[ View Thread ] [ Post Response ] [ Return to Index ] [ Read Prev Msg ] [ Read Next Msg ]

BGonline.org Forums

BQ: Hit or not hit?

Posted By: Timothy Chow
Date: Sunday, 24 April 2011, at 2:57 a.m.

In Response To: BQ: Hit or not hit? (leobueno)

My recommendation is to focus on large errors rather than getting unduly obsessed with cutting down the variance with massively long rollouts of close plays.

If GNU 4-ply says that your checker play is wrong by 0.100 or more, then it's very likely that your checker play is a big blunder. This is something you should work hard to understand.

On the other hand, suppose you do a 46656-game rollout of two checker plays and find that Play A beats Play B by 0.012 and the standard deviations are 0.003. Now you have a lot of statistical confidence that GNU thinks that Play A is slightly better. But should you expend a lot of effort trying to understand why Play A is slightly better? You could, but your game will probably improve faster if you spend more time on your large errors than on this kind of hair-splitting. If all your errors are less than 0.020 then you'll be playing at a world-class level already.

Messages In This Thread

 

Post Response
Your Name:
Your E-Mail Address:
Subject:
Message:

If necessary, enter your password below:

Password:
 

 

[ View Thread ] [ Post Response ] [ Return to Index ] [ Read Prev Msg ] [ Read Next Msg ]

BGonline.org Forums is maintained by Stick with WebBBS 5.12.