bug-gnubg
[Top][All Lists]

## Re: [Bug-gnubg] Double Decisions

 From: Robert-Jan Veldhuizen Subject: Re: [Bug-gnubg] Double Decisions Date: Wed, 22 Aug 2007 15:56:45 +0200

On 8/22/07, Peter Carlson <address@hidden> wrote:
I dont really have any precise data.  It's more of a feeling.  Obviously
if gnu thinks it is in a losing position it wont offer a double.  I
guess what my main point was that once gnu offers, if you press hint,
the hint will > 75% of the time - based upon actual game playing -
advise you to take the double.

In money games, one can expect from doubling theory that with perfect play, 2/3 of initial doubles are takes (for redoubles, it is 1/2). Simulations with GNUBG 2-ply playing against itself indeed get very close to this number.

Then almost 100% of the time I'll lose,
following the hints for every play.

That sounds like sample bias. You should probably win at least 1/4 of games when you correctly take in a money game (depending on associated gammon rates).

For example:

gnu offers double
hint says take, so I take
gnu rolls and plays
I roll and press hint and I have *usually* less then 20% chance of
winning.  I follow each hint for every roll and I end up losing.

In money games, one should often take with winning chances of 25% (depending on gammon rate and recube vig this number can be higher or lower).

So again the real question is would gnu offer a double if it didn't
think it was in a significantly better position to win.  And if it is in
a better position to win, which I have to believe it is, why would the
hint offer to take?

Because when you drop, you also lose: 1 point right away. Without considering gammons, the basic takepoint is therefore at 25% game winning chances. Because after taking, you have a chance to redouble while opponent cannot double you anymore, the gammonless takepoint is actually as low as around 21.5%.

Hope this helps,

--
Robert-Jan Veldhuizen
(Zorba on FIBS)