| |
BGonline.org Forums
BQ: Hit or not hit?
Posted By: leobueno In Response To: BQ: Hit or not hit? (leobueno)
Date: Sunday, 24 April 2011, at 12:57 a.m.
I am trying to use GNUbg as a coach. Whenever I do a rollout of a position, I would like GNUbg to tell me which is the best play. I can then think-through the position and attempt to figure out *why* play A is the best play, thereby developing my understaning of the game.
This is obviously not as good as seeing your (Nack's) and other experts' take on the position, but at least it gives me a starting point.
So, if GNUbg says that play A is the best play, I would like to be fairly sure that play A is indeed the best play.
I realize that since GNUbg uses a simulation technique, the best we do is obtain a level of confidence in GNUbg's conclusion. This is what I am trying to ascertain: how sure (statistically confident???) can I be that play A is the best play.
In the example in question, play B (8/4 6/4) is less than 3 JSD's away from play A (15/11 13/11). How "confident" can I be that play A is better than play B? The last thing I want to do is develop my understanding of the game based on flawed results from GNUbg.
I realize the concept of statistical confidence is often misunderstood, so when we couple that idea with GNUbg's numbers, we need to make sure we all understand what we are discussing.
So, what can we say about GNUbg saying that 15/11 13/11 is better than 8/4 6/4 based on the posted rollout results?
| |
BGonline.org Forums is maintained by Stick with WebBBS 5.12.