nuu1212
(Banned)
- 282
- Germany
Inspired by a recent critical video by TidgneyRacing, I want to outline a simple solution to the FIA points system.
The general thought is that DR ratings as well as FIA points must follow the same simple and perfect mathematical system like ELO in chess!
I guess PD tried to follow this example for DR, but they failed to get the details correct.
1. It makes absolutely NO SENSE to cap the rating
In the chess example, what would be the point of all Grandmasters being capped at e.g. a 2600 rating!?
2. The FIA Points must reflect your performance with respect to the ratings of all other participants and thus must be DR ratings, too!
There is no other fair way to award points other than to directly use the ratings and the relative performances during a specific race!
Tidgney was on the right track when suggesting that the average DR of a race must set the baseline for that split.
I have no interest in working out a complete formula/algorithm here. Space is limited and I am not getting paid for it.
I guess the problem has already been solved for chess tournaments, as e.g. in this reference
http://www.chessmetrics.com/cm/CM2/Formulas.asp
The bottom line is that we have to use the result of this one race to estimate the driver performance (which must obviously be done directly in DR rating)
A simplified linear version would look like this: (this is just a hint, take with a grain of salt before slaughtering!)
Say a lobby has DR ratings from 6000 to 7000 with an average of A=6600 and a standard deviation of SD=250.
Drivers would then be awarded points in the range from A-x*SD to A+x*SD from last to first, where x is a fixed weight (x=2...3 should be good values)
Resulting in the following points for 11 drivers from 1st to 10th place: 7100 - 7000 - ... - 6200 - 6100.
The actual race time differences could even be accounted for by not using a linear spacing...
The issue of how to account for different splits will always remain, but it can be minimized by choosing the details wisely.
There could also be an algorithm making use of the overall results.
3. Sticking to DR rating as the result of a race also solves the problem of multiple entries to FIA races
Your final points should just be the average of your individual entries.
Using your maximum performance as the final result also is a mathematically sound option, but probably not as fair as the average.
It is just a matter of taste at this point.
4. One more thing:
The fact that FIA points are nothing other than DR ratings could (and should) probably be disguised from "the masses" by simply using a fixed factor. Such that e.g. a DR=70000 race performance would result in DR/20 points for an end result of 3500 points!
The general thought is that DR ratings as well as FIA points must follow the same simple and perfect mathematical system like ELO in chess!
I guess PD tried to follow this example for DR, but they failed to get the details correct.
1. It makes absolutely NO SENSE to cap the rating
In the chess example, what would be the point of all Grandmasters being capped at e.g. a 2600 rating!?
2. The FIA Points must reflect your performance with respect to the ratings of all other participants and thus must be DR ratings, too!
There is no other fair way to award points other than to directly use the ratings and the relative performances during a specific race!
Tidgney was on the right track when suggesting that the average DR of a race must set the baseline for that split.
I have no interest in working out a complete formula/algorithm here. Space is limited and I am not getting paid for it.
I guess the problem has already been solved for chess tournaments, as e.g. in this reference
http://www.chessmetrics.com/cm/CM2/Formulas.asp
The bottom line is that we have to use the result of this one race to estimate the driver performance (which must obviously be done directly in DR rating)
A simplified linear version would look like this: (this is just a hint, take with a grain of salt before slaughtering!)
Say a lobby has DR ratings from 6000 to 7000 with an average of A=6600 and a standard deviation of SD=250.
Drivers would then be awarded points in the range from A-x*SD to A+x*SD from last to first, where x is a fixed weight (x=2...3 should be good values)
Resulting in the following points for 11 drivers from 1st to 10th place: 7100 - 7000 - ... - 6200 - 6100.
The actual race time differences could even be accounted for by not using a linear spacing...
The issue of how to account for different splits will always remain, but it can be minimized by choosing the details wisely.
There could also be an algorithm making use of the overall results.
3. Sticking to DR rating as the result of a race also solves the problem of multiple entries to FIA races
Your final points should just be the average of your individual entries.
Using your maximum performance as the final result also is a mathematically sound option, but probably not as fair as the average.
It is just a matter of taste at this point.
4. One more thing:
The fact that FIA points are nothing other than DR ratings could (and should) probably be disguised from "the masses" by simply using a fixed factor. Such that e.g. a DR=70000 race performance would result in DR/20 points for an end result of 3500 points!
Last edited: