Home | Drivers | Owners | Tracks | Sprint Cup | Xfinity | CWTS | KNPSW | KNPSE | IndyCar | ARCA | F1 | Tudor | Random
Comments on this blog (2) (moderated)
Equation Racing writes:
"The Big 8 at Darlington"
Posted by Ol Don on May 7, 2011
Viewed 621 times


For those not familiar with Equation Racing, I'll give you a brief introduction before I dive into what I'm calling The Big 8 at Darlington.

Equation Racing is a system that I built using five years of historical, loop-level data from the Sprint Cup Series (provided by the kind folks at racing-reference!). The system is essentially a statistical model that looks at key variables that can help predict winners, such as driver ratings, recent performance, qualifying position, and so on. The model also considers how well drivers do by track type -- short, cookie cutter, road. It even considers how well drivers do by track shape (D-shape, Tri-oval, etc) and track design (length of straightaway, angle of turns, etc).

I run the model against each race track because the variables that predict performance change quite a bit depending on the track. Some tracks are very predictable and others not at all. I also run the model between the upcoming race and the most recent previous race, because driver momentum is such a key input.

All of this leads to a ranking of drivers for the upcoming race. I usually run the model before qualifying to get a sense of who should perform well and often update it after qualifying to adjust for those running particularly well that week.

It's kind of a geeky way to look at racing. But if you think about it, racing generates tons of data. For instance, let's just pick Richmond International Raceway since the Series was just there. Each driver (that finishes on the lead lap) generates 400 laps of data each race. Multiply that by 2 races per year (which they run at Richmond) and by 5 years of historical data that I have, and suddenly for each driver you have 4,000 data points at RIR alone! That's a lot to use and is a huge help when predicting winners.

So... with all that build up, who does the model like tonight at Darlington? There seems to be two top tiers (4 drivers in Tier 1 and 4 in Tier 2) and then the rest of the drivers fall way back in the model ratings. I should say that this is not necessarily how I personally would rate these drivers this week, but subscribing to data-based decisions is my motto, so I'll go with what the model says.

Here are the projections for the top tiers:

Tier 1
Jeff Gordon
Denny Hamlin
Jimmie Johnson
Greg Biffle

Tier 2
Ryan Newman
Carl Edwards
Kyle Busch
Kasey Kahne

If betting were legal, I would put money that the winner comes from one of these 8 drivers. And I'd probably parlay that wager by betting that the top 3 comes from these drivers, as well. And I'd probably lose those winnings on the back jack tables immediately after. But that's just hypothetical talk.

Alas, time to post this or no one will believe that it was written before the actual race.

Good luck to all who are picking drivers in office pools, fantasy leagues and in Vegas (Nevada residents only).

Signing off,
Ol Don

Opinions expressed in blogs are those of the individual bloggers and do not necessarily represent the views of racing-reference.info.