish prepared a pawn push on the king side. On move 50 Stockfish started with the h pawn, driving a black bishop away to the queen side where it captured a pawn. The g pawn followed and Stockfish opened the king side, however evals came down. There were many piece exchanges and the black king ran to the center. The game reached a QRN vs QRN position with black a pawn up. Stockfish chased the black king to the queen side, the engines exchanged queens and Leela gave a knight to queen a passer. Evals were close to 0 and the game was adjudicated, PV showed Stockfish reducing to a tablebase draw.

Games 53-54 started with a rare sideline of the Dutch classical variation. In game 53 early exchanges opened files on the queen side. The exchanges gradually continued while evals came down, the game was adjudicated in a drawn rook ending on move 42. In game 54 the engines opened the c file, there were no piece exchanges until Stockfish gave a rook for a knight on move 18. Stockfish opened the king side and exchanged queens, its eval came down faster than Leela's. Stockfish chased the exposed black king with its minor pieces, it had a strong bishop pair in the center. Eventually Leela gave a rook for a bishop and got rid of the white bishop pair, Leela gave a knight and went 3 pawns up. Evals were close to 0 and the game was adjudicated.

Games 55-56 started with a rare sideline of the QGA Smyslov variation. In game 55 there were no pawn exchanges after the start, the engines exchanged most of the minor pieces and stabilized their pawn lines. Starting from move 29 the engines shuffled and evals came down. On move 73 Stockfish gave a bishop for two pawns and a rook for a bishop, it attacked the white king and captured more pawns. The game was adjudicated in a QR vs Q with black 6 pawns up. In game 56 the engines exchanged minor pieces and a pair of pawns that opened the d file. After exchanging queens the engines reduced to a RB vs RB position and evals came down. The game was adjudicated when Leela lowered its eval for the draw rule.

Games 57-58 started with a 22-ply line in the English, Nimzo-English variation not seen in human chess, where black is a pawn up after taking the white g pawn. In game 57 Leela castled long and developed a rook through the g file. In a long PV agreement Leela regained the pawn, the engines had long pawn lines across the center and queen side, with a black knight filling a hole. Leela's eval increased over 1 while it considered an attack on the king side, but the engines mainly shuffled for a long time. On move 93 the engines exchanged a pair of bishops and resumed shuffling. Leela's eval slowly came down, on move 121 Stockfish moved a pawn and shut down any attempt to open the queen side. On move 171 Leela gave a pawn in the center, the position opened up with exchanges and Stockfish ended the game in repetitive check. In game 58 Stockfish kept its king in the center, it pushed the c pawn forward and opened the queen side. The engines exchanged queens and Stockfish went a pawn up and created an advanced passer. Leela gave a rook for two minors and the game reached a RR vs RBN position. Evals were close to 0, after exchanging a pair of rooks the game was adjudicated.

Games 59-60 started with a rare sideline of the Pirc defense. In game 59 the engines castled in opposite directions and there were a few exchanges in the center. Leela pushed pawns on the king side, in a long PV agreement the king side was opened, the black king was exposed and the game reached a QRR vs QRR position by move 25. 

Stockfish had a weaker pawn structure and its king was not as safe as Leela's king. After a while Leela went a pawn up and evals started to increase. Leela moved its king forward, sheltered by two pawns, and pushed the f pawn forward. The engines opened the center and Stockfish attacked the white king with its queen, Leela exchanged queens and reduced to a double rook ending. 

Stockfish pushed the h pawn forward but had to abandon it when the white king advanced to support the f pawn. Leela pushed the f pawn to the 7th rank, then gave a pawn and created a second passer on the queen side. Stockfish could not stop both passers, it lost material and the game ended in a tablebase win. 

In game 60 Stockfish locked the center before castling long. A minor piece exchange forced the black king to move without castling and Stockfish's eval started to increase. The black king walked to then king side, Stockfish pushed its king side pawns until they locked the black pawns. Leela pushed pawns on the queen side, one reached a3.

Stockfish walked its king to the center, away from any danger. Leela tried to protect its weak a and c pawns while threatening the white b and e pawns. After some shuffling Stockfish exchanged a pair of rooks and traded the its b pawn for the black a pawn, creating a passer on the queen side. The white queen moved forward and captured two more black pawns on the queen side. 

