Trying to compile statistic on this years races

[btgoss]

I am trying to revisit an idea I had, and I may have posted about this in the past, but I am going to throw it out there again.
I am trying to work on a statistics program that will weight the finishes of all drivers in a particular class/division. Similar to baseball stats, I am trying to work on a formula that will allow me to rate the drivers, and eventually come up with a "best in class."

I want to start with the modifieds first, since they have raced a fair amount already, and there is a small amount of travelling that goes on in this class. (For my purposes travelling refers to racing at tracks and against an assortment of drivers.)

I am trying to decide what would be the measurable variables.
Obviously starting and finishing position would just some of the variables.
Home track advantage
Number of cars that start the feature
Size of track
Number of cautions
Cars on lead lap

I am open to any other ideas. The point being that I think we can come up with a statistical champion even across tracks. It should be something that is very close to the way the NCAA let's the computer pick the college brackets for March Madness.

I would like to reward consistency over wins, and pure racing over qualifying well and outlasting the field, since I am trying to show the best racer in a class.

I am open to any thoughts or suggestions. I will also admit that the math is a bit beyond me at this point, but I am going to hopefully recruit a big brain or two to help get me started.

[W. J.]

This may not be as easy as you thought at first. Even with a system of weights, the number of years vary wildly. Some drove for 20-30 years, others on 5-6-7 and then moved on or out. Some competed in upwards of 35-45 or more races a season, others less than 20. Short of finding someone like that professor on the TV show 'Num3ers', I don't know if you'll trully accomplsh what you've set out to do here. But good luck and let us know if you need any particular kind of assistance.

[Geno]

In my opinion, without making things super complicated- take the starting spot, then finishing spot of a driver for each feature. Tally them up for the year, and then divide the overall finish by the overall starts and that will give you the “batting average” that you’re looking for. It might give you a good start.

[btgoss]

W.J I am only going to start with this year. It gives me the best chance of getting the information I would need. To try and go back in time would be a challenge best tackled after I get the math straight, and get it running on the web or in a spreadsheet.

Geno I had thought about doing that way as a start, the quick and dirty way, but it leaves out so much.

Here is a situation. Ted Christopher wins the pole for a mid-week race at Waterford. Marissa Neiderauer qualifies and starts last, in 28th. Christopher wins the race going away. Marissa gets a top ten, on the lead lap. For her average that should mean more then Christophers win. At least in my opinion. Obviously number of cars finishing, number of cautions, and other factors like that would then detract from her finish. I want to try and express the big picture of finishes in statistical form.

[Geno]

A little more complicated way would be to set up categories based on a points system. The most cars passed gets x amount of points, as does a win, etc. and have the points distributed throughout the field. Then take the driver’s accumulated points from that night and divide it by the total possible points from all categories combined. Just trying to throw out a few ideas.

[W. J.]

Here's a problem. If Marisa, in your example, gets more credit (points?) than Christopher, the calculation is already skewed, because that sounds like you're penalizing someone for having a good race (winning the pole and the race). That's where the Cup point system always worked, it awarded consistancy, but didn't hurt the winner. You need to keep that in mind when coming up with a formula for yours.

[btgoss]

Being consistent is only part of what I am trying to show, and I do think Marisa should see a benefit from running well at an unfamiliar track, unless that finish was the result of cautions and cars dropping out. More importantly, Ted Christopher winning at a familiar track after having started on the pole is not as impressive, and it would be made less impressive under the same conditions that would detract from Marisa's top ten.

( I would like to point out, I used Marisa as a quick example, not knowing if she has in fact ever seen Waterford, much less raced a modified there. Additionally I know she is under financed compared to Christopher, adding to the odds that she would not finish well at this mythical race. These examples do work as they are about the two most possible extremes in terms of making a statistical system.)

So I want a system that is more then just wins and loses. I eventually want to do power rankings, and almost come up with a line on who runs well at which track.

I know it sounds very complicated, but I look at all the statistics that are generated for baseball and football and it isn't that bad.