Single Seater will be "forecasting" each race of the 2016 IndyCar season, with win probabilities posted to the site after qualifying is over. These will be up either late Saturday or early Sunday morning -- if a driver scratches from a race, an additional update will be posted.
How It Works
Our model is a simple one, only taking into account three variables. I chose a simple model for two main reasons:
- It's easier to update and keep track of. There aren't 20 different variables I have to gather together to make a prediction, so I can easily post them for each race.
- There is so much randomness in racing. Crashes, mechanical failures, caution flags -- these can all have a drastic affect on the race, and they're almost impossible to predict. Instead of trying to with a ton of variables, I just skip over it altogether. Also, it helps to distinguish the signal from the noise -- data that isn't very predictive. I've chosen variables that are historically very predictive of race success, and that's it.
These variables are added together for each driver and a winning percentage is determined.
I've done research before on qualifying position and how it relates to race win probability. These same values are used as the baseline win probability for a driver. These values are different for road/street courses and ovals, as the probability of winning from different places is different depending on the type of track.
The number of wins a driver has over the past three years is the next variable. If a driver has won 20 percent of the races the last three years, you could say there's a 20 percent chance he'll win the next one.
Prior Performance At The Track
How a driver has fared at a track in the past is also included. If a driver has won two of the last three races at a track, most of the it means he's pretty good there. This variable rewards drivers who have a "good" track and knocks down drivers who have never had a good result there.
That's it. You can find the race predictions for each 2016 race here.