KCEC
(Kirr's Chess Engine Comparison)
A tournament of original free chess engines
April 21, 2012
Testing summary:
Total: 122,915 games
played by 187 programs
1277 CPU days (X2 4600+)

White wins: 49,970 (40.7%)
Black wins: 42,759 (34.8%)
Draws: 30,186 (24.6%)
White score: 52.9%

Draw Rate

Overall draw rate

Out of 122915 KCEC games played so far 30186 ended as a draw. This makes the overall draw rate of 24.6%.

Draw rate vs rating difference

Naturally the chance to get a draw is higher when both engines are of similar strength:

Each point represents a match of two engines.

Linear regression (blue line in the graph above):

Draw rate = − R.Diff / 21.20 [±2.01] + 28.44 [±0.47]

Draw rate vs average rating

As expected, strong engines make more draws:

Each point represents a match of two engines.

Linear regression (light blue straight line in the graph above):

Draw rate = Av.Rating / 72.16 [±5.68] − 8.22 [±2.60]

Exponential curve (purple curve in the graph above):

Draw rate = exp((Av.Rating − 1829.8 [±212.7]) / 348.74 [±73.28]) + 18.56 [±1.22]

Draw rate vs rating difference and average rating

If you try to use both Rating difference and Average rating, you may end up with something like this:

Draw rate = − R.Diff / 30.83 [±4.41] + exp((Av.Rating − 2184.3 [±120.8]) / 237.57 [±46.63]) + 23.64 [±0.96]

Created in 2005-2012 by Kirill Kryukov
Updated on April 21, 2012