FireKeepers Casino 400 Projections

Michigan International Speedway (Sunday, June 12 2016)

Click image to zoom.

Welcome to FantasyDriverAnalytics! The projections shown above are based upon a model that uses the DraftKings scoring algorithm and historical data to provide guidance for this week’s race. Some have claimed that scoring projections for this fantasy sport are not helpful due to the number of variables that can impact a driver’s performance, but I disagree. Analytics can help a player make better decisions based upon all of the available data.

The order of drivers on the y-axis is based upon the DraftKings scoring algorithm’s place differential, laps led, and finishing position components. While I would love to include fastest lap data as well, I have not identified a good source of historical data to include in my model. The values plotted on the x-axis in the box-and-whisker plot should be interpreted as follows: the box shows the range of finishing positions that would fall within the variance explained by the model. The whiskers extend to include each driver’s variance that is not explained by this week’s model, but present in the historical data.

Having the benefit of analyzing my own results of past races privately, I will share a few thoughts and observations on the projections above. Most significantly, Kyle Busch may outperform the model’s expected DraftKings points. He has struggled recently, and he has struggled at Michigan, so it is unsurprising that he is ranked as poorly as he is. However, the model also shows some significant potential upside for Kyle. The model would explain a top-10 finish and his unexplained variance may put him in contention for a top-3 finish. He’s not a driver to rule out at any track, though these statistics also show that he is a risky pick this week.

Whereas Kyle Busch may appear to some to be ranked too poorly, some may think that Clint Bowyer is ranked too strongly. However, 3 out of Clint’s last 4 starts at Michigan have resulted in top-10 finishes. He has also benefitted from stronger finishes recently following his season-worst 38th place finish at Texas a couple months ago. These results help boostClint’s overall ranking, however the model shows plenty of potential downside for Clint as well. Just as Kyle Busch may be a risky pick this week, Clint Bowyer may also be a risk.

The statistics that appear to most strongly predict Kyle Busch’s finish this week include his starting position, long-term performance at Michigan, recent consistency of finish, recent consistency of laps led, and recent laps led. For more detailed statistics about Kyle Busch, click here. Clint Bowyer’s strongest predictors appear to be his starting position, consistency and number of laps led at Michigan, and his recent average finishing position. Click here for more detailed statistics about Clint Bowyer.

I don’t subscribe to Accupredict, so I can’t tell you how their full lineup compares to these projections, but their teaser shows the following top 5 finishers predicted for tomorrow’s race:

  1. Truex
  2. Logano
  3. Edwards
  4. Keselowski
  5. Harvick

Stay tuned to hear how well this model fares against Accupredict’s numbers! Projections will be published each week, followed up by analysis of modeled results vs. actual results. My goal for now is just to generate some discussion.

If you have questions, ideas, or commentary on the model, feel free to post them here. Let me know what you think!