Re: [Bug-gnubg] RE: Bug-gnubg Digest, Vol 80, Issue 6

bug-gnubg-bounces+massimiliano.maini=amadeus.com@xxxxxxx
wrote on 06/07/2009 18:38:49:
> > From: Massimiliano Maini massimiliano.maini@xxxxxxxxxxx wrote"
> > ...
> > In matchplay, imagine a situation where a very long match goes
up to DMP 
> > without
> > any particular luck for the two players. Then , at DMP bearoff,
one roll a 
> > super joker
> > and wins. In terms of Total Luck this will appear clearly (like
a +50% in 
> > MWC), but in 
> > normalized luck this will not. The same situation in a much shorter
match, 
> > despite
> > having the same total luck (+50%) could have a "Go to Vegas"
rating 
> > instead of
> > a "lucky" or a "none" rating.
> > ...
>  
>   Suppose that the first game in that match had also been decided
by
> a superjoker. That game 
> counted one towards the goal just like the last game. You just 
> didn't realize its importance at the time. 


That's true, but what about a joker
that brings you a gammon win instead
of a normal win ? For sure it hasn't
got the same value in game 1 of a
long match or at DMP ... Raw (unnormalized)
luck must take into account
cube value and, for match, the score.

In money play, a superjoker with the
cube at 4 is surely worth more than
the same superjoker with the cube at
2 ...
At the moment in gnubg, the "luck
in EMG" associated to the two is the
same, which is fine by the definition
of EMG luck. What is not fine (imo)
is that the luck rating is computed
from the EMG luck.

I have handcrafted a simple example
of a 2 games money session.
In each game there's only 1 bearoff
position.
In game 1 white holds a 2 cube and
gets a lucky roll, subsequent rolls
are nothing special.
In game 2 we have the same position
with black and white swapped and with
the difference that the cube (owned
by black) is at 4. They play the same
sequence of rolls (swapped, of course:
black gets the lucky roll).

The two players have had, over the
two games, the same normalized luck
and hence the same luck rating, but
in terms of net luck (in points),
black has won more thanks to luck:
its luck happened when the cube was
at 4 (for him) instead of at 2 (for
white).

In the overall stats, the luck rating
is the same for the two players
(Go to bed, in my example) but the
luck adjusted result is not zero.

A player witnessing the session would
tend to say that black has been
luckier than white, in agreement with
the luck adjusted result and not
with the luck rating.

I can send the sgf file on demand.

On top of modifying the way the luck
rating is computed (using raw luck
instead of normalized luck), I would
also advise, in the Luck panel,
to write the two lines:

         
                    gnubg
               MaX    
            

Luck total EMG (Points)    
  -0.597 ( -2.208)     -0.622 ( -1.498)
Luck rate mEMG (Points)    
  -85.3  ( -0.315)    -103.6  ( -0.250)


As:

         
                    gnubg
               MaX    
            

Luck total Points (EMG)    
  -2.208 ( -0.597)     -1.498 ( -0.622)
Luck rate Points (mEMG)    
  -0.315  ( -85.3)     -0.250 ( -103.6)

Since for most practical purposes,
it's raw luck that matters and not
normalized luck. In the above, "Points"
becomes "MWC" for match.

MaX.

P.S.
BTW it sucks that no attachment, no
matter how small, can be sent to
the mailing list, can we configure
this ?
_______________________________________________
Bug-gnubg mailing list
Bug-gnubg@xxxxxxx
http://lists.gnu.org/mailman/listinfo/bug-gnubg