ParFax: Are There Truly Horses for Courses?

James Sherill runs the Rexford Buzzsaw, a site that looks at second-level golf statistics.  In an effort to bring on the Golden Age of golf Sabrmetrics, I've invited James to do occasional golf stats posts for us here at Waggle Room.

Common sense holds that, in golf, there are "horses for courses." Those golfers that for whatever reason seem to own one certain venue. Take, Tiger Woods, who has won 7 times at Firestone Country Club’s South Course. It shouldn’t be too hard to look around the golf world and find everyone and their uncle telling you how Tiger owns Firestone. But, does he really? Does this magical "horses for courses" phenomenon exist?

Luckily, this week, the WGC-Bridgestone event will be played on a course that under various names has hosted an elite-fielded stroke play event since 1976. That gives us a great sample to judge if there is such a thing as horses for courses.

I went back and recorded every event dating back to 1977. I then gave a player a standardized score based on the field average, as I do in my rankings. Then I compared historical performance as well as number of rounds on Firestone’s South Course with results. In other words, for the 1999 tournament I compared performance and rounds from the 1976-1998 tournaments with 1999 and so on, all the way up to 2009.

Here are the graphs of what I found:

And here are the r and r-squared values:

What this tells us is there is a small correlation between playing well in the past at Firestone and number of rounds played (it’s negative because more rounds played leads to lower scores) with results at firestone.

However, this does not mean course form is worth anything. It’s quite possible that historical performance at Firestone is linked with results simply the best players tend to play well over time. In other words, there appears to be a correlation, but it could simply be good players playing well because they are better. Furthermore, better players, since entry into Firestone’s event is linked with performance around the world, tend to play more rounds at Firestone because they are better. So, in other words any experience related correlation could just be that players that are better would have more rounds at Firestone, because they are better.

So, I decided to look a little deeper. I went back to 2004, as far back as I could go and still get a 2-year ranking from my rankings, and looked at a few more comparisons, starting with 2-year average compared to results. Here is what I found:

An r-value of around .4 means there is a moderate correlation between overall skill and performance at Firestone since 2004 and quite similar to the r-value I got for overall skill when I looked at This means that there is a much higher correlation between overall skill and performance than course history. However, there is still a decent amount of randomness over four rounds of golf.

Next, I broke down the course history even further. I compared the historical performance at Firestone and number of rounds played to the results relative to a players’ two-year average. So, if a players score is an average of -1.35 standard deviations per round better than the field and the players overall average is -.35, I now treat it as -1, instead of -1.35 like in the above comparisons. The graphs below show play in relation to average compared to all-time historical performance and number of rounds played:

Now, there is almost no correlation. This means most of what I observed above from historical performances was largely based on the relationship between good players tending to play better over time. A player’s history at Firestone had about the same correlation as one random tournament.

*You might also notice the y-intercept comes at .30. That’s because on average this field is much stronger than the average PGA Tour player, which my rankings are adjusted to. The average player over the past 5 years in this event is around .28. That means a first time player at Firestone can be expected to on average be about .02 standard deviations or around .05 strokes per round worse than the average. The guys with the most rounds can expect to be around .2 strokes per round better than average.

So, when you’re trying to fill out your fantasy team, how much should you put into performance at Firestone. Well, not that much. There is almost no correlation between performance(or rounds played) on this course to results. If anything, simply playing rounds at Firestone is probably just as important as playing well at Firestone. I used the equation from the rounds played at Firestone vs. performance in relation to average graph to change the odds for this week.

Here’s the top-25 based on just 2-year performance(based on rankings through British Open):

And the top-25 with an adjustment for rounds played at Firestone:

As you can see that roughly .25 stroke per round difference between a Firestone rookie and doesn’t translate to much overall, partly because most players in this field have played here before. The biggest winners from adjusting for rounds played (even though it is small) are the best players in the world over the past 10-15 years. Obviously, they have been good enough over a long period of time to play in this event a lot.

This is only one event. It’s possible that while course history at Firestone has almost no meaning, that’s only at Firestone. However, anything that I’ve ever looked at on this issue seems to be that course form is entirely overrated. Sure, if you cherry pick Tiger Woods it looks like there are "horses for courses." But, taking one outlier and trying to make a rule based on that is ridiculous. When you judge the whole field over a much bigger sample size, course form seems to have very little effect.