New Metric

F1 Shift.net

Points Scorer
Hey guys,

Long time reader, first time poster. Anyways, since this season has been so crazy and seems consistency is the key the championship, I created a simple metric to help measure which driver is performing the best. I call it Total Driver Performance (clever I know). It combines Average points per race and average positions gained/lost per race to come up with the TDP. Formula: average points per race divided by 25 + average positions gained/lost divided by 23. That total is then divided by two. Let me know what you think and if you have any suggestions, please let me know.

Driver		Points	Total Driver Performance
Alonso 76 0.3186
Vettel 73 0.2796
Webber 73 0.2506
Hamilton 63 0.2390
Raikkonen 51 0.2207
di Resta 21 0.1570
Rosberg 59 0.1532
Perez 22 0.1204
Vergne 4 0.0894
Glock 0 0.0833
Maldonado 29 0.0604
Button 45 0.0594
Senna 15 0.0572
Karthikeyan 0 0.0507
Massa 10 0.0406
Kovalainen 0 0.0290
Hulkenberg 7 0.0270
Petrov 0 0.0217
Pic 0 0.0217
Kobayashi 19 0.0162
de la Rosa 0 0.0145
Ricciardo 2 -0.0114
Grosjean 35 -0.0609
Schumacher 2 -0.2180
 
It's an interesting concept, and there's some interesting results (di Resta, Glock and Grosjean, for example).

Now, you asked for suggestions, and I'm afraid I get myself far too interested in these things so you're probably about to get an essay!

It's difficult to improve without making the system much more complicated, but I think the average positions gained/lost has a large dependence on starting position. It's impossible for the driver starting last to lose places and impossible for the driver starting first to gain places. To try and make it more fair, I'd suggest somehow giving extra weighting to places gained in the first few positions and extra weighting to places lost in the last few positions.

I've attached a little excel spreadsheet with a way you could do this and a few examples. Basically, rather than dividing by 23, you could divide by the number of places it was possible for the driver to gain (if he moves up the field) or by the number of places it was possible for him to lose (if he moves down the field). A driver starting 2nd and finishing 1st then gets +1 in this system, versus +0.09 in the original system, for example (more examples in the spreadsheet).

This still disadvantages the guy who starts on pole, however. In fact, it probably provides a massive advantage to drivers who start say in the top 5 and move up a couple of places. So, this system may not be the right one to use either, but you see the sort of thing I'm suggesting!

You could also try to include something that compares the performance of a driver to that of their team mate. This could be something like (for driver A vs. team mate B):

(Driver Performance A - Driver Performance B)/(Drive Performance A + Driver Performance B)

This gives a value between -1 and +1, +1 being when driver A completely outclasses driver B and -1 when the reverse is true.

My suggestions probably lead to a reduced contribution from the actual points though, so you may want to increase this but you can have a play around... if you're interested in anything I've said, feel free to ignore me!
 

Attachments

  • newmetric_weighting.xlsx
    10.2 KB · Views: 156
Good data, it's always hard to quantify this season as there are so many variables, but to improve it would it be better to factor in relative car performance, mechanical failures and overtakes? Because, from what I can remember, Karthikeyan has lucked into many of his finishing positions, mainly because of mechanical failures, and he has performed very few actual overtakes for position. But great concept!!!
 
Thanks for all the suggestions and ideas, I really appreciate it. I know my metric isn't perfect, but it gives an idea anyways, so incorporating you're thoughts can make this thing work! sushifiesta I will work with the starting position idea and see how I can use that, I think that's great. I'll take a look at the weighted system too. mjo I'm not sure how relative car performance could be factored in, but I'll try to think of something!

I also have an overtake/overtaken metric (based on the data collected here) that I'll post up later tonight.
 
Here's the overtaking metric. This is nothing fancy at all, but still fairly interesting. Thanks to the great data collectors at this great site for letting me use their info. Simple: Overtakes-Overtaken divided by career GP's competed in. Keep in mind I did not put an emphasis on weather for this, so all overtakes or overtaken carry the same weight:

Driver GPs Overtakes Overtaken Total Overtake Performance
Hamilton 95 193 43 1.5789
Vettel 87 89 32 0.6552
Massa 159 219 116 0.6478
Perez 24 89 74 0.6250
Hulkenberg 24 41 28 0.5417
Alonso 184 171 76 0.5163
Webber 183 187 95 0.5027
Button 215 191 118 0.3395
Schumacher 294 158 78 0.2721
Maldonado 25 81 75 0.2400
Raikkonen 163 87 49 0.2331
di Resta 25 76 71 0.2000
Grosjean 13 17 15 0.1538
Rosberg 114 135 127 0.0702
Kobayashi 46 107 107 0.0000
de la Rosa 92 32 39 -0.0761
Vergne 6 23 25 -0.3333
Petrov 44 81 104 -0.5227
Senna 32 39 71 -1.0000
Kovalainen 96 92 189 -1.0104
Ricciardo 17 27 50 -1.3529
Glock 80 60 172 -1.4000
Karthikeyan 32 7 77 -2.1875
Pic 6 2 22 -3.3333
 
Has Vettel only taken part in 32 gp's? And has Karthikeyan really overtaken 32 people?
Very good data again.
 
I have worked with the Total Driver Performance metric and have a pretty good solution. Instead of using 23 as the possible positions gained, I used average starting position minus 1. This still does not take into consideration mechanical failures or relative car performance, but right now I don't have the time or energy to go through and figure out who's race ended based on mechanical failure, driver error, ect. So here's the new and (slightly) improved TDP. Again, open to suggestions.

TDP.JPG
 
Back
Top Bottom