Long-Term Metrics vs. Recent Form
Recent form is definitely a divisive topic — in any DFS sport, really, but especially PGA. Because the courses change each week, it is often necessary to create a brand new Player Model to fit both that course and that specific tournament. But one thing that always comes up is this: How much do you weight long-term versus recent metrics?
During the Open Championship a couple weeks ago, I found that DraftKings salaries are less correlated with recent Adjusted Round Score (Adj Rd Score) than with Long-Term Adj Rd Score — a unique metric that measures average strokes per round and adjusts for course difficulty and field strength. It’s the best proxy for talent or recent form in PGA DFS. This information doesn’t necessarily mean that long-term metrics are more important than recent ones, but DK salaries do have a very high correlation with Vegas odds to win the tournament, so by a weird transitive property we can infer that Vegas prefers long-term metrics . . . and Vegas tends not to lose money.
However, because of the lower correlation between DK salaries and recent metrics, that means that there likely might be more inefficiencies to be found by exploring the metric. Or put another way: If DK priced players perfectly according to their recent play, there would be no edge or value to be gained by using that statistic, since simply looking at a player’s price would give you an indication of recent play. Since that’s not the case, we can perhaps spot some edges to be found by looking at recent play and more specifically the difference between a golfer’s recent and long-term play.
To do this, I simply looked at important statistics found in our Player Models — Adj Rd Score, Greens in Regulation (GIR), Driving Distance (DD), Driving Accuracy (DA), Putts Per Round (PPR), and Scrambling percentage (SC) — and measured the difference between long-term and recent data. Here’s what I found:
Here’s an example in case you’re unsure of how to interpret the table: Rickie Fowler has a +0.5 difference between his long-term and recent Adj Rd Score. That’s not great, but note that his elevated recent mark of 69.4 is the same as Scott Piercy’s, who has seen a -0.7 difference in his Adj Rd Scores. Fowler has a worse mark, but his baseline was much different.
Some Analysis
There’s way too much data to hit on every golfer, but I will highlight a couple intriguing points. Hideki Matsuyama has seen an incredible +2.5 increase in Adjusted Round Score — a truly miserable mark. Now there are two paths of thought: Do Matsuyama’s bad recent marks indicate impending regression, as BABIP does in MLB? Or do the marks indicate that struggles will likely continue? I lean toward the latter because of his other stats: He hasn’t been playing badly because of struggles with putting or other historical issues. He’s not hitting a higher percentage of GIR, perhaps a more disturbing problem.
Also, let’s look at Brooks Koepka, who is battling an ankle injury this week and had to withdraw from the Open Championship just two weeks ago as a result. This is a pretty glaring issue when you look at his data: He has lost 21.8 yards of distance on his drives. In this week’s PGA Fantasy Flex podcast, PGA Director Colin Davy said that DD is very important at Baltusrol GC, the site of this week’s tournament. In that case, if Koepka’s injury is stopping him from getting the distance necessary to compete at this course, then it’s very hard to make a case to have any exposure to him.
There are other intriguing pieces of data here. Dustin Johnson is a golfing god right now. He has seen positive upticks in every single category, and some of those upticks (like the one in DD) have been quite obscene. Thomas Pieters has also seen a huge difference in his recent Adj Rd Score. The data suggests that he’s in great recent form. As a distance guy who already potentially fits Baltusrol, he’s an intriguing tournament play at only $7,000.
Enjoy combing through the data and good luck this week. It’s the last Major of the year. Make it count!
