Re: [Bug-gnubg] Luck rates ?

[Christian Anthon]

Hi Max,

yes I mean EMG. To clarify the argument:

suppose you are playing a 8 point match. In the first game your are
really unlucky in terms of EMG, but because the cube isn't turned you
end up loosing just three points. In the second game you are a bit
less lucky in terms of EMG (compared to the bad luck in the first
game), but now the cube is turned to 8 and you end up winning 8 points
for the match. Now your EMG luck is negative for the match, but your
match equity luck could be very positive.

I suppose that you should consider your self lucky for the match, and
therefore prefer the match equity numbers. However, I like to have the
numbers the way they are to keep them comparable to the error rates.

Christian.


On Nov 12, 2007 2:16 PM, Massimiliano Maini
<address@hidden> wrote:
>
> "Christian Anthon" <address@hidden> wrote on 12/11/2007
> 14:09:13:
>
>
>  > Hi Max,
>  >
>  > the first numbers in the luck rate is calculated from the money
>  > equities while the second is from the match equities. So while
>  > unusual, it is certainly possible to have different signs for the
>  > two.For example with -0.068 (+16.885) you have been unlucky in terms
>  > of the money equity, but your luck was in a game with a lot of match
>  > equity at stake, so overall that number is positive.
>
> Hi Christian,
> when you say "money equity", do you mean EMG (Equiv to Money Game) equities
> ?
> Because if that's the case I do not expect sign changes (at least not big
> ones),
> the "normalization" should take care of the "match equity at stake" when
> the luck happened.
>
> And if it's not the case (i.e. the 1st number is really computed from
> pure money equities), then the 1st number has a very small meaning in
> a match context ... maybe we should just drop it and display only the
> percentages.
>
> MaX.
>