Google Groups no longer supports new Usenet posts or subscriptions. Historical content remains viewable.

Dismiss

GNUBG match equity table

0 views

Skip to first unread message

Zorba

unread,

Dec 20, 2003, 1:14:13 PM12/20/03

GNUBG 7pt Match Equity Table

Method: Full rollouts of initial position, 6480 trials x 2 (or less
when SE are low enough).

Settings: 0-ply checker play, 2-ply 25% (100% for Post-Crawford) cube.

Note: 95% confidence intervals mostly around or below 0.100% MWC, the
extra digits in the MET are for calculation purposes and to preserve
as much accuracy as possible when extending the table.

The scores are to be read as "n points-to-go", more commonly known as
"n-away"

Post-Crawford

% MWC Trailer 1 2 3 4 5 6

1 50.000 48.811 31.770 30.588 18.538 17.763

(Pre-)Crawford

% MWC 1 2 3 4 5 6 7

1 50.000 68.145 75.077 81.768 84.451 89.025 90.898

2 31.885 50.000 59.964 67.184 74.524 80.195 84.430

3 24.923 40.036 50.000 57.373 65.026 71.424 76.514

4 18.232 32.816 42.627 50.000 57.745 64.533 70.186

5 15.549 25.476 34.974 42.255 50.000 56.828 62.867

6 10.975 19.805 28.576 35.447 43.172 50.000 56.315

7 9.102 15.570 23.486 29.814 37.133 43.685 50.000

This MET is somewhat close to (especially) the Jacobs MET, but there's
also some notable differences. You can use "Menu: Analyse -> Match
Equity Table" to see the MET your gnubg is using, and click "Load
table..." to select different ones (*.xml files).

--
_
/
_ orba