[Editor’s Note: We’re just as surprised as you are that we’re publishing NASCAR DFS advice, but we figure that anything sitting at the intersection of sports and numbers is fair game. We hope to expand our coverage of growing DFS sports going forward. Until we figure out what we’re doing with these sports we’re going to offer the content for free.]
RotoDoc holds a Ph.D. in mathematics and tackles sports data to give you a competitive edge in your fantasy games.
The daily fantasy sports industry has continued its booming growth, both in volume and breadth of content. Recently, DraftKings has introduced daily fantasy NASCAR to their portfolio to give daily fantasy nuts like myself another route to degeneracy. While in its infancy, there’s a significant edge to be found in daily fantasy NASCAR. I was able to exploit this edge in just the second and third weeks of DFS NASCAR, taking down the $30k High-Line GPP then followed that up a week later by placing all four of my entries in the $25k High-Line GPP. So how am I exploiting this edge? With data, of course!
NASCAR Prediction Model
DraftKings uses four scoring metrics in DFS NASCAR:
- Finishing position (43 points for first, down to 1 point for 43rd, with a 3 point bonus for first place)
- Place differential (starting position – finishing position)
- Laps Led
- Fastest Laps
As is my norm, I created a prediction model to predict these metrics for each driver and for every race. Just like my DFS NFL model, I used a machine learning technique called a random forest to train and test my model against historical NASCAR data. Data I used included:
- Loop data, which includes things like Driver Rating, average running position, and quality passes
- Did not finish data
- Team and manufacturer data
- Qualifying results
- Practice results
- Driver age
- Track information
I will create an article on each of these items in the near future describing the significance of these metrics, but suffice to say they all play a role in predicting NASCAR results.
So who are the plays this week when NASCAR pays a visit to the Tricky Triangle?
Top Plays at Pocono
Kevin Harvick ($11,700): Kevin Harvick is my model’s top play this week, projecting him at 4.99 points per dollar. Harvick is projected to lead the most laps, and accumulate the most fastest laps as well. This is driven mainly by his starting position, his 97.0 driver rating over his last eight Pocono races, and his 126.3 driver rating over his last eight flat track races. Compare this to the guy starting ahead of him, Kyle Busch, who has a 86.2 and 95.5 driver rating respectively, and it’s likely Harvick dominates the early portions of the race. Look for him to be there at the finish too.
Kurt Busch ($10,100): Kurt is a sneaky solid play this week. His second best average driver rating at Pocono, behind only Dale Earnhardt Jr. combined with a consistently fast car (no worse than 10th in any practice or qualifying session), means he should be in contention for the win.
Brad Keselowski ($9400): Keselowski is projected at 4.48 points per dollar, which mainly stems from an expected finishing position of 10.4 according to the model. This also places him 4th among expected finishing position, behind only Harvick (8.2), Joey Logano (9.2) and Kurt Busch (9.8). The other two start farther forward than Keselowski, meaning Keselowski doesn’t suffer as much as the other two in the place differential category.
Kyle Larson ($8300): Kyle Larson’s starting position of 27th provides an opportunity as he should make up a significant number of positions. Considering he has the 10th best average driver rating over the last eight Pocono races, as well as the 14th best combined practice time, a top 15 finish seems quite reasonable for Larson.
Casey Mears and Ricky Stenhouse Jr. ($6500 and $6400): Either of these two drivers make for decent cheap plays mainly due to the fact that their rank in practice was higher than their qualifying rank. Mears practiced 20th fast in happy hour and Stenhouse 27th fast, while they qualified 26th and 34th respectively. Each of these two can make up 6-10 spots for a nice place differential score to add on to their race finish to make 4x value.
Avoid if Possible
Martin Truex Jr. ($10,300): It might seem odd recommending avoiding the man who won the previous race at Pocono this year, but consider this. Truex may be highly owned simply because of his prior win, and his car has not been as fast this week as it was back in June. He was 9th quickest in happy hour, 11th quickest in combined practice, and qualified 13th best. He’s not likely to lead many laps with only 6 laps led projected. To make 4x value at this price point, he’ll need a top 8 finish which is doable. But if you want him to win you a GPP, he’ll need a top 3 and a significant amount of laps led, which is unlikely. He’s a borderline cash game only play.
Tony Stewart ($7900): Stewart has been poor this year, and don’t expect his season to suddenly turn around here. A fifth place starting spot just lends itself to him tumbling down the order and giving you a significant negative value in the point differential category. He ranks only 13th best in last eight average driver rating at Pocono, his average finish this year is 22.8, and his combined practice speed was only 22nd best.
Austin Dillon ($7500): Similar to Stewart, Dillon had a fantastic qualifying, but his race stats just don’t support his starting position. His average finish of 21.7, his 20th best average driver rating at Pocono, are at least tempered somewhat by an 8th best combined practice speed. There’s still too much risk here, so Dillon is only a contrarian GPP play if using multiple entries.
|Martin Truex Jr.||10300||11.1||6.0||7.2||32.9||1.5||3.6||1.9||40.0|
|Dale Earnhardt Jr.||9200||14.3||6.2||7.6||29.7||1.5||3.8||0.7||35.8|
|Ricky Stenhouse Jr.||6400||26.7||0.0||0.5||17.3||0.0||0.2||7.3||24.8|
|Sam Hornish Jr.||6400||27.1||0.0||0.2||16.9||0.0||0.1||2.9||20.0|