Re: [Bug-gnubg] error rate and match winning chances

[mailinglists]

Hi Nis,

first of all thanks for a great explanation. Indeed I did  
misread/misinterprete the LAR before.

Zitat von Nis Jørgensen <address@hidden>:

> Mueller Achim skrev:

> Christian wrote something which could be misread. Jan's description is
> more precise. Estimated Win % = 50% + LAR

The reason why I'm diving deeply into it is the following. I try to  
figure out whether I'm able to beat the rake on certain backgammon  
servers. Of course some reliable numbers and calculations wouldd help  
here enormously. A few examples from last night. I'm always player B,  
besides the last one (3p) all matches to 1p, a * shows who won:

Player:                           A         B
Snowie error rate:                4.5      0.8
Luck adjusted result:           -13.13   +13.13*
Error based abs. FIBS rating:    1667.8   1991.9

Snowie error rate:                3.5      1.7
Luck adjusted result:           +13.50   -13.50*
Error based abs. FIBS rating:    1693.0   1910.1

Snowie error rate:                5.4      0.3
Luck adjusted result:           -11.94   +11.94*
Error based abs. FIBS rating:    1576.3   2026.0

Snowie error rate:               21.3      0.1
Luck adjusted result:           -10.38   +10.38*
Error based abs. FIBS rating:    1619.3   2046.6

As you might have guessed already, this is from one tournament.

My key question now is: how much of an advantage do I need to beat a  
rake of 10%, 15% or 20% (that's what they take on certain bg servers  
for low level tournaments).

   Fibs formula (rounded)     LAR     Luck Rate

1)    59%                    63.13%   36.87%
2)    56%                    36.50%   63.50%
3)    62%                    61.94%   38.06%
4)    62%                    60.38%   39.62%

The first column is something I can deal with, the second and third  
not. I do not have the Snowie MWC present, but I bet they totally  
differ from the column LAR.

Ciao

Achim