• Analytics Blog
Sept 20th, 2018
A golfer's "strokes-gained" is defined as the difference between their score and the field's average score on a given day. What we call a golfer's "true strokes-gained" adjusts these values to take into account the strength of a field. In our opinion, true strokes-gained is the best summary of golfer performance: it rates a given performance based on the number of strokes better or worse it is than what we would expect from some benchmark golfer (we will use the average PGA TOUR player as the benchmark throughout this article).

On the PGA TOUR, golfers are not compensated with money, or FedExCup points, according to their true strokes-gained (nor should they be). Instead, money earned is a function of a player's finishing position and the purse of the particular event. This article explores the relationship between true strokes-gained and earnings; that is, the relationship between performance and compensation.

To start, we analyze earnings at the event level. We are interested in the extent to which the same performance over 4 rounds is rewarded differently at different events on the PGA Tour. What should we expect? Clearly the World Golf Championships (WGC) and other no-cut events (e.g. the Tournament of Champions, the TOUR Championship) reward poor performances more than other events. You or I could show up and, as long as we didn't run out of balls (no small task), we would take home a sizeable cheque for finishing in last place. But what about earnings associated with high-quality performances (e.g. > +1 true SG per round)? The answer is not obvious because tournaments with higher quality fields tend to offer larger purses, making it harder to finish in the top spots while also paying more for doing so. Below we show the overall relationship between true strokes-gained (in units of strokes per round) and tournament earnings (left panel), as well as expected earnings as a function of true strokes-gained for a few different types of events (right panel):

Notes: Data includes all official stroke-play events on the PGA Tour from 2015-2018. Panel A plots average true strokes-gained and money earned at an event, and includes a polynomial fit. A few true SG outliers are dropped to condense the plot. Panel B shows fitted curves for three groups of events from 2015-2018: WGCs (not including the Match Play), the Honda Classic, and 'low field strength events' (we include the Barbasol, the OHL Classic at Mayakoba, and the Sanderson Farms). True SG performances above +4 are not shown in the right panel, as the relationship to earnings is uninteresting after this point (larger purse = higher predicted earnings). Only 4-round performances are included in this analysis (i.e. we don't include missed cuts).
Let's start with the right panel. As expected, WGCs have the highest expected earnings for lower true strokes-gained performances. Interestingly, it is the low field strength events that have the highest expected earnings for performances above about +1.2 true strokes-gained and below about +4 true SG (greater than +4 is not shown on the graph - this is the level of performance beyond which it's fairly likely you are winning any tournament). The Honda Classic, which should be thought of as an event with both a medium-sized purse and medium field strength, offers the lowest bang for a golfer's buck for true SG performances below +2, while the WGC events provide the lowest compensation for performances above +2 true SG and below +3.5. For those wondering, expected earnings at majors were pretty middle of the road.

Shown on the left panel is the overall relationship between money earned and true SG at a tournament. This can be thought of as answering the question "What are the expected earnings from a given true SG performance without knowing the PGA Tour event at which it was shot?". It's worth mentioning a couple extreme earnings outliers over the last few seasons:

Chris Kirk won the 2015 Crown Plaza Invitational with a true SG of +2.79 per round, netting a cool $1.2M. Conversely, Alex Noren averaged +2.75 true SG per round at the 2017 WGC-HSBC to finish in T12th, earnings just $130K. We would have expected both of these performances to yield about $330K without knowing the events at which they were shot.

Matt Jones finished in T41st at the 2015 Humana Challenge with a true SG of +1.2, earning about $20K. (The Humana seems to be an event that doesn't pay well for good, but not great, performances - this is due to the fact that the leaderboard is often very bunched.) Also in 2015, Alex Cejka was part of a 5-man playoff for the ages at the Puerto Rico Open (incredibly, 37 players finished within 5 shots of the winning score in this slugfest). Cejka ultimately won that playoff, earning $540K for his efforts, despite averaging just +1.18 true SG for the event! We would have predicted earnings of about $50K for both of these performances. The 2015 Puerto Rico Open not only had a weak field, but it was also an incredibly bunched leaderboard - this latter point is what made it possible for Cejka to take home a lot of money with a mediocre performance.

Now let's move on to some analyses at the season level. This next figure shows the relationship between a golfer's season-long earnings and their season-long average true SG. We restrict the plot to golfers who played at least 25 events (a strict cutoff, but one meant to eliminate big differences between players' earnings that are only the result of playing more or fewer events) and who had true SG performances between -1.5 and +1.5. This is done to more clearly display the part of the relationship we want to focus on.

