There are a lot of golf stats out there – A LOT. The PGA Tour publishes 458 different statistics, some dating back to 1980! That doesn’t include the PGA Tour’s Shotlink data, which tracks the conditions and distances of every shot through the green measured to the nearest yard and every putt to the nearest inch!
No doubt there are numerous insights to be derived from all this data, but the massive number of statistics can be overwhelming. Without a clear approach to evaluating statistics, it is hard to discern the signal from the noise (and with a large number of variables, even traditionally valid statistical techniques can lead to false correlations, see this excellent piece from FiveThirtyEight). A player looking to improve her performance must determine which statistics to track and discern how those numbers should inform her practice regimen or on-course strategy.
My last several articles have focused on optimizing strategy when faced with a choice between conservative and aggressive play. This post begins a new focus on the major statistics in golf and how they can be used to shoot lower scores. However, before we can consider which statistics are important for improving performance, we must first consider what improved performance looks like.
What is the goal?
There are several reasons for gathering and using statistics. The PGA Tour and its broadcast partners use stats for their entertainment value. Gamblers and Fantasy Sports players are interested in how statistics give them a betting edge. I am more interested in statistics for their use in improving player performance.
It may obvious, but an important place to start when devising a performance improvement plan is to ask, “What does improved performance look like?” The answer to this question may be different for different types of players and may imply different statistics for measuring performance.
For Tour players, some potential goals may be:
- Win more tournaments
- Win more majors
- Earn more money / Keep Tour card
- Make Ryder Cup / President’s Cup team
For high performing amateurs, the goals may be:
- Win more tournaments
- Qualify for Local/National Events
- Make school team / Walker Cup / PGA or LPGA Tour
For average golfers, the goals may be:
- Win weekly casual matches
- Win club event (net or gross)
- Have a lower handicap
- Shoot lower scores
Some of these goals are related, while some are specific to a player’s current abilities. In some cases improvement will achieve multiple goals at once. Regardless of skill level, clearly determining the goal is more likely to lead to a plan that actually works.
Choosing a performance metric
After determining the goal, the next step is to find a performance metric that is best suited to measure progress toward that goal. For some cases the proper metric will be obvious – for others measuring progress is hard. Generally speaking, the chosen metric should have these features:
- Closely related to performance goals: Analytics are more effective the more closely the underlying goal can be measured.
- Shows intermediate progress: It will typically take time before a goal is met – there needs to be a way to tell whether the prescribed practice regimen or game plan is productive.
- Provides lots of data: Small sample sizes are often misleading.
- Easy to record or measure: Practically speaking, if the metric is time consuming to track, it will not be regularly used.
Let’s consider a few performance metrics and how well they satisfy these qualities:
- Handicap: If your goal is a lower handicap, this is a great performance metric to use. Your handicap can be updated after each round, providing feedback every time you play. It can be tracked over time to measure improvement, and it is automatically calculated at your club.
- Wins: Winning is a potential goal for players of all levels, and tracking wins is the most direct way to measure success. The challenge in measuring wins is that, unlike other sports, winning is quite rare in golf. Only one player in a field of 156 wins each week on tour, and many players go years without a win. If your goal is to simply win your weekly head to head match against a buddy, you probably win at a higher rate, in which case wins may be a good metric. For most players though, wins do not provide a lot of data or show intermediate progress between wins, and thus may not be as useful.
- Tour Earnings: This applies only to Tour players, but Tour Earnings is a good metric to use even if the performance goal is not just to make the most money. Because every player making the cut gets a paycheck, earnings are updated much more frequently than wins. Furthermore, the size of the paycheck relates to a player’s finish, which provides more information than a binary win/loss record. Finally, earnings are directly related to Ryder Cup points and to staying on Tour, so earnings can be useful in measuring progress toward those goals as well.
There are plenty of other metrics a player might use, such as top 10 finishes, birdies made, or rounds in the 70s. However, there is one metric that is used more than any other in to measure performance, which I will discuss below in detail.
Does a lower scoring average mean improved performance?
In most cases, a claim about a particular statistic’s significance involves evaluating its effect on scoring average. Scoring average is usually considered the benchmark for evaluating performance. It is a relatively intuitive statistic, and it makes sense that if the goal for any one round is to shoot the lowest score possible, then a lower scoring average is a good way to achieve that goal.
I do not disagree with using scoring average as a measure of performance, but I think more care needs to be made in recognizing that in many cases a lower scoring average is not the end goal. It is often correlated with the end goal, but scoring average by itself does not tell the whole story. An example of this discrepancy is illustrated below:
Handicap Index
Handicaps are a good example of how scoring average is a related but not identical measure of performance. As noted above, many average players measure improved performance by a lower handicap. However, there are several differences between computing a scoring average and computing a handicap index.
