Golf Analytics

How Golfers Win

Don’t Trust a Hot Putter

I have written a lot about predicting putting performance. I’ve found that 1st round putting performance provides no information to predict 2nd round putting performance. I’ve found that players who improve their putting at the beginning of a season tend to retain little of that improvement going forward. I’ve found that putting improvements driven by making more long putts tend to disappear the following season because making long putts is largely a matter of luck. I’ve found that putting performance over a quarter of the season provides almost no information on how well a player will putt in the next quarter of a season. All of these studies indicate that putting performance is highly affected by luck.

What I have not done is test the influence of very recent putting on one week of putting performance. If one believed that putting came in streaks – perhaps someone got “hot” or “in the groove” for a few weeks before returning to normal – you would expect very recent performance (prior week or prior month) to have a strong impact on putting performance in the current week’s event. As I’ll show, that is not the case. My results over several different methodologies again show that short-term putting is driven mostly by luck and that recent putting performance provides almost no information on how well a player will putt going forward.

Methodology
I collected the following data from 2011-2015 for all qualifying PGA Tour golfers: 1. their average putting performance in strokes gained versus the field over the previous three seasons (2008-2010 for 2011, 2009-2011 for 2012, etc.), 2. their putting performance in the month prior to every event they played, 3. their putting performance in the week prior to every event they played, and 4. their putting performance in each event they played. I ignored majors and Tour events outside the US for which strokes gained data was not available.

Once I had matched that data for every qualifying golfer for each tournament they played, I discarded any data points for which the player had not played the week before. These two samples of performance (played prior week and did not play prior week) were not different in a statistical sense. Each data point contains putting data from 1. prior seasons, 2. the prior month, and 3. the prior week.

Influence of the prior week
I had nearly 3100 pairs of tournaments to compare. Just comparing performance in the prior week to the current week yielded almost no predictive value (R^2 = .006). For every stroke better than the field a player putted in the prior week, they’re expected to putt 0.08 strokes better in the next week.

prior week to current SGP

Influence of the prior month
Using the same 3100 pairs of tournaments and comparing performance in the prior month to the current week again yielded almost no predictive value (R^2 = .006). For every stroke better than the field a player putted in the prior month, they’re expected to putt 0.09 strokes better in the next week.

prior month to current SGP

Influence of the prior three seasons
Using the same process and comparing performance over the previous three seasons to the current week yielded more predictive value (R^2 = .057). For every stroke better than the field a player putted over the previous three seasons, they’re expected to putt 0.91 strokes better in the next week.

priorseasons to current SGP

Combined model
Including all three pieces of data and predicting the current week resulted in the following equation:

Putting estimate = (0.02 * Prior week) + (0.04 * Prior month) + (0.86 * Prior seasons)

This indicates that when attempting to predict putting performance, how a player has putted in the previous few seasons provides about fourteen times more information than how they have been putting recently.

What this means for this week

Putting performance is almost entirely driven by how well a player putts in general, rather than any hot or cold streak in recent weeks. Looking at the field this week, Nick Watney sticks out as a guy whose poor putting has cost him recently. Of players with at least 6 rounds of putting data in 2015, Watney is 2nd to last despite being generally a fairly average putter by the numbers. Poor putting likely cost him at least a spot in the playoff at Torrey Pines on Sunday.

Kevin Streelman’s putter has also been cold to start the season – he’s putted about half a stroke worse than expected over his first four events. He’s wasted three strong tee to green performances at Kapalua, Waialae, and Scottsdale by putting poorly, and managed only a T30 as his best finish there.

Advertisements

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s

%d bloggers like this: