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

BGonline.org Forums

Small Cube Error Reported by XGR++. Why?

Posted By: Henrik Bukkjaer
Date: Friday, 7 February 2014, at 12:23 p.m.

This is a position from a match I have just reviewed for botnorm approval in the DBgF.

Freddie Noer played 1.98 (XGR++ Gigantic) in a live clocked 17-pointer, in a round of the Danish Team Championships vs. Henrik Hansen (a strong intermediate player, who would be expected to play a PR between 5 and 8).

What I just stumbled on, has nothing to do with the analysis of the actual match. But it's a position during post-Crawford play, where Freddie's opponent doubles at his second opportunity (in what XGR++ evaluates as an "Automatic Double / Take"):

is Freddie Noer

score: 16
pip: 152
17 point match
pip: 156
score: 8

is Henrik Hansen
on roll, cube action?

Analyzed in XG Roller++ No double Double/Take
Player Winning Chances: 47,51% (G:9,53% B:0,29%) 47,44% (G:9,76% B:0,32%)
Opponent Winning Chances: 52,49% (G:14,13% B:0,55%) 52,56% (G:14,51% B:0,57%)
Cubeless Equities -0,035 +0,147
Cubeful Equities
No double:+0,144 (-0,003)
Double/Pass:+1,000 (+0,853)
Best Cube action: Double / Take

eXtreme Gammon Version: 2.10, MET: Kazaross XG2

Now, my question is this: Why would XGR++ give a penalty of 0,003 here, to a player NOT doubling?

I would certainly not double this position myself (well maybe I would in this particular case playing Freddie Noer, but for 95% of my opponents I wouldn't).

Looking at the two strongest exchanges that could happen from here (33 or 44, followed by a dance), XGR++ correctly evaluates those as Double/Take, thus no market loss in the position.

The odd thing is, that XGR++ provides the worst evaluation of all, from 1-ply to 7-ply all mini rollouts included. I've seen this before in post Crawford game cubes - is it in general a problem related to XGR++ few games when looking at positions that are about potential market loss?

Especially it surprises me that XGR++ does worse than XGR+, since they both operate with variance reduction, and have the same truncation point (7 moves).

I cannot recall how many games XGR++ and XGR+ rolls out, maybe that's where the answer lies? Anyone have those numbers?


Messages In This Thread


Post Response

Your Name:
Your E-Mail Address:

If necessary, enter your password below:




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

BGonline.org Forums is maintained by Stick with WebBBS 5.12.