First, a USGA handicap index uses not the scores themselves but what are called the differentials, a number representing roughly the difference between player’s scores and what a scratch player is expected to shoot (a number that is different for each course and tee). Using the differentials attempts to account for the fact that a 90 at Pebble Beach is different than a 90 at your home course.
Second, the handicap index is computed using the best 10 differentials among a player’s 20 most recent rounds. This is because the index is supposed to be a measure of potential performance, not average performance (there are a few more technical details on how the exact number is determined, you can read more about it here).
These differences in computing scoring average vs computing handicap index mean that there is not a one-to-one correspondence between the two statistics. Consider two imaginary players, Dave and Steve, whose most recent scores and differentials are shown below, with the top 10 differentials in bold.
| Dave’s Scores | Dave’s Differentials | Steve’s Scores | Steve’s Differentials |
|---|---|---|---|
| 81 | 10.1 | 82 | 11.6 |
| 85 | 16.2 | 86 | 17.2 |
| 87 | 17.8 | 90 | 19.4 |
| 86 | 13.8 | 88 | 17.4 |
| 83 | 13.3 | 93 | 25.3 |
| 85 | 15.9 | 85 | 17.8 |
| 82 | 10.7 | 92 | 19.5 |
| 89 | 19.4 | 91 | 18.9 |
| 93 | 21.1 | 81 | 12.1 |
| 92 | 19.7 | 87 | 18.1 |
| 95 | 20.3 | 83 | 15.5 |
| 86 | 15.0 | 92 | 22.8 |
| 92 | 18.5 | 80 | 7.9 |
| 93 | 21.1 | 91 | 24.1 |
| 80 | 8.7 | 95 | 25.1 |
| 84 | 12.5 | 89 | 19.3 |
| 82 | 10.2 | 84 | 14.6 |
| 81 | 10.1 | 82 | 10.2 |
| 97 | 29.1 | 79 | 9.8 |
| 85 | 15.9 | 88 | 17.4 |
These two players’ scoring averages are exactly the same (86.9), but Dave’s handicap index is 11.5 while Steve’s is 12.8 – over a full stroke worse! Because the handicap is computed using only the 10 best differentials, Dave’s 97 does not affect his handicap index, but it does raise his overall scoring average. Additionally, Dave’s differentials are in general lower than Steve’s, suggesting that Dave’s rounds were played on tougher courses, lowering his handicap further.
The example above illustrates that handicap is related to scoring average, but it is not a perfect correlation. If either of these players wanted to lower his index by 5 strokes, he would almost certainly lower his scoring average as well. But if the goal is only a slightly lower handicap, it may not necessarily be accomplished via a lower scoring average.
In defense of scoring average
While scoring average is not a perfect measure of performance, it is used so frequently because nearly all the performance goals above are highly correlated with a lower scoring average. This is especially true for large scale improvement goals: it would be insane to see a 70s player with an average over 100, or a multi-major winner with a scoring average higher than a high school player.
This is similar to a concept in physics called length scale analysis: certain effects are significant only at large scales (such as the gravitational forces between planets) while others only matter at small scales (quantum mechanical interactions of electrons). In physics, it is important to understand when each effect is significant in order to explain the mechanisms behind a behavior. Similarly, in golf scoring average is a particularly good measure of performance when considering large-scale goals. For small scale improvements, there may be better metrics for judging performance.
To summarize, let us consider how well scoring average meets the criteria enumerated above:
- Closely related to performance goals: Scoring average is more effective when the performance goals involve a large change in scores.
- Shows intermediate progress: Scoring average is very good here, as each round changes the average. One can also compute a local average based on the last 10, 20 rounds to evaluate how the average is changing.
- Provides lots of data: Uses every round played, so the sample size can be large if the rounds played is large
- Easy to record or measure: One of the easiest.
Scoring average is a great performance metric for a lot of goals. However, one must remember that a low scoring average itself is usually not the goal, and in some circumstances scoring average is less useful.
Applying it
If you attempting to achieve a new golf goal, consider what success looks like to you. Try to find a metric that will help you evaluate your progress toward your goal. Sometimes your goal will have an obvious and straightforward metric, such as tracking your handicap index. If your goal is harder to measure, like tournament wins, you may have to use a related statistic that is highly correlated with your goal. Scoring average is often a good tool to use as a default, but note that it may not correlate perfectly with your goals, especially if your goal involves only a marginal change.
Once you have settled on a metric for evaluating progress toward your goal, you can start investigating how different components of your game affect your metric. In future posts, I will discuss some of these statistics and what they can tell you about how to achieve your goals.