Monday, October 24, 2016

Season 9, rapid tournament stats

Draw rate, wins

Final draw rate was 27.3%. Putting all engines in one group resulted in huge differences in engine strength in many games, most of which were decisive. Looking at the subtable of only the top 10 engines the draw rate was 63.3%. The top 3 engines played 6 games, all draws.
There was a small preference for white wins. This may have to do with the openings, or perhaps it is just the general bias for white in chess.

Game termination

The three most common game termination causes were:
66.2% - TCEC win rule
12.6% - TCEC draw rule
9.0% - TB position

There were 37 cases of time loss or crash and 4 cases of mate on the board. The most curious case was a loss due to an illegal move by Delphil, I've never seen something like that happen to an engine on TCEC.

Moves per game

Median= 55
Average= 61.7

The distribution has a long tail due to several long games. For example there were 4 games with more than 200 moves, the maximum was 274 for Jellyfish - Chiron, round 50. On the other hand there were a few very short games, the minimum was 10 moves for Critter - Laser, round 10.

Time per game (hours)

Median= 1:01
Average= 1:01

The time distribution is almost symmetric, when playing many moves the engines played faster.


There were 2-move book openings in this stage chosen by Cato. The first letter of the ECO codes was distributed as follows:

Open games (C) and indian openings (E) were under-represented, very similar to stages 1a and 1b that also had 2-move openings. Is this a Cato preference or do the engines prefer to transpose to other openings? 

If we use the opening 'family name' (using format FAMILY_NAME: VARIANT....) the top 3 are:
Sicillian - 151 times
English - 107 times
QGD - 78 times

The engines had a lot of freedom to choose the opening variant. Only 40.7% of the game pairs repeated the same ECO code twice, and only 21% repeated the same opening variant. In 15.7% of the game pairs the ECO family was not repeated twice.

Reverse pairs, wins

With such a low draw rate it's hard to draw conclusions from this histogram. There were 22/496 openings that had a two win 1-1 biased result, and some of the openings were one-sided but only a few. I think luck and weak engines are more important factors here.

Reverse pairs, same moves 

The reverse games diverged very quickly, half diverged in the first move and 83% diverged at or before the third ply. Only 7 games repeated more than 6 plys.

No comments:

Post a Comment