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

BGonline.org Forums

XGR++ versus n-ply for bearoffs

Posted By: Rick Janowski
Date: Friday, 28 November 2014, at 3:55 p.m.

In Response To: XGR++ versus n-ply for bearoffs (Timothy Chow)

I have compared known equity positions in Sconyers' Bearoff Database (which is fully calculated out for unlimited games) with XG analysis and find that the evaluations from XGR++ and then XGR+ are superior to the 7-ply results. Also the accuracy is astonishing, rarely out by more than 0.005 to my recollection. I believe the reason lies in their use of mini-rollouts which gets a much better handle on cube ownership, both for the favourite and particularly so for the underdog.

If XGR++ is the most superior method of analysis for unlimited games then it naturally follows that they will also provide the best analysis tools for other matchscore positions where cube ownership is relevant.

7-ply might be superior in evaluating cubeless winning chances but the differences if any will always be negligible I believe.

We should thank Xavier and Neil Kazaross for creating these remarkably efficient hybrid analysis models, that enhance the bot's evaluation capabilities considerably.

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.