BGonline.org Forums

XGR++ errors in holding games

Posted By: Bob Koca
Date: Sunday, 27 November 2011, at 5:38 p.m.

In Response To: XGR++ errors in holding games (mtuhtan)

The cubeless numbers are very close and even slightly better for the cuber in the XG++ rollout. It is giving the taker too much cube equity. The full rollout "sees" the position through and catches the idea that the wins from hits are often a sudden big market loss.

Is it possible to model better cube power here? Can't a neural net learn how much cube power exists for position types?

Messages In This Thread

• Are full rollouts obsolete?
  Chuck Bower -- Wednesday, 23 November 2011, at 3:28 p.m.
  • Are full rollouts obsolete?
    David Rockwell -- Wednesday, 23 November 2011, at 3:57 p.m.
    • Are full rollouts obsolete?
      Bob Koca -- Wednesday, 23 November 2011, at 8:18 p.m.
      • Are full rollouts obsolete?
        David Rockwell -- Thursday, 24 November 2011, at 3:58 a.m.
  • Are full rollouts obsolete?
    Timothy Chow -- Wednesday, 23 November 2011, at 7:46 p.m.
    • Are full rollouts obsolete?
      Bob Koca -- Wednesday, 23 November 2011, at 8:16 p.m.
      • Are full rollouts obsolete?
        David Rockwell -- Wednesday, 23 November 2011, at 10:53 p.m.
  • Are full rollouts obsolete?
    Neil Robins -- Wednesday, 23 November 2011, at 9:29 p.m.
  • XGR++ data
    eXtreme Gammon -- Thursday, 24 November 2011, at 1:33 a.m.
    • Question for Xavier and Stick
      Paul Weaver -- Thursday, 24 November 2011, at 2:25 a.m.
      • Question for Xavier and Stick
        eXtreme Gammon -- Thursday, 24 November 2011, at 3:02 a.m.
  • XGR++ errors in holding games
    mtuhtan -- Sunday, 27 November 2011, at 11:45 a.m.
    • XGR++ errors in holding games
      Bob Koca -- Sunday, 27 November 2011, at 5:38 p.m.
      • XGR++ errors in holding games
        Timothy Chow -- Sunday, 27 November 2011, at 7:20 p.m.