Last week, we looked at the general shape of FanDuel’s golf scoring system and how that might influence roster construction. This week, we’ll move from the general shape to the specific shape as we discuss the concept of first-half versus second-half golfers and how they differ from a practical standpoint.
First- and Second-Half Correlation
Let’s take the sample projections from last week’s tournament and make a simple plot: Each player’s projected first-half FD points versus his second-half FD points. In general, we expect a linear-ish relationship. Let’s see how it looks:
The general relationship is about what we expect. It’s the golfers who stray far from the diagonal line who interest me. Here’s the quick way to interpret this graph: The more a golfer is below the diagonal line, the better he is in the first two rounds relative to average. The more he’s above the line, the better he is in the later rounds.
In my opinion, finding value in FD’s new scoring system will hinge on the ability to pick out cheap players who score strong in the first half. The second-half cohort will often consist of stars, and rightfully so, since they offer the best chance of making the cut, so the optimal way to build and distinguish lineups (via our Player Models) will be to roster first-half players who excel relative to the field and their ownership, the latter of which you can study once contests have started via our new DFS Ownership Dashboard.
What Do First-Half Golfers Look Like?
First of all, is the concept of a “first-half golfer” even valid? What does that player look like, and is there any basis for believing that first-half overperformance is a repeatable, predictable skill? Anecdotally, we can think of a couple players who always seem to come out of the gate firing and then fade down the weekend. (Brendan Steele and Ryan Palmer among others come to mind.) There are all sorts of hypothetical reasons why some players might be better in the first half: Physical exhaustion, inability to stay focused over multi-day stretches, etc. There’s probably enough here at least to warrant that we take seriously the concept of first-half golfers.
In addition, we’ve actually already covered a data-driven rationale for such a player previously: The cheap cut maker. Basically, the first-half golfer is the cheap cut maker. Here’s the simple version of why they’re the same: In order to make the cut, a golfer must play well in the first half. All else being equal, the cheap cut makers DFS players have used on DraftKings are likely to be the strong first-half golfers they roster on FD.
The Difference in Process
Even though strong first-half players tend to be cheap cut makers, the research process on FD is likely to be different than on DK for one simple reason: Round-by-round results are really annoying to research unless you have that data already aggregated. (And, yes, of course we do at FantasyLabs.)
For example: How often do you use the quick shorthand for checking a player’s results (T5, T33, etc.)? That shorthand just doesn’t exist for most results for rounds 1-2 and 3-4 only. Additionally, if you try to figure out what’s predictive of performance in each rounds, there’s a weird twist: You’re using the same aggregated stats (long-term adjusted round score, strokes gained, etc.) to predict two different quantities — first-half and second-half performance. So far, no one’s written the guide on how those weights might change depending on the section of a golf tournament under consideration. It hasn’t been done because no one’s needed to do it — until now.
So that’s exactly what we’re going to do in the next article: We’ll look at the weights we have in our Player Models and consider how to optimize them for first-half vs. second-half performance.
