Racecar Winning Probability Index - click to find out what it means

[moonman239]

Racecar Winning Probability Index -- click to find out what it means

So I just developed an equation that calculates how likely a racer is to win a particular race. It is not a probability ratio, which is why I'm calling it an index.

RWPI(racer) = GRI(racer) / GRI(top competitors)

where the GRI (Good Racer Index) of a racer is:

GRI(racer) = (wins / total races) * avg. time

In the case where the racer competes with racers he has competed with in the past, and those competitors (out of the top) have won better places than him a few times, the driver's RWPI becomes

RWPI(racer) = (GRI(racer)/GRI(top competitors)) / ((bp / tcr) / cic)

where
bp refers to how many times the racer has gotten a worse place than anyone of the said competitors,
tcr = the number of races they've had in common
cic = the number of said competitors

 

[zli034]

First look, if we consider 2 drivers with identical number of races. And the number of wined race also identical. For example, they all had 10 races before, and all won 5 races. Then the RWPI reduces to a ratio between total time of 10 previous races of driver1 divided by total time of 10 previous races of driver2.

If the driver1 uses 50 hours total time to finish his 10 previous races, and the driver2 uses 200 hours total time to finish the driver2 10 previous races. The RWPI becomes 50/200, which is 25%. So lower RWPI means high probability to win.

If we consider they have different number of winning, same number of total races, and same amount of time to finish the same number of races. Then the higher RWPI means higher probability to win.

We need to have a valid input range for the RWPI function to be consistent.

 

[moonman239]

I suppose we could fix the algo so it uses the average place of the driver and standard deviation, instead of wins over total. So the GRI, then, is mean place plus deviation, divided by 2, multiply that by the average speed of the racer in all his races. Then the lower the number, the higher the prob.

 

[zli034]

The GRI is an index of a driver's ability. GRI is high means the driver's performance is high.

So GRI should not be proportional to average time. Like you showed on the first post. GRI increases when average time increases. Doesn't make sense.

GRI should proportional to 1/average time.

 

[pantaz]

Every race, even on the same track, will have different lap times. Even tracks equal in length (rare) will have different lap times due to factors such as track layout, temperature, barometric pressure, age/condition of track surface, etc. Also, the average speed and length of time a race takes to complete also varies according to unpredictable events, such as collisions resulting in yellow flag conditions.

 

[xxChrisxx]

Time to make the caveat that this only works in a spec series. Car >> Driver in series that is not spec.