| |
BGonline.org Forums
eXtreme Gammon question
Posted By: Timothy Chow In Response To: eXtreme Gammon question (Bob Koca)
Date: Wednesday, 26 August 2009, at 8:48 p.m.
For example supose you are only giving it problems in which you are pretty sure that play A is much better than play B. The higher scoring play will be correct more than what the confidence level says.
If that's what the bots do, then calling it the confidence level is rather misleading. It should be clear exactly what statistical decision is in question when a confidence level is reported.
Suppose two plays are not exactly equal. As the rollout progresses the two individual deviations get smaller but the JSD gets larger and larger.
What is the j.s.d. if not the root mean square of the two individual deviations?
| |
BGonline.org Forums is maintained by Stick with WebBBS 5.12.