Leela managed to capture the d pawn, but after a pawn exchange the white e pawn became a passer. Stockfish reduced to a knight ending, Leela couldn't prevent a queening and was mated. There were two white wins in this game pair, Stockfish leads 11-10.

Season 24 superfinal games 41-50

After 50 games Stockfish leads 10-9 with 31 draws. Stockfish is back in the lead after two game pair wins. There were two game pairs with two draws and one with two white wins. At the half way point the match can still be won by both engines, a tiebreak is also a possibility.

Stockfish's win in game 46 came after a very long queen ending, with Leela's eval below 1 for 30 moves, even when Stockfish already saw a tablebase win in its PV. I haven't seen such a discrepancy of evals for a very long time, and it is even more surprising in a superfinal. 

Games 41-42 started with an 18-ply long rare sideline in the KID Fianchetto variation. In game 41 there were many early exchanges played in PV agreement after the start. The game reached a QRR vs QRR position with white a pawn up on move 25. Leela had a pawn majority on the queen side, the king side pawns stabilized and the engines exchanged a pair of rooks. Evals came slowly down, Leela created a passer and pushed it forward. Stockfish countered by exposing the white king and reducing to a drawn queen ending. In game 42 there were no exchanges after the start except one pawn exchange on the king side. The black queen moved forward on the king side, starting from move 19 the engines played out a long PV agreement, exchanging pawns and minor pieces in the center. Evals came down with most pieces still on the board. The engines gradually exchanged pieces and the game was adjudicated in a RN vs RN position.

Games 43-44 started with a rare sideline of the French Advance variation, rarely seen in recent human games. In game 43 the engines exchanged a pair of bishops after the start, Leela pushed the h pawn and moved its king without castling. The engines traded minor pieces on the king side, Stockfish kept its king in the center uncastled as well. Leela went a pawn up and exchanged queens, evals started to come down. The engines exchanged pieces until reaching a RN vs RN position and the game was adjudicated. 

In game 44 Stockfish pushed pawns on the king side, again the engines exchanged a pair of bishops but there were no other exchanges after the start. Leela pushed passers on the queen side and then castled long, a pawn exchange opened the a file. The engines formed a diagonal pawn wall blocking the center and queen side except the open a file, Stockfish castled short. Pawn exchanges opened the king side, each engine created a passer. All the rooks were exchanged through the open files, evals remained well below 1 but Stockfish's eval started to increase after move 39. Leela's eval stayed under 1 for 4 moves while Stockfish's eval jumped over 2, on move 48 Stockish captured the black passer.

The queen side was empty of white pieces, perhaps Leela thought it would have an effective counter but now its eval jumped as well. Leela sacrificed a knight for a pawn and the b pawn became a passer on the 2nd rank. Stockfish rushed back with its queen to block, then slowly sent its other pieces back to assist. The white king became exposed and Leela delayed with checks. Eventually Stockfish captured the passer on move 72. 

Now Stockfish was a knight and pawn up with no compensation for Leela. Stockfish gave the passer to exchange queens, the game reached a BN vs N position. It took another 15 moves before the game ended in a tablebase win. Stockfish wins the game pair, the score is 8-8.

Games 45-46 started with a rare sideline of the closed Sicilian defense. In game 45 the engines exchanged pawns and minor pieces after the start, the white central pawns were cleared and the black pawn structure was weaker than the white. Both engines castled short, on move 19 Stockfish moved its queen forward on the queen side and attacked the a pawn. Leela abandoned its a pawn and pushed the h pawn forward. Stockish's eval came down as it took the a pawn, it felt confident its queen can retreat in time to help on the king side. Leela pushed the h pawn forward, at first its eval increased but then came back down as the engines followed a long PV agreement. Leela opened the king side and exposed the black king, the black queen covered the 7th rank and the white rooks didn't find a way to join the attack. The game reached a RRN vs RRB position, black was a pawn up with a weak pawn structure but its pieces kept the pawns protected. Evals came down, Stockfish gave a rook for a knight and pawn and the game was adjudicated.

In game 46 the engines repeated the reverse for 10 plies, then diverged but transposed back so the real divergence was after 31 plies. Leela had the queen forward on the queen side, it decided not to take the white a pawn immediately, instead Leela attacked the center. In a very long PV agreement the engines cleared the center pawns, Stockfish's eval increased very slowly. Then in a series of exchanges in yet another PV agreement the engines reduced to a queen ending with white a pawn up.

