Welcome to the third chapter of Finding Frankenstein, a PGA experimental series on FantasyLabs.
Finding Frankenstein is where I leave all my opinions and narratives at the door and rely only on what the tools at FantasyLabs tell me. First, I experiment with the unreal amount of data by using our Trends tool to run trends for literally every possible metric specific to rounds played at that week’s course. Next, I use those results to construct a tournament-specific model in our Player Models tool. Finally, once the model is built, I take my experimentation even further, trying to create a monster player capable of taking down a tournament.
The Zurich Classic of New Orleans Recap
Last week’s Frankenstein was Jamie Lovemark, who won the tournament after 53 holes. Unfortunately, he needed to be the winner after 54 holes — 56 if you want to get technical. Brian Stuard outlasted Lovemark on the second playoff hole, after eliminating Byeong Hun-An on the first playoff hole.
Lovemark finished T2 at fifteen under par with 90.5 DraftKings points. He was owned at around 14 percent in all the large tournaments. With the chalky Justin Rose missing the cut, Lovemark ended up being a great tournament play. Admittedly, the Mad Scientist in me was slightly disappointed Frankenstein didn’t notch his first victory, but it was a great week nonetheless.
In hindsight, the Zurich Model was built very well. Eleven of the twenty-five golfers in consideration for last week’s Frankenstein Title finished inside the top twenty, with six inside the top ten, including Lovemark and An (who were both in the playoff). Seventy-two percent of the Frankenstein-eligible golfers made the cut and played the weekend.
Of course, I was hesitant to weight Putts per Round too heavily despite its strong showing in the Trends tool, and then Stuard went out and won the tourney putting his brains out.
Enough of the trip down memory lane. It’s time to build another monster. We’re moving onto the Wells Fargo Championship at Quail Hollow Club in Charlotte, NC.
Creating the Frankenstein Model for the Wells Fargo Championship
A total of sixty trends were created and evaluated for this week’s Wells Fargo Frankenstein Model. Henrik Stenson matched 43 of them. But that doesn’t necessarily mean Stenson is going to end up being Frankenstein. Some of the trends demonstrate a negative Plus/Minus, and you don’t want your golfers to match those. So first we need to eliminate the bad golfers and then accentuate the positive ones.
If somebody told you that you should target golfers with a history of excellent scrambling ability at Quail Hollow, the data is calling them a liar! Long-Term Scrambling was not profitable over any sample size. In fact, the harder I pushed towards the elite end of scramblers, the worse the results became. Needless to say, it will not be used in the construction of the Frankenstein Model.
In analyzing the results this week, my biggest take away was that, although there are many profitable trends over large sample sizes, only a few trends are more consistent than a coin flip. When the data is narrowed down to the elite end of performance at Quail Hollow, a lot of the trends do become more profitable and incredibly consistent.
Since Long-Term Greens in Regulation (GIR) was the most consistent over a large sample size and increased steadily in Consistency towards the top end, it will be weighted most in this week’s model.
Recent Adjusted Round Score showed reliably gainful results over multiple variations of sample sizes and got freakishly appetizing at the top end so that will be next up in the weighting.
If you plan to build your own Frankenstein model this week, all the other sliders included in order of weighting are as follows: Long-Term Eagle, Recent GIR, Long-Term Driving Distance, Long-Term Birdie, Long-Term AdjRd, Long-Term Field, Long-Term Missed Cut, Course AdjRd and Recent Driving Accuracy.
My findings show that although the narrative this week is that Quail Hollow is a “bomber’s paradise” (which to some extent the data supports), the data demonstrates that it doesn’t need to be the most heavily-weighted metric, hopefully differentiating our Frankenstein from the herd.
Creation of the Monster
Per usual, the top-tier golfers must be removed from Frankenstein consideration, so any golfer priced $9,500 or higher is out this week. We can’t have a beautiful stud be Frankenstein. We are trying to create a hideous wretch worthy of catapulting your tournament team to the top of the leaderboard.
Last year’s Quail Hollow winner Rory McIlroy, possibly still hung-over Rickie Fowler, the very handsome Adam Scott, the cut-making-machine Henrik Stenson, and the chalky Justin Rose, Patrick Reed, and Phil Mickelson have all been priced out of consideration for this week’s title.
It’s worth noting that McIlroy and Stenson were rated Nos. 1 and 2 in this week’s Wells Fargo model. The 25 remaining Frankenstein options this week were:
Not considered in our model was the number of Pro Trends. The data proved that golfers with at least four Pro Trends were much more dependable and incredibly rewarding when compared to golfers with fewer than four. This is where we will begin our creation of the monster:
The minimum Pro Trends filter eliminated only three golfers, one of whom was everybody’s favorite DFS dude, Will Wilcox. That’s two weeks in a row in which Wilcox was in consideration for Frankenstein but was eliminated early. Sorry, Will, but I have to trust the data.
Recent Par-5 performance could not be considered in the model, but the data has shown that the high-scoring golfers at Quail Hollow excel in this metric. Eight more golfers were wiped out based on their Par-5 abilities, and we were down to fourteen eligible golfers.
Recent Adjusted Round Score was weighted heavily in our model, but I want to dive deeper into the results at this point, because it returned the highest Plus/Minus of any metric and had top-five Consistency. This week’s Frankenstein needs to be on the elite end of this metric, so that’s next up in the creation of the beast!
Seven additional golfers were eliminated at this point, including last week’s Frankenstein, Jamie Lovemark.
As stated above, Long-Term GIR was weighted most heavily in our model because of its Consistency over a large sample size. Recent GIR was considered in our model but not weighted quite as heavily. However, it returned a Plus/Minus of +10.25 and Consistency of 60.6 percent on the elite end of the sample. This is where we will continue to trim the fat from our monster, to be sure that he’s in elite playing shape for Quail Hollow.
At this point of formation, Charles Howell III, Gary Woodland and Hideki Matsuyama were all eliminated. Honestly, the human in me was secretly rooting for Matsuyama to become Frankenstein this week, but the Mad Scientist must rely on the data!
We are down to the Wells Fargo Championship Frankenstein Final Four:
Eagles are an integral part of high scores on DraftKings. Elite Long-Term Eagle Score also proved to be an incredibly consistent and profitable trend at Quail Hollow, so this is where we will continue our creation. Some of the past results matching this trend include last year’s winner, Rory McIlroy, and his redonkulous 137 DK points.
Frankenstein Revealed
Of the four remaining qualified golfers, only J.B. Holmes matched the Long-Term Elite Eagle Score trend. Holmes is our Frankenstein for the Wells Fargo Championship. He is the hideous wretch whom the data has deemed most suitable for taking down a tournament this weekend.
Holmes won this tournament in 2014, but missed the cut last year. In fact, he has missed the cut 42 percent of the time at this tournament, adding some frightening variance to our monster.
So load up a Frankenstein lineup, and see what kind of monstrous results you can achieve. Feel free to berate me on Twitter throughout the tournament if Frankenstein doesn’t play up to your standards.
Good luck this weekend.