Notes: Data includes 2015-2017 PGA Tour seasons. Plotted are a player's total earnings against their season-long true strokes-gained average (in units of strokes per round). Only seasons in which at least 25 events were played are included. We also restrict the plot to true SG performances between -1.5 and +1.5; in actuality, season-long true SG varied from about -3 to +3 in our sample.
The relationship between season-long true strokes-gained and earnings is exponential, due to the highly nonlinear payout structure of PGA Tour events. We've highlighted a few notable players from 2017 - Xander Schauffele and Si Woo Kim both earned much more than expected given their season-long true SG, while Stewart Cink earned much less. The reason for this is obvious: Schauffele and Kim both won big events (The TOUR Championship and THE PLAYERS, respectively) while otherwise playing poorly (in the case of Kim) and just mediocre (Schauffele) in 2017. Cink had no very high-end finishes, but was consistent, racking up 10 top-25 finishes. Also noted on the graph are three members of the 2018 Web.com graduating class; using their season-long true SG averages in 2018, we've indicated what their expected earnings would have been on the PGA Tour. The leader of the regular season Web.com money list (with $534K), Sungjae Im, would have been expected to earn about $1.6M on the PGA Tour in 2018. Now there is some food for thought: Im was compensated about 1/3 of his PGA Tour counterparts for what we would call equivalent "work".

To further examine the interplay between performance and compensation on the PGA and Web.com tours, we next look at what kind of season-long performance is required to gain membership on the PGA Tour. As a reminder: the top 125 on the regular season FedExCup points list keep their cards on the PGA Tour, while just the top 25 on the season-long money list on the Web.com tour get upgraded to the Big Boy tour. This next figure visualizes what it takes to gain full status on the PGA Tour:

Notes: Data includes PGA Tour and Web.com seasons from 2014-2018. True SG averages are calculated from all regular season stroke-play events (up to start of FedExCup playoffs on PGA Tour, up to start of Web.com Finals on Web.com tour); "rank" is a player's regular season FedExCup rank on the PGA Tour, and money list rank on the Web.com tour. Plotted data points include all seasons in which the golfer played at least 60 rounds on the PGA Tour, and 50 rounds on the Web.com tour. Fitted curves are higher-order polynomials.
This figure plots a golfer's rank on the regular season points/money list as a function of season-long true SG on the Web.com and PGA tours. The results are interesting: you may need to squint to see it, but the fitted curves shows it is slightly easier to gain full status on the PGA Tour through the Web.com circuit than it is to keep that status once you are on the PGA Tour. A season-long true SG of about -0.35 is expected to yield a money list rank of 25 on the Web.com, while a -0.2 true SG average is expected to be 125th on the PGA Tour. This is a very small difference and is quite surprising; we thought there would be a bigger gap here. (It's worth mentioning that we did have issues fitting this data, so the gap could be slightly larger than what is implied by the fitted curves.) Given this result, why does it seem so common for Web.com grads to fail to retain their cards in the following season on the PGA Tour? One possible explanation is that many Web.com grads require outlier seasons to graduate to the PGA Tour. Then, in the following season, they experience some regression to their mean performance level, which ends up not being good enough to keep a PGA Tour card.

Phew! We have actually covered a lot of ground in this blog, and plan to revisit some of these topics in the future in more detail. Let's wrap things up with a few takeaway messages: 1) There is a lot of variance in earnings at PGA Tour events for what we would deem equivalent performances (i.e. the same true strokes-gained average over 4 rounds). This means that where a player happens to play well (which is basically luck) can have important effects on their season-long compensation. 2) For very good performances (greater than +1.5 but less than +4 true SG per round), low field strength events offer the highest expected earnings. 3) At the annual level, there is also a lot of variance in earnings conditional on averaging the same true SG throughout the year. This is explained mostly by the fact it's much better to have a few great performances than it is to be consistently good throughout the season (due to the top-heavy payout structure of PGA Tour events). 4) Web.com players are compensated much less for equivalent performances to their PGA Tour peers. 5) It is slightly harder to maintain a PGA Tour card by finishing in the top 125 on the regular season FedExCup points list than it is to graduate to the PGA Tour by finishing inside the top 25 on Web.com money list.