The ending was very difficult to analyze. For the viewers and me the real story was that Stockfish was confident it could win, its eval kept increasing with each move, while Leela felt safe and its eval even drifted down slowly. This went on for almost 30 (!!!) moves, with endless checks from both engines. Leela of course avoided a queen exchange, Stockfish's plan was to walk with its king to the queen side and attack the black a pawn. By move 66 the white king made it to the queen side, still on the back rank. Stockfish already saw a tablebase win according to its eval, Leela's eval was still under 1 (!!!!!).

The white king moved up the board and Leela's eval finally started to react. Stockfish did not stop to capture the a pawn, the white king continued to move towards the black king, pausing on move 79 to move a pawn and reset the 50 move counter. Stockfish pushed the f pawn forward to the 7th rank, the white king move in front of the pawn to avoid more checks, and finally Stockfish queened on move 97, with mate following. Incredible game !! Stockfish wins the game pair, it leads 9-8.

Games 47-48 started with a sideline of the Kan Sicilian, Marocszy bind, Reti variation, played in high level human chess. In game 47 the engines exchanged a pair of knights after the start and the white queen moved forward. Most pieces were concentrated in the center, after some shuffling the engines opened the c file and exchanged a pair of rooks. After some more shuffling there was a series of exchanges that resulted in a queen ending. Leela was up a pawn but evals were close to 0 and the game was adjudicated. In game 48 the engines exchanged pawns and minor pieces after the start, both engines had a passer on the queen side and both queens move forward. Evals came down as Stockfish captured the black passer and exchanged queens. Leela blocked the white passer, the engines shuffled for a while and then reduced to a RBB vs BRN position. Evals were close to 0, Leela traded the rook for two bishops and the game was adjudicated.

Games 49-50 started with a sideline of the Czech Benoni defense, played in high level human chess. In game 49 the center was blocked after the start and Leela attacked it from the king side. There was a long PV agreement that started with a series of exchanges, the engines opened the f file and both engines created a passer in the center. Leela temporarily gave a rook and attacked the black king, Stockfish gave the rook back and stopped the attack. After stablizing the queen side pawns the engines started to shuffle, evals were stable around 1. Stockfish's eval increased after Leela pushed the h pawn, it jumped over 2 after Leela pushed the g pawn as well and moved its h pawn to h6, trapping a black bishop in the corner. The engines exchanged a pair of rooks and both kings ran to the center, only then Leela's eval increased as well - after move 70 !! In a series of exchanges the engines reduced to a same color bishops ending.

Leela gave the g pawn and the h pawn became an advanced passer. Leela used its bishop to keep the black bishop in the corner, Stockfish had 3 connected passers but no way to support them. The black king had to block the white passer and also to block the white bishop from getting to the vulnerable queen side pawns. Stockfish had to make a choice and it chose to abandon the queen side and capture the central passer. There was a pawn race, both engines queened a pawn. Stockfish was first but Leela queened with check and captured the black bishop in the corner. Leela was on time to stop another black pawn promotion, after Stockfish ran out of checks Leela exchanged queens and won with its remaining passers.

In game 50 there was one pawn exchange on the king side after the start, on move 20 Leela initiated an attack on the blocked center from the king side and Stockfish thought this was an inaccuracy. Evals started to increase in a PV agreement, the engines opened the g file, Stockfish exchanged queens and went a pawn up. Then in another PV agreement there was a series of exchanges and the game reached a RN vs BNN position.

Stockfish moved its rook forward behind the black pawns on the queen side and moved its king and knight forward on the king side. Leela couldn't keep all its pawns and pieces safe and Stockfish captured the black f pawn. After another pawn trade the engines reduced to a R vs NN position, Stockfish with 3 passers. 

Leela captured two white passers with its knights, Stockfish captured the remaining black pawns with its rook. The white a pawn was unstoppable, Stockfish queened and mated. There were two white wins in this game pair, Stockfish leads 10-9.

