Nascar Stuff

Intro To The Contenders

Statistically Speaking

Just a few meanderings, observations and prejudices regarding the actual Nascar world about which I know little. But I watch and I can pretend like all the rest. And I can put my little right here. For instance, I just sort of stumbled on a driver rating system that I kind of like. It uses no esoteric, hidden stats accessible only to the elite and the "pay-fors". You've heard of Bill James? He made tons of baseball stats and material available to the everyman in the form of a "Baseball Abstract". The first copy I bought was mimeographed and mailed by the man himself. It was only a matter of time before a big publisher signed him and bound editions found their way to the bookstores where I waited every year.

Of course, what followed were the wannabees and the statistical purists and the theorists and the capitalists who wanted to make sure that you paid more for all the newfound wisdom. But the imitators failed to realize that James brought more to the table than numbers. He brought soul and a sense of humor to logic. He had a wonderful style of writing that transcended logistics. See, he made money but his was a labor of love... the ones that followed were looking to make a buck by privatizing information that volunteers had been helping James put together for years. Most of them took themselves way too seriously.

At amy rate, I have seen links to driver ratings lately that ask for money before they'll let you view them and thus be "in the know". To tell the truth I find the whole driver rating system kind of farcical. Average mph during race, stuff like that. Ridiculous. You know what their stat system is going to tell you? Basically what the Driver Points standings tell you. I have something that is probably just as accurate. You know what the price is? Nada, zippo, zilch. Because for one thing I don't really believe any of it.  Too many variables.  There's no way to create a basic model, a control-set for this experiment. Maybe set up all the drivers on emulators? Rotate drivers and crews and cars? Give everyone the same car such as the IROC races? They still can't all use the same crew. How about you make all the teams take timed 30-second pitstops? Give everyone the same number of cars and the same amount of money? You really think so?

What I did was take and award points for all races and used these to tabulate a race score. Just like per-race points right? Well, yes and no. I won't go into a deep discussion here but to summarize: I awarded points for start and finish position for the purpose of awarding some points for qualifying well. In some races it can really mean a lot to start near the top. And a good qualifying position shows general preparedness by team and driver. I also created a "difference" number, the difference between start and finish. This awards points for starting badly and being able to climb up through the field; adversely, starting well and falling back subtracts points. You have to keep up with the track and the driver must be able to communicate what he needs to his crew. Not to mention keeping the advertising in the sun and the wheels on the pavement.

Here's something else I did: I gave 5 points each for a lead lap finish and for running at the finish. I guess one might call those points for competitiveness and endurance. And finally the kicker: I gave big points for percentage of laps led. Several races have just been dominated by a driver but bad luck prevented him from closing the deal. Pit decisions, fuel issues, being wrecked from behind after a restart, an engine gone south. It doesn't matter with my system--if you dominate a race you are rewarded.

That's the long and short of it. There is one thing to keep in mind when you see the race scores: it's not based on 100 (although 100 is an excellent score). If you were to grade race scores on a bell curve anything above 75 is a very good score. 75 is not an "gimme". In fact, the average score for a race is 46.7. The standard deviation is huge (28.4) which means that a typical race score may range from around 18 to roughly 75. That equates, as a rule, to a finish between 9th and 35th. Which gives you an idea of how much luck and 'randomness' are involved. Contenders in this years chase average from 63.94 to 77.38 for all of the races. By the way, using my scoring system Kasey Kahne would be out and Tony Stewart in. Sorry, that's just the way it is. See the section on Kasey in the Contenders article to see why five wins doesn't mean you automatically deserve a place in the top ten.

One thing that science and the corporation can't overcome is randomness. Randomness and humaness. Sometimes the best indicators are subjective relics--such as our eyes. But since we need logic so badly we use stats. Every now and then a stat hints at a truth. But don't take it too seriously--a stat should be a toy and not a tool. Wrenches are mechanics' tools. Grab a wrench. Wheels are driver tools. Grab a wheel. Stats are fools' tools. Grab a reality. And when you've stabilized visit the Contenders & Pretenders link on the sidebar to your left. More information than you ever wanted is now available.