Houdini 2.0c T3-setting test (10000 games)

[pohl4711]

The T3-setting played 10000 games (LS-rankinglist conditions, same opponents as Houdini 2.0c default) and scored 0.6% weaker than the default-engine. So it will not be in the LS-rankinglsit - who cares about a setting, which is not better than the default-engine? (-0.6% in 10000 games is inside errorbar - so it is not sure, if the setting is weaker, but it is sure, that it is not stronger!).

Best - Stefan

[ernest]

...and yet it started (see CSS- Forum):
Houdini 2.0c x64 - Houdini 2.0c T3 188.0 - 212.0 +104/=168/-128 47.00%

Good work, Stefan: nothing like statistics!...

[pohl4711]

The 10000 games test of the T3-setting was for the LS-rankinglist and did not include games against Houdini 2.0c default. The test of the settings (inclusive T3) against Houdini 2.0c default was a test in advance to find the best setting for the final testrun for the LightSpeed-rankinglist.

Best - Stefan

[ernest]

OK Stefan, I understand!
Indeed, with 10000 games, the statistics are significant...