Season 24 superfinal games 31-40

After 40 games Leela leads 8-7 with 25 draws. Leela won 4/5 game in white, in three of these game pairs Stockfish managed to win as white but in one it only drew. As a result Leela is in the lead again. Of the last 12 game pairs Leela won 3 times and Stockfish not once. This superfinal is more exciting than any we have seen in the last seasons, usually the winner (Stockfish) was clear by this stage of the match. 

Leela has been playing a lot better than in the premier division. I'm not sure why that is, the chat mentioned a few configuration changes. I noticed a few things that are new to me. First there are a lot less instamove blunders. These caused many losses for Leela in past games, where it continues to play its PV with only a few seconds thinking time, only to find the move was a blunder in the next move. Another thing I noticed is that when thinking for a long time, even while following its PV, Leela sometimes changes its mind about the move to play. This usually is combined with an eval jump while still thinking. There were a few cases I saw where this was a deciding move that surprised Stockfish and changed the eval trend of the game. Lastly, many Leela wins come after a long period of stable evals around 1. Sometimes Stockfish's eval jumps first, as if it cannot wait any longer to see what Leela would do, but it knows this is a mistake. Whatever it is, Leela obviously has improved a lot and is a strong opponent to Stockfish, which it hasn't been for quite a while.

Games 31-32 started with a sideline of the Scandinavian defense, played at the highest level of human chess. In game 31 the engines started with a PV agreement in which a pair of minor pieces was exchanged and the engines castled in opposite directions. A pawn exchange opened the d file and created pawn majorities for each engine on the side it castled. Stockfish pushed pawns on the queen side, Leela's eval jumped on move 24. Stockfish's eval increased slowly as more pieces were exchanged, the game reached a QRN vs QRN position with white a pawn up.

Stockfish avoided exchanging queens, it exchanged knights instead. Stockfish didn't try to defend the b pawn and attacked the king side. Leela defended its king side pawns, it captured a pawn in the center and reduced to a rook ending. Stockfish regained one pawn but Leela created connected passers on the queen side.

Stockfish couldn't stop the passers, Leela queened and then reduced to a winning king and pawns tablebase win.

In game 32 Leela immediately attacked the king side with its queen, it gave a bishop for a rook. Stockfish castled long, then targeted the black queen and developed its pieces faster. On move 24 Leela thought for 11 minutes and its eval jumped. Stockfish thought Leela's move was not accurate and its eval jumped as well. 

Leela captured a pawn and opened the d file, then castled long as well. In a long PV agreement Stockfish captured a knight as well as pushed the black king to the corner, then in a series of exchanges the engines reduced to a BN vs R ending. 

Stockfish pushed pawns on the queen side while keeping its king side pawns protected. Leela's eval increased very slowly, Stockfish saw it was winning many moves ahead. Eventually Stockfish exchanged pawns on the queen side and created a passer. Leela blocked the passer on the 7th rank with its king, thereby trapping its king in the corner. With mate threatened Leela lost its rook and the game. There were two white wins in this game pair, the score is 5-5.

Games 33-34 started with a line diverging from the KID anti Gruenfeld, not seen on human chess. White started a pawn up. In game 33 the engines exchanged pawns and minor pieces on the queen side, opening the a file. Stockfish avoided exchanging rooks, it doubled its rooks on the b file and attacked a white pawn. Evals came down, the center opened and Stockfish moved its queen forward to attack the black king. The engines exchanged a pair of rooks, Stockfish traded pawns to capture the white b pawn. In a series of exchanges the engines reduced to a queen ending, Leela was a pawn up but evals were low and the game was adjudicated. In game 34 there were less exchanges after the start, Stockfish had a passer on the a file. Stockfish pushed the passer to the 6th rank and doubled its rooks behind it. Exchanges opened the center, then Stockfish gave a rook for a knight and pawn on the queen side. In a series of exchanges the engines reduced to a BB vs RB position. Stockfish was two pawns up with connected passers on the queen side but evals were low and the game was adjudicated.

Games 35-36 started with a rare sideline of the French Winawer opening, Alekhine-Marocsy gambit. In game 35 the black king moved without castling. The center was blocked and there were no pawn exchanges until move 19. The engines shuffled for a while with stable evals, Leela doubled rooks on the b file. On move 26 Stockfish thought for 17 minutes and it changed its move from what it expected in its PV, evals started to increase. In a long PV agreement the engines opened the queen side and reduced to a QRB vs QRN position with white a pawn up.

