I have setup a 500 games tournament between the new Houdini 2.0 and the older Houdini 1.5a to see what the results will be.
I set it up for a 20 minutes tournament, thus about 40 minutes game each, with ponder on and ran it on a 32 bit XP PC.
Unfortunately my PC crashed after 175 games. I have started it again, but here is the results thus far.
Houdini 1.5a won 43 games, Houdini 2.0 won 51 games and 81 games were drawn.
Houdini 1.5a scored 83.5 points (47.7%) and Houdini 2.0 scored 91.5 points (52.3%).
White have won 55 times and black 39 times. (A little bit against stats... would have expected a closer battle between white and black?)
The ELO difference is 16 with 175 games. I read somewhere that Robert Houdart said that the ELO is about 15? If so, then that is also what I got thus far.
I hope Robert respond on the amount of times that white has won more than black. Is that a significant statistic or does it have something to
do with the software programming, that gives white a little bit of unfair advantage over black?
In conclusion, with 175 games I feel confident that Houdini 2.0 is indeed stronger than Houdini 1.5a. It is up to everone to decide if that margin is
way stronger or not that much. My personal feeling is that on that high level of competition (above 3000 EL), even a 1 ELO higher rating, is actually
just as important/stronger than say a 10 ELO rating higher, on a 2000 ELO scale. In my eyes Houdini 2.0 has taken over the reigns from Houdini 1.5a
as the strongest chess engine currently available. Long live Houdini!
