StatCar Racing: Bristol

August 22, 2008 2:10pm CDT

When in doubt, pick Kyle Busch. The Joe Gibbs racing dynamo who won the first Car of Tomorrow race at Bristol last spring comes in atop this week’s rankings.

Review of Last Week

Correlation between pre-race ranking and driver finish position was 48 percent, fairly strong considering several traditionally strong drivers like Jeff Gordon and Denny Hamlin suffered crashes or engine problems which relegated them to terrible finishes. Carl Edwards delivered on his top two ranking with a dominating victory and David Ragan and Brian Vickers provided fantasy owners with the high finishes that we were expecting.

Comparing StatCar to the Fanball Cheat Sheets

To provide some perspective on the performance of the StatCar model with other rankings available, the Dan Beaver and I decided to compare the performance of the StatCar rankings with the his cheat sheets. Performance was compared using the average difference between driver pre-race ranking and driver finish position.

Last week the StatCar model had an average place differential of 8.8 compared to 10.0 for the Fanball Cheat Sheets. The Cheat Sheets, however, had more picks that had a difference of one position or less, meaning when they were accurate, they were dead on.

Interestingly, for the season the results are reversed. The Cheat Sheets have a lower average difference at 9.7 compared to 10.6 for the StatCar model, while the StatCar had 12 percent of selections that had a difference of one position or less compared to 10 percent for the Cheat Sheets.

We will continue to track for the rest of the season, but for right now, it is probably too close to call. To make it more interesting, Dan and I have a five dollar weekly bet on which of our top ranked drivers will finish higher. Because Carl Edwards beat out Jimmie Johnson last week, he’s up a half a sawbuck on me. Hopefully we will get back to even this week as my Kyle Busch should beat his Greg Biffle.

This Week’s Rankings

There were four variables that were predictive of driver finish position at last year’s Sharpie 500: pass differential in past races at Bristol, percent of laps driven in the top 15 in the Ford City 500 earlier this season at Bristol, finish position in the Crown Royal 400 at Richmond, and laps led in the Best Buy 400.
Taken together, these variables indicate that how a driver has recently performed on short tracks or high banked concrete tracks is predictive of how they will fare this weekend. The variables and their impact coefficients are below.

Variable, Impact Coefficient
Constant, 26.0
Career Pass Differential at Bristol, -0.12
Percent Laps Top 15 - Spring Bristol, -0.09
Percent Laps Led – Dover, -0.37
Finish – Richmond, 0.19

The formula for predicting rankings’ headliner Kyle Busch is therefore:

Expected Finish = (1)(26.0) + (95)(-.12) + (-.09)(51.8) + (-.37)(39.5) + (.19)(2) = -4.3

How can Kyle Busch have a negative expected finish? Remember that the StatCar model computes how on average several variables impacted driver finish position in last year’s race. Because Busch has had such a remarkable season in all of these variable areas this year, his expected finish is statistically “off the charts.” Expect him to run up front this week.

And don’t forget, I have put my money where my pen is.

Rankings