Wednesday, October 17, 2012

Analyzing Miss Bias on Tour

[image]

One of the questions to ponder when it comes to game improvement is the impact of a ‘1-way Miss’ in your game. Considering this lately, I decided to look at this from a statistical analytics standpoint. When 2012 Pro Golf Synopsis comes out, each PGA Tour player analysis will denote the golfer’s miss tendency.

But the question still remains ‘how important is it?’

Thankfully, the PGATour.com’s Web site has the miss tendencies of each player on Tour broken down in the shots off the tee that miss the fairway and the percentage of whether or not they miss right or left of the fairway.

With that, I would like to start off with some general analysis, looking at these metrics:

• Adjusted Par-4 Scoring Average
• Adjusted Par-5 Scoring Average
• Birdie Percentage
• Bogey Avoidance Percentage

Since we are talking about miss tendency on par-4’s and par-5’s, I do not see a reason to look at par-3 scoring average. And while the standard par-4 and par-5 scoring average is a generally a good indicator of a player’s success on Tour, an even better indicator is to adjust those scoring averages based upon the strength of the fields they play in and other factors. While the PGA Tour has done a good job of adjusting scoring average for the strength of the field, I have a more accurate adjustment.

And in 2013, I hope to have an even more accurate way of adjusting each player’s scoring average to better depict their performance and to even more accurately project future earnings and provide individual game analysis.


***


The first thing I wanted to check was to see how well players with a miss bias (left or right) did in the metrics I bullet pointed above. One can use either the right miss tendency or left miss tendency metric measured by the Tour. Here’s a link to the left miss tendency:

http://www.pgatour.com/r/stats/info/?02422

[image]

Hopefully I don’t need to point out the obvious, but if Marc Leishman misses left of the fairway 35.1% of the time, then that means his tendency on misses is to miss to the right, 64.9% of the time. Thus, I want to see if the metrics above were affected if a golfer missed left or if they missed right and if Tour golfers were better off missing left or missing right.

Here’s the correlation coefficient data I came up with for 2012 (for the statistical minds):

Adjusted Par-4 Scoring Average: -0.0184331

Adjusted Par-5 Scoring Average: 0.08249

Birdie Percentage: 0.024434

Bogey Percentage: -0.036218

For those who do not understand correlation coefficients, I’ll try and give a brief explanation. The numbers are used to help mathematically determine if there is a correlation between two conditions. The closer the number is to +1.0, the stronger the direct correlation is. The closer the number is to -1.0, the stronger the indirect correlation is. And the closer to 0, that means there is no correlation.

For example, if I own a store that sells lemonade and I run the numbers to see the relationship between lemonade sold and outdoor temperature and get a number of +0.75, that means that there is a strong correlation between temperature and lemonade sold and the hotter the temperature, the more likely I will sell more lemonade. One can then use that number to help project how much lemonade will be sold if they can get the projected temperature.

OTOH, if I am looking for the relationship between weather and hot soup and find the number to be at -0.75, that’s telling me that there is a strong indirect correlation. Meaning, the hotter the temperature the less likely I will sell hot soup. But on the other hand, the lower the temperature, the more likely I will sell hot soup (and not sell lemonade).

And if I am looking for a relationship between coffee and temperature and get a number like 0.0013, that means there is no real rhyme or reason as to when coffee sells based on temperature.

Thus, when we look at those correlation coefficients, they are all near zero which means there is no correlation between golfers favoring missing right of the fairway or missing left of the fairway and with actual scoring.

However, we start to see a small, but noticeable change when look at the correlation coefficient of miss tendency and my personally made metric, Driving Effectiveness:

+0.10747

That being said, that number still practically means there is no correlation overall to where the golfer misses and their effectiveness off the tee.

Part of my conclusion is that there is so much more involved with par-4 and par-5 scoring averages along with birdie and bogey percentages (i.e. putting, iron play, etc) that miss tendency has little to do with it.


***


The rise in the correlation coefficient between miss tendency and driving effectiveness is peculiar and thus I wanted to check it out. My initial theory is that the closer the miss tendency is to 50/50 is actually NOT good. But, being on the extreme is not good either.

I wanted to look at the miss percentage of the top-25 and bottom-25 players currently in my driving effectiveness metric.

Below are the miss tendencies of the top-25 players in my driving effectiveness metric. I have sorted the miss tendency in order to help more clearly illustrate the point.

Miss Bias
59.1% R
56.1% R
55.2% R
53.4% R
53.1% R
52.5% R
52.2% R
51.1% R
48.2% L
47.5% L
47.3% L
47.3% L
46.9% L
46.7% L
46.0% L
45.5% L
45.4% L
44.8% L
44.2% L
43.8% L
43.6% L
42.9% L
41.7% L
40.5% L
39.9% L


There are 3 things I notice here:

1. 17 of the top-25 players in driving effectiveness had a left miss bias.
2. Only 1 player had a miss bias of more than 10% (10.1%)
3. Only 4 players had a miss bias within 2.5%.

Here’s a look at the miss biases of the bottom-25 players in my Driving Effectiveness metric.

Miss Bias
65.9% R
60.9% R
60.7% R
57.2% R
57.2% R
54.8% R
53.0% R
52.6% R
52.2% R
51.4% R
51.2% R
51.1% R
47.9% L
47.8% L
46.9% L
46.7% L
46.7% L
45.7% L
45.3% L
45.3% L
45.3% L
41.2% L
40.3% L
39.1% L
36.7% L

Looking at these numbers we see:

1. 13 of the 25 had a left miss bias (versus 17 of the top-25)
2. 5 of the 25 players had a miss bias over 10%
3. 6 of the 25 players had a miss bias within 2.5%


***


Before I come to more confident conclusions, I want to see how miss tendency impacts accuracy. Since my Driving Effectiveness metric is based on distance, fairway percentage and distance to the edge of the fairway, distance may skew the real impact that a miss tendency bias will have. Thus, I created a formula to determine total accuracy using fairway percentage and distance to the edge of the fairway. However, I’ve weighted distance to the edge of the fairway to properly reflect its impact on scoring.

When I ran the correlation between Total Driving Accuracy and Miss Bias, the correlation came out to 0.00144079229411184. Again, that means there is no real rhyme or reason when we look at the Tour as a whole between miss bias and Total Driving Accuracy. And I got the same type of numbers when comparing it with Fairway Percentage as well.

So…let’s take a look at the top-25 and bottom-25 again.

Here’s the top-25 in Total Driving Accuracy Metric and their miss bias:

Ben Curtis 59.1% R
Tom Gillis 56.7% R
Graeme McDowell 53.4% R
Tim Clark 51.4% R
Ryan Moore 51.1% R
Jim Furyk 51% R
Jeff Maggert 50.6% R
Zach Johnson 50.5% R
Chez Reavie 50%
Matt Kuchar 49.7%L
Jerry Kelly 49.2%L
Colt Knost 48.6%L
Gary Christian 48%L
K.J. Choi 47.6%L
Richard H. Lee 47.5%L
Jason Dufner 47.3%L
David Toms 46.5%L
Brian Davis 46.4%L
John Huh 45.5%L
Chris DiMarco 44.3%L
Russell Knox 44%L
Hunter Mahan 43.6%L
John Mallinger 43.4%L
Heath Slocum 40.5%L
Mark Wilson 38.4%
L

What’s interesting here is that there are more players within that 52.5% bias either left or right. In the top-25 Driving Effectiveness players, there were