Re: [Bug-gnubg] Redouble ?

[Joern Thyssen]

On Sat, Jul 26, 2003 at 04:04:53AM -0400, address@hidden wrote
> Hi,
> thanks for answering my last bug report so fast and very detailed indeed !
> 
> I don`t know if this is a bug at all, more likely I am to blame of
> ignoring simple mathematical facts...but I have a faint feeling that
> maybe there's something wrong with gnubg's cube analysis (at least
> the fibs opponent of my nephew and I had missed the alleged redouble
> proposed by GNUBG 0.14 Jul 2 2003).
> 
> IMHO it is not a redouble in a money game, but how about the redouble in
> match play, especially in a 5-point match trailing 0-1 ?
> Who of your guys can do the math exactly ?
> Unfortunately I don't know much about taking points and match
> equity tables, so could you please check if this redouble was correct ?

You can use gnubg for this: go into Anayse->Market window. For player NN
the double point is 50% and the cash point is 70% (the numbers may vary
a bit with the match equity table). According to the cube analysis you
have 58%, so you're above the double point, so you should consider
doubling. It's obviously a take since the 58% is less the the cash point
of 70%.

Whether it's a double or not depends on the market losers or market
losing sequences. You can use Analyse->Temp map (cube decision) to see
the market losers. Anyway, gnubg estimates you have enough market losers
to double at this match score.


BTW, this position is so simple that gnubg has the exact equities for
money play (use Analyse->Evaluate):

Player       Opponent
Position          216           124

Cubeless equity               : +0.1571
Owned cube                    : +0.2722
Centered cube                 : +0.2121
Opponent owns cube            : +0.0971


So in money game:

No double (owned cube)            :      +0.2722
Double, take ( 2 * opp.owns cube) :      +0.1942

so as you say, it's not a redouble for money.


Jørn