Draw rate, wins
Final draw rate was 59.8%.
Game termination
The three most common game termination causes were:
34.8% - TCEC win rule
27.7% - TCEC draw rule
16.1% - SyzygyTB
16.1% - 3-Fold repetition
Ginkgo crashed 4 times in the division and was disqualified, Fritz crashed once. One Leela game was stopped due to a network failure and was later replayed.
Moves per game
Median= 63.5
Average= 71.0
There were 16 games longer than 100 moves, the longest was 308 moves (Leela - Fritz, round 28, Leela won, longest win in TCEC history).
Time per game (hours)
Median= 2:06
Average= 2:03
Openings
There were 6-move book openings in this stage chosen by Cato. The first letter of the ECO codes was distributed as follows:
The 6-move book determined the opening in most cases, 82.1% of the game pairs repeated the same ECO code twice, 76.8% repeated the same opening variant. All of the game pairs repeated the ECO first letter twice.
Reverse pairs, wins
Reverse pairs, same moves
Pairs of reverse games diverged very quickly, 41.1% diverged immediately out of book, 82.1% of the pairs diverged at most after 1 move. The longest repeated sequence of moves was 15 plys (Leela - Jonny, rounds 1 and 8, Two knights defense, two draws)
Possible minor correction: game length = 71.4m. The 'schedule log' overcounts by one ply if game ends with btm but is ok if game ends with wtm.
ReplyDeleteI'm not sure what the cause is here. I'm not using the schedule log, the data is the ply count in the pgn file (divided by 2) and I think it is accurate. It's not a typo, that was the calculated average. There may be a problem with the Leela game that was replayed, I fixed the pgn manually, hopefully without error.
Delete