Threads: Evaluating The Single Seater’s Elo Ratings and Predicting DNFs

I wrote two Twitter threads over the past week or so that I wanted to share here in case anyone doesn’t follow on Twitter or missed them.

The first is a thread on evaluating how well The Single Seater’s Elo projection model did this year. I looked at the percentage of matchups it correctly predicted as well as a calibration plot of the actual vs. projected results.

I've had some time to look at how our Elo ratings (https://t.co/IOsb7B1nP1) compared to the actual results of matchups and races this season, so here's a short thread with some initial findings.

— Single Seater (@thesingleseater) November 30, 2020

The second thread looked at whether or not DNFs in IndyCar are really as unpredictable as people say they are. I looked at this from both the season and race level and found some interesting results.

Whether or not a driver will DNF during a race is typically thought of as a random process – there are lots of moving parts and drivers often get caught up in accidents through no fault of their own. But is it a *truly* random process, or can we predict when drivers will DNF?

— Single Seater (@thesingleseater) December 4, 2020

I might dig a little deeper into these topics in the future and write a full post, but if anyone has anything particular they would like expanded on please let me know in the comments here!

Header Photo: Chris Owens/IndyCar