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

BGonline.org Forums

Variance Reduction Question

Posted By: Phil Simborg
Date: Friday, 4 March 2011, at 3:28 p.m.

In Response To: Variance Reduction Question (Casper van der Tak)

This whole concept of where to believe the bots and where not to is difficult. I believe we are at the point that with XG we can have strong confidence in a huge percentage of the rollouts, but it's that remaining 1 percent or so that makes us question everything.

For the "purists" who want to know the answer definitively to the Nth degree, I'm not sure we will ever have a bot they will trust, but for MOST OF US, I have one word: CHILL. You not only can trust the rollouts, you can also pretty much trust XGR+ and not even waste your time with rollouts for most positions. (It might give you the wrong answer occasionally, but most of the time even when wrong it will be close.)

When XG came out, I took 30 positions that Snowie and GNU got wrong and found that XG only got 2 of them wrong.. (I published with positions and results here a couple of years ago).

Occasionally I do find some XGR+ positions that do change with rollouts, but I have yet to find one that changed "significantly." If others do, THAT would make an interesting, and very helpful database and THAT would be a great tool for Xavier to find out where he needs to improve his program.

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.