Introduction

This article will lay out some data behind Spectrum, what it is and what makes it valuable as a DFS Tool and betting tool for identifying "true" performance skill and undervalued players. First let's take this article from the approach of the current state Spectrum Data which is all of the Strokes Gained categories. Spectrum has the opportunity of revising nearly every statistic, many of which are in the works and will be in production soon, but for brevity let's focus on the Strokes Gained categories. Before we dive into it, let's do a refresher on Strokes Gained.

Strokes Gained Basics

Strokes Gained is a broadly used term in the PGA gambling space, but it refers to any of the categories of stats as shown below. SG:Total is essentially the scoring average relative to the field (in reverse), for example if a player shoots -2 on a round when the field's scoring average was -1, then their SG:Total would be +1. Mark Broadie (the inventor of Strokes Gained) took this breakout a step further with the addition of SG:Putting and SG:T2G. Basically this allowed PGA Tour fans to see how a player performed in their SG:Total value, broken out between Putting and everything else Tee-to-green. Since then, the PGA Tour further broke out SG:T2G but separating SG:OTT (Off the tee), SG:APP (Approach) and SG:ARG (Around the green). This is where the first issue arose in Strokes Gained metrics and the way they are used today.

Strokes Gained Math

The math for Strokes Gained is quite simple, however. Each shot has a PGA Tour baseline. Let's use putting for our first example, referencing the SG:PUTT chart from below:

So let's make an example, the player hit their approach shot to 30 feet. They have 2.00 stroke expectation here. So for their next shot, they are a bit too aggressive and run it 5 ft by the hole. The expected strokes gained from 5 ft. is 1.15, so for this single 30 ft. putt, the player's SG:PUTT calculation is: 2.0 (expectation from 30ft.) - 1.15 (expectation from 5 ft.) - 1 (the stroke they took) = -0.15

So for this example, the player lost 0.15 strokes putting. They then make their 5 ft. putt, which is then: 1.15 (expectation from 5 ft.) - 0 (ball is holed, no expectation remaining) - 1 ( the stroke they took) = +0.15

So on this 30 ft. putt the player lost 0.15 strokes on their first putt then gained 0.15 on their second putt, exactly matching the 2 stroke expectation from 30 ft. leading to a SG:PUTT metric of 0.00.

The same math applies to other SG categories, for example SG:OTT. The player begins on the tee for a 450 yard hole with the expectation of 4.1 strokes to complete the hole. The player then hits a 300 yard drive into the fairway and has 150 yards remaining. The benchmark from 150 yards is 2.9 strokes, so the math is as follows for SG:OTT on this drive: 4.1 (benchmark from the tee box) - 2.9 (benchmark for 150 yards remaining) - 1 (the stroke they took) = +0.2

So this player gained 0.2 strokes OTT on this single tee shot. Strokes Gained adds some more complexity as each hole and round is then adjusted for the field. For example, perhaps this hole plays straight downhill, so hitting it 300 yards is quite easy and arguably short. This player may not have gained anything in the perspective of this hole against the field.

But that is the basics of Strokes Gained. Simply calculating remaining expectation to hole out from their original expectation of their shot.

Strokes Gained Issues

Strokes Gained is an aggregate statistic when viewed at the round level and event level. This means, that the round value for a player's SG:OTT is the sum of all tee shots they hit. Why is this bad? Well first, let me say Strokes Gained tells a great story about a player's performance. Understanding why they shot 74, because they lost 3 strokes off-the-tee is very helpful for analyzing a players round. But this does not help us in regards to predicting their next round's expectation for our purposes of DFS or betting.

Why? Because golf is riddled with outliers, and the concept of penalties for golfers are largely unfairly weighted. For example, a player can play 17 holes (13 tee shots) and have gained exactly 0.1 strokes on every single tee shot. Currently their Strokes Gained Off-the-tee is +1.3 as they hed to hole 18. Pretty strong performance so far. Unfortunately they hit their tee shot into the left hazard, leading to a penalty drop and a loss of -1.5 strokes for that tee shot. Now they have significantly shifted their SG:OTT from +1.3 to -0.2...MASSIVE! 

While positive outliers for tee-shots are rare, they are very common for Approach shots, Around the green and even Putting, although negative outliers are equally likely. Spectrum helps calculate each player's historical skill-level to determine outliers. For example, Nick Dunlap losing 0.8 strokes off-the-tee may not be an outlier, whereas for Bryson DeChambeau losing 0.3 strokes OTT that could be identified as an outlier.

Aggregation is the biggest enemy for us when it comes to researching and analyzing players true skill. Simply adding up all shots and assuming that is their future expectation is foolish, and frankly dangerous to our decisions.

Additional concerns with strokes gained is their generalized locations for benchmarking. We know not all shots are created equal, but the PGA Tour benchmarks assume that to be true. We know that the left rough on #10 at TPC Sawgrass is much worse than the right rough due to the angle. So this further leads to skewed values that 2 players who hit it in the rough the same distance from the pin should have the same benchmark. That simply isn't the case and golf data is riddled with these scenarios. Others include discrepancies in lies (sitting down vs. sitting up), behind a tree or open shot, pin location and wind and weather.

In short, Strokes Gained does a fantastic job of explaining why Tiger Woods gained 1.5 strokes during his round but Matt Kuchar lose 0.5 strokes. You can see exactly where the difference is in that single round performance. The problem arises when you take a closer look at the many variables inherent in golf, the various penalty scaling, and the aggregation that one single poor shot (or poor decision) can shift a player's data 1 or 2 shots.

Spectrum Data

This is where Spectrum comes in, and the exact reason why it was created back in 2019. Spectrum data completely cleans every player's round to identify and recalculate outliers, both positive and negative. Let's take a look at some sample data to help you understand the deviation in a typical player's round.

This player had a reasonable event at the Travelers Championship until a debacle on Hole 13 where they lost 6.1 SG:OTT on that single hole. With SG, this all gets aggregated the same. Spectrum Identified this and instead of a round that shows the player lost 7 strokes OTT, Spectrum shows they actually lost only 0.4 SG:OTT. This would otherwise not even be noticeable. But what happened next?

The very next week this player finshed T22 while gaining over 6 strokes Tee-to-green. 

Let's take a look at another example this one is really fun: