When I was a young boy, I used to love conducting my own experiments. Once I actually set a large portion of a cow pasture on fire during one of those experiments. Don’t worry – no cows or crab-apple trees were harmed in the blaze, but the big red fire truck had to come and extinguish the flames. Now that I’m a grown-ass adult with children of my own, I prefer to conduct my experiments in a less volatile environment – FantasyLabs. I figure it’s probably not a great idea to teach your kids that accidentally setting animal farms ablaze is fun.
Seriously, that’s why I love FantasyLabs. It allows idiots like me the ability to make fully customizable models with an insane amount of data (something I could never hope to do left to my own devices), giving me the opportunity to experiment under a much safer, and often times more profitable environment. If you’re looking for opinions on who to play this week, trust me – you can find a myriad of well-thought-out opinions from various PGA touts who are way smarter than me and probably never set a field on fire, so I won’t pretend to know more than they do and won’t give you any of my opinions. Instead I will rely on the data, and see what hideous creature I can create from it.
I’m hard at work in the FantasyLabs, analyzing trends to try and create the perfect model for the Valero Texas Open, and in turn find a monster player to help me take down a tournament. I’m trying to find Frankenstein, if you will. Finding Frankenstein is all about experimenting with the massive amount of data available in FantasyLabs to create a model and then find a player that might not be getting the love he deserves. A player that may look like a hideous wretch on the outside, but actually has the ability to launch your team to the top of the leaderboards.
My first step for finding Frankenstein in the FantasyLab was to run trends for every possible Long Term and Recent Form filter possible specific to rounds played at TPC San Antonio. My goal was to identify what data I should weigh most heavily in my Frankenstein Model. Although I ran trends for every filter possible (literally), my concentration for this first step was on “Score” filters, because these are the sliders available for us to weigh in the FantasyLabs Models.
Going into this experiment I had some ideas on what I thought would or should get weighed heavily in the Frankenstein model. I assumed Recent Greens in Regulation (GIR) would be a huge part of it, but apparently, I really am a dope and it’s becoming clearer by the minute how I managed to set a field on fire with G.I. Joe action figures. The top 70-plus scores of Recent GIR played at San Antonio don’t translate to a positive Plus/Minus at all.
Because of the tight and narrow fairways that are often talked about for this week’s track, Recent Driving Accuracy was something else I assumed would be weighed heavily in my Frankenstein model. And, I was wrong again. This is why I said I wouldn’t give you my opinions and would rely on the data.
I was consistently surprised with the results of the trends I ran (and I did run all of them), specific to rounds played at TPC San Antonio. Nothing came back as terribly consistent over larger sample sizes, so I took the trends that were both most profitable in terms of Plus/Minus and consistency and narrowed them down to smaller sample sizes to see if golfers at the top range were indeed more profitable more consistently, and they were. The Frankenstein model will be built most heavily around Long-Term Birdie Score.
Next up in the Frankenstein Model weighting will be Long-Term Adjusted Round Score. Colin Davy’s somewhat controversial (at least on Twitter) stat is something that I love using in models because: A) Colin Davy is so much smarter than me, and B) it’s something only FantasyLabs users have access to. So if we have access to something created by a really smart dude that other PGA Experts aren’t considering, it allows us to find intelligent, yet contrarian value. This is how I can be contrarian without being stupid. And we’ve already established I’m pretty good at being stupid.
I won’t bore you with each and every specific trend result I decided to weigh in the Frankenstein Model, but know that Recent Driving Distance, Long-Term Missed Cut Score, Recent Adjusted Round Score, Pro Trends, and, because this is a tournament model, upside were all part of the creation of our monster model.
Once the model was built, it was onto finding Frankenstein. And let me mention that Frankenstein shouldn’t be a Branden Grace, Zach Johnson, Matt Kuchar, or Patrick Reed, although they are all somewhat externally hideous in their own endearing way. Frankenstein can’t be a player you would roster without even knowing what a model was. No friends, we are looking for a hideous wretch to separate our teams from the crowd and take down a tournament, so for the purpose of this experiment I am removing all players priced above $10,000 from consideration.
J.B. Holmes, Jimmy Walker, and Brooks Koepka actually ended up being three of the top four highest-rated players in the Frankenstein Valero Model. And both Patrick Reed and Branden Grace also showed up in the top eight, but they, along with all other 10k-plus golfers, aren’t the type of hideous wretch we’re attempting to find. So let us look at what we had left to work with.
In order to find Frankenstein, I continued to rely upon the data to narrow down our choice of monsters. In my model building research I noticed that players with at least five Pro Trends at San Antonio performed considerably better than those who had four or less. So this is where we will start our deconstruction of the beast.
Pro Trends are another exclusive FantasyLabs feature that non-users don’t have access to. So again, it gives us the opportunity to build a unique monster, one who can be both contrarian and smart because it’s driven by data-based research and back-tested results. The Pro Trends filter wiped out 15 of our 25 options and left us with 10 golfers in the running to be this week’s Frankenstein. We were forced to say goodbye to some interesting names like Patrick Rodgers, Jhonattan Vegas, Francesco Molinari, and Kevin Chappell, who has two top-10 finishes in his last three starts.
Long-Term Par 4 Score was another profitable filter at San Antonio that we couldn’t weight specifically in our model, so we’re going to look there next to in order to continue construction of this wretched beast!
Players who had both a Long-Term and Recent Adjusted Round score under 70 also showed up with a profitable Plus/Minus in my research so I went there next, seeing if we could complete construction of our monster. The young man with the fancy hat and the really repulsive Brendan Steele, who won this event in 2011, weren’t up to task, and we were left with only Luke Donald and Billy Horschel.
One thing we know about TPC San Antonio is the winning golfer will go pretty low, so birdies will be a good thing for us. The last five winners of the event have an average score of 10 under par. My next step in creating our monster was to look for players who have had big birdie numbers both recently and over the long-term. According to my trends research, 13 seemed to be the magic number for positive results, so I created a Trend for golfers who averaged more than 13 birdies both recently and long-term and it proved to be an extremely profitable trend.
We’re left with only one remaining qualified match and we’ve finally found Frankenstein! Billy Horschel is our Frankenstein for the Valero Texas Open. He is the hideous wretch that the data has deemed most suitable for taking down a tournament or two this weekend. Horschel has finished third here in two of the last three years, but he also missed the cut in 2014. So why not load up a Frankenstein lineup or two, and see what kind of monstrous results you can achieve … just try not to set anything on fire when you do.
