| |
BGonline.org Forums
unusual match
Posted By: Henrik Bukkjaer In Response To: unusual match (Timothy Chow)
Date: Thursday, 14 July 2011, at 12:38 p.m.
Not considering variance reduction but thinking of CMC's observation.
Doesn't the gap of 2.5 percent tell us something (given enough matches). The bot plays at it's best level and analyze at the same level. It says "I should win x%" but it only wins y%. That difference must be the net worth of what the bot doesn't understand correctly? Why else would there be a difference? The actual result is a fact (again, given enough games) that's the inescapable truth about what happens when "skill/style x" plays "skill/style y". If the bot over evaluates it's own side, then there must be room for improvement!
This could be interesting: See such data broken into different sets, playing different levels of opponents and playing different styles/game types.
| |
BGonline.org Forums is maintained by Stick with WebBBS 5.12.