[Bug-gnubg] BUG: Calculating gammon prices for cubeless matches

[Ian Shaw]

I have found a few problems when trying to calculate a MET for cubeless
matches.

I have doubling, Crawford and Jacoby off.

I have rolling settings as 0-ply, with cubeful disabled on the General
tab and the First Play Both tab. (What's the difference?)

I rolled out as initial position with each side on roll, to generate the
attached MET, which I have completed for a 4-point match.

Problem 1. Gammon Price

At 1-away 3-away, I get a gammon value of 0 for both sides. This is
correct for 1-away, but 3-away's gammon value should be 0.55 (using
32.3% for 1-away 3-away).

I get the same result whether or not the game is set as the Crawford
game.

Problem 2. 1-away 5-away

When I roll out or evaluate MWC for 1 away, I get a value of about 57%.
This is clearly wrong; it should be at least 89%. Rolling out scores up
to 3-away 4-away have yielded believable numbers, as you can see from
the attached MET. I haven't changed any settings.

I suspect that the two problems are related. At 1-away 3-away, I think
that gnubg assumes the current game will be doubled sooner or later, so
gammons become irrelevant.

-- Ian

Cubeless.xml
Description: Cubeless.xml