The engines completed playing the PV agreement. Leela tried to attack the black a pawn, Stockfish blocked the diagonal line of attack with its rook. Leela pushed a pawn on the king side and opened the h file, then threatened to shift the rook there and capture the black knight. Stockfish gave a rook for a bishop to prevent this, now Leela could capture the a pawn as well. 

Leela pushed its passer forward, it gave some material to queen the passer, game over.

In game 36 the engines played out PV agreements after the start, the moves played were not the same as the reverse game but again the black king moved without castling and the center was blocked. A pawn exchange opened the c file, there were only a few exchanges after the start. Stockfish also didn't castle its king, it developed the king side rook through the h file and walked its king to the corner. On move 23 Stockfish's eval started to increase slowly, reacting to what it considered inaccuracies by Leela. On move 30 Leela's eval also jumped, it didn't expect Stockfish's move.

There was only one pawn exchange on the king side, the files remained closed. Stockfish managed to get a protected bishop to g5. The engines cleared all the pawns on the queen side, Stockfish moved both its rooks forward. A series of exchanges reduced to a QRB vs QRB position.

Material was equal but Stockfish's pieces were better placed and its king was safe, while Leela's king was exposed to attacks. Stockfish moved its rook to the 7th rank, its queen was active across the board. Leela gave a pawn and exchanged queens, Stockfish had two passers facing the black king. Leela managed to capture one passer, but then lost material and was mated. There were two white wins in this game pair, the score is 6-6.

Games 37-38 started with a line that diverged out of human games from the Italian game. In game 37 there was a pawn exchange after the start, then the engines started playing out a long PV agreement. Stockfish changed a move in the middle, the game continued in a different PV agreement with a few minor piece exchanges. Evals were stable around 1, in another long PV agreement the engines opened the center and exchanged queens, then Leela went a pawn up. On move 38 Stockfish thought for almost 10 minutes and its eval jumped, Leela reduced to a RRN vs RRN position.

Leela had a pawn majority on the queen side, after exchanging a pair of rooks it created a passer on the b file. Leela pushed the passer slowly and carefully forward, it took 25 moves to reach the 7th rank. After queening it took Leela another 15 moves to end the game in a tablebase win. 

In game 38 the engines repeated the reverse game for 7 plies, after the games diverged evals started to come down. The engines opened the center and exchanged minor pieces, then there was a period of mostly shuffling. A series of exchanges cleared all the queen side pawns, evals were close to 0 when queens were exchanged. The game reached a RB vs RN position and was adjudicated. Leela wins the game pair, it leads 7-6.

Games 39-40 started with a rare sideline of the Queen's Indian, Petrosian variation. In game 39 the engines opened the queen side with pawn and minor piece exchanges, almost all pieces on the board moved there. Evals were under 1 and stable, most moves were in PV agreement. On move 35 Leela changed its mind and made a move not in its PV, its eval started to increase and so did Stockfish's. In a series of exchanges the engines reduced to a RRB vs RRB position.

Evals continued to increase as the engines traded pawns. Starting from move 47 the engines played out a very long PV agreement, Stockfish gave a rook for a bishop and the engines were down to two pawns each. 

The endgame was very long. Leela slowly walked with its king to the queen side, on move 84 Leela  managed to capture the black c7 pawn. Stockfish lost both its pieces to stop a queening, game over.

In game 40 Stockfish chose to avoid many exchanges after the start, the engines only opened the b file and exchanged a pair of rooks. Evals were over 1 and slowly increasing, Leela tried to hold a long pawn line. For a while it was unclear whether Stockfish would attack on the queen side where the black a pawn was weak, or attack the black king. On move 27 Leela accelerated the decision when it pushed the g pawn and created contact with the white pawns, Leela's eval also jumped as a result.

Stockfish opened the king side, in a long PV agreement the engines exchanged minor pieces in the center and the black king walked to the center. Stockfish went a pawn up and created a passer on the king side. Stockfish moved its pieces to the king side, after some shuffling the game reached a QRN vs QRN position.

Stockfish exchanged the major pieces and reduced to a knight ending. It pushed the passer and Leela let it promote to a queen, mate followed. There were two white wins in this game pair, Leela leads 8-7.