How predictable are PGA Tour players?

MAMARONECK, NEW YORK - SEPTEMBER 16: Dustin Johnson of the United States plays a shot as Justin Thomas of the United States and Claude Harmon III looks on during a practice round prior to the 120th U.S. Open Championship on September 16, 2020 at Winged Foot Golf Club in Mamaroneck, New York. (Photo by Jamie Squire/Getty Images)
MAMARONECK, NEW YORK - SEPTEMBER 16: Dustin Johnson of the United States plays a shot as Justin Thomas of the United States and Claude Harmon III looks on during a practice round prior to the 120th U.S. Open Championship on September 16, 2020 at Winged Foot Golf Club in Mamaroneck, New York. (Photo by Jamie Squire/Getty Images) /

Even the best players on the PGA Tour do not perform consistently from week to week.

You can check out the jump-off to the reliability conversation in part one of this series. Between 2016 and 2020, 75 players completed enough rounds each season to qualify to be ranked in the numerous statistical profiles on the PGA Tour. How many of those 75 produced scores that fit within a two standard deviation range of fewer than 10 strokes?

Such a player would be a figure to watch, especially to those interested in the predictive side of wagering. That player would be even more interesting if he combined a small range of scored predictability with a low scoring average.

That player would be the best of both worlds: one combining talent with reliability.

More from Pro Golf Now

Unfortunately for those who like a friendly wager on a Tour event, those players do not presently exist.

Not a single one of the 75 players whose performances comprised the 2016-20 study had a single-digit two standard deviation range of predictability. The most predictable player during that time – with a two standard deviation range of 10.84 strokes – was Aaron Baddeley. And his 70.7 scoring average was nearly a half stroke above the group average of 70.32.

That means the most predictable thing about Baddeley was his average-ness. That’s not valuable knowledge.

Of the 10 players with the narrowest performance range, not one also ranked among the group’s top 10 in scoring average. Coming closest to meeting that goal was Paul Casey who combined a 3.04 reliability rating – 10th best – with a 69.86 stroke average, 11th best.

To a surprising degree, putting your money on a player with a low scoring average also meant accepting relatively high volatility.

There are no clearer examples of this than three recent major champions at the top of the scoring list, Dustin Johnson, Justin Thomas, and Brooks Koepka.

Johnson ranks No. 1 in scoring average for the period at 69.34. But the 3.33 stroke standard deviation of his performance range surpasses 13 strokes, from a low of 62.68 to a high of 76.0. In fact, on 22 occasions since 2016 Thomson has shot 75 or worse.

It’s the same story with Thomas, whose 69.42 scoring average ranks behind only Johnson.

But the first standard deviation — the reliability rating — of Thomas’ round-to-round performance was 3.31 strokes, about a tenth of a stroke higher than the group average. Extended across a second standard deviation – a range that gives us a 97 percent probability of inclusion – that means all we can say with certainty about Thomas from round to round is that his score is likely to fall between 62.8 and 76.0 strokes. That’s a 13.2 stroke range. That’s not much to bet on.

Koepka, a four-time major winner, can be even more frustrating to trust from round to round, much less from tournament to tournament.  His 69.85 average was 10th over the span of that five-season period.  But his 3.35 reliability rating ballooned his two standard deviation range to 13.4 strokes. The low end of that range, 63.15, is mighty sweet. The high end, 76.55, will get you cut from any field on the continent.

And you’ll never know from one PGA Tour round to another which version of Brooks Koepka will show up.

As previously noted, this performance spread is not unique to Johnson, Thomas, and Koepka; it is essentially a shared condition on tour. And as a general proposition, the relationship between a player’s average score and the reliability of his actually producing that score is wafer thin…less than 7 percent, according to correlation theory.

For the record, here’s the data on the 10 players included in the five-season study who produced the lowest scoring averages over that period. Each line includes the player’s average score, his reliability rating – that is, his one standard deviation spread – plus the 97 percent confidence spread – low, high, and width of the range – for each player.

Rank      Player                                  Avg.      Std. D.     2SD L      2sd H     2sd R

1              Dustin Johnson                69.34     3.33        62.68     76.00     13.32

2.             Justin Thomas                  69.42     3.31        62.80     76.04     13.24

3.             Webb Simpson                 69.47     3.20        63.07     75.87     12.80

4.             Rory McIlroy                      69.50     3.21        63.08     75.92     12.84

5.             Jordan Spieth                    69.67     3.15        63.37     75.97     12.60

6.             Rickie Fowler                    69.69     3.16        63.37     76.01     12.64

7.             Hideki Matsuyama          69.69     3.29        63.11     76.27     13.16

8.             Tony Finau                         69.74     3.08        63.58     75.90     12.32

9.             Justin Rose                        69.83     3.18        63.47     76.19     12.72

10           Brooks Koepka                  69.85     3.35        63.15     76.55     13.40

If I were a betting man, that data might suggest to me that the safest bet from week to week is Finau. He’s not as likely as Johnson, Thomas, or a few others to go way low, but his modest 3.08 reliability rating – 17th best among the group of 75 — suggests he’s also not as likely to throw in a major clunker. That’s also true of Casey.

Casey’s 69.86 stroke scoring average combined with his 3.04 reliability rating puts his 97 percent confidence range between 63.78 and 75.94.

The problem is the wafer-thin relationship between reliability and performance. The 10 players included in the study with the smallest reliability range were Aaron Baddeley, Matt Jones, Bill Haas, Vaughn Taylor, Nick Taylor, Ryan Moore, Matt Kuchar. Peter Malnati, Chris Stroud, and Casey.

As a group, those most predictable players competed in 1,164 tour events between the 2016 and 2020 seasons. They earned a fraction over $79 million and racked up a credible 152 top 10s.

But they only actually won a total of 10 times in those 1,164 starts. That’s a winning percentage below 1 percent.

What those who wager on golf are looking for is the next Tiger Woods. The problem is that nobody has combined a low scoring average with a high solid level of reliability since Woods.

At his performance peak – between 1998 and 2002 – Woods played 471 official rounds in Tour-recognized events. (Between 2016 and 2020, Patrick Reed set the standard with 486 such rounds.) Woods’ stroke average over the course of those 471 tournaments was 69.09; that’s about a quarter stroke per round superior to Dustin Johnson, the 2016-20 leader, and about a stroke and a quarter below the average for the 75 players included in the study.

He combined that with a 3.08 reliability range, a number that would rank 17th in the modern game and comfortably inside the 3.22 group average.

Next. Golf Tips: Top Five Guys Golf Destinations to Consider. dark

Could any current PGA Tour player re-create Woods’ combination of low average and reliability? That’s a topic for a third installment. Check back tomorrow for an answer.