Probability of Win 2008

[Schummy]

Three weeks without GPs (even if we have MM's videos to pass the time) is a bit too much. So I have rescued a kind of stat I did last season; it is about calculation of probability based on results in these three GPs and last season's results.

In the next table the last column is the estimated "probability of win". Below that appears probability of win for teams.

....D07 T07 07  | 08
============================
RAI 31% 25% 27% | 34% => 30%
MAS 19% 25% 23% | 28% => 25%
HAM 25% 24% 24% | 25% => 25%
KOV  1% 24% 16% |  0% =>  9%
HEI  1%  1%  1% |  6% =>  3%
KUB  0%  1%  1% |  7% =>  3%
ALO 24%  1%  8% |  0% =>  5%
PIQ  0%  1%  1% |  0% =>  1%
============================

FER 55%
MCL 34%
BMW  6%
REN  6%

Columns are:

• D07: Driver's %performance in 2007
• T07: Team's %performance in 2007
• 07: Driver's 2008 a priori performance, weighting his 2007 performance (1/3) and his present team 2007 performance (2/3).
• 08: Driver's 2008 present performance in these 3 GPs.

Some details are as follows. "07" and "08" columns are combined (in a bayesian way with weights 4 and 2.85).
Performances are similar to wins, but a win equals to 0.8 and a second position values 0.2. Past GPs are discounted (faded) a 5% per GP.

As can be seen, Ferrari and McLaren are clearly in the top, with advantage for Ferrari. BMW and Renault are the other two teams with discernible probability of win (Renault is there because of Alonso's performance).

[Schummy]

Two GPs have passed and now I have a bit more data about probability of win.

The table is now arranged a bit differently in its structure of columns.

     CW  PW   EFW    FWs    TWs   ETW   PMW
RAI   2  25%  3,26  1 - 6  3 - 8  5,26  48% 
MAS   2  23%  2,96  0 - 5  2 - 7  4,96  40%
HAM   1  20%  2,59  0 - 5  1 - 6  3,59  12%
KOV   0   9%  1,16  0 - 3  0 - 3  1,16   0.2%
KUB   0   7%  0,88  0 - 2  0 - 2  0,88   0.1%
HEI   0   6%  0,73  0 - 2  0 - 2  0,73   ---
ALO   0   5%  0,60  0 - 2  0 - 2  0,60   ---
ROS   0   2%  0,25  0 - 1  0 - 1  0,25   ---

CW: Curent number of wins (after GP Turkey).
PW: Probability of win.
EFW: Expected future wins in the rest of the season.
FWs: Probable (95%) interval of future wins.
TWs: Probable (95%) interval of total wins in the season (CW+FWs).
ETW: Expected total wins in the season (CW+EFW).
PMW: Probability of being the most winning driver in the season.

The key of those calculations is the estimation of PW. It is done as formerly said in the first post. It is just an orientating value based in a particular model and sample values of this season.

Usually the driver with most wins gets the title (just *usually*), so PMW is a good guess for the probability of winning the champioship. The data would show there is a close fight between KR and FM, but LH also has a chance. Interestingly (or perhaps obviously) the title chase looks already limited to three drivers.

The next GP, Monaco, is the most particular one in the schedule, so perhaps things can change a bit in two weeks time.

[duke_toaster]

Nice stats. I think that Monaco and Canada could swing the pendulum back towards McLaren, but who knows ...