NASCAR DFS Strategy and Driver Sim Scores – Las Vegas
Before we dive into NASCAR DFS strategy for Las Vegas, including the new sim scores I’m introducing, let’s recap Atlanta.
Last weekend at Atlanta was a bit of a rough weekend for me in DFS. I went underweight on Kevin Harvick in GPPs, which obviously hurt considering he dominated most of the race. However, I was significantly overweight on Brad Keselowski, Matt Kenseth, Kasey Kahne, and Chase Elliott to help me claw my way back to a loss of only $1.7k, when things could have been far worse.
That hit still leaves me with just under $30k in profit so far in 2017.
Enough of last weekend, let’s dive into the Las Vegas slate strategy. I’m also bringing back my driver similarity scores. For a primer on those, you can read the in-depth version of how they work for NFL in our free articles describing them here and here. As a quick description, let’s say you want to see how Russell Wilson might perform against the Arizona Cardinals. We can take Wilson’s stats, and compare them to past and present players with similar stats and physical attributes to Wilson. We then find from these players which ones faces defenses that had similar stats to Arizona’s current defense, and bam, we have Game Level Similarity Projections. We take the top 20 player/defense comparables, and find the low (25th percentile), median (50th percentile), and high (75th percentile) fantasy output for that group of 20. This then becomes Wilson’s low, median, and high projection, allowing the low and high to be 25th, and 75th percentile projections.
So how do we do this for NASCAR? Instead of using defenses, the NASCAR version of the app uses tracks. This let’s us compare how drivers with similar stats performed at similar tracks to a chosen driver and track. For example, if we want to see how drivers historically similar to Kyle Busch have performed at tracks historically similar to Las Vegas, we can do that. Here are Busch’s comps:
In the 20 comps, eight of them are from Las Vegas races in the last three years, while 12 of the comps are from other 1.5-mile ovals from drivers that were similar statistically at that time to Kyle Busch’s current stats. The stats include:
- Driver Rating
- Laps Led
- Fastest Laps
for each of the following:
- All Tracks
- Current track
- Similar tracks
The number of races for the history will be sliders that the user can set when we complete the app. For now, I’ve used 18, eight, and eight as the number of races for all, current, and similar tracks. Note, when I say current track, I mean the current track that the historic driver raced at. So for Kyle Busch we are looking at his last eight Vegas races, but for the comp Tony Stewart at Atlanta in 2008, we’re looking at Stewart’s eight most recent Atlanta races prior to that race in 2008.
Additionally, there is also finishing position baked in over the last 18 races, which helps identify drivers that might finish better or worse than their driver rating indicates. For the track data, I’ve used the following:
- Length (in miles)
- Banking in the turns
- Banking on the straights
- Asphalt or Concrete surface
- Track surface age
- Track type (i.e. road course, restrictor plate, large oval, etc.)
For the surface age, Las Vegas was last paved in 2006, making it 11 years old. Atlanta was repaved in 1997, meaning it was 11 years old in 2008, which is why it is an apt comparison to have Tony Stewart’s 2008 Atlanta as a comp. Stewart had very similar stats in 2008 to present day Kyle Busch, and the tracks were similar in surface age and length.
Let’s dive into the Las Vegas Driver Sim Scores, and then talk about how to use them for NASCAR DFS strategy for Las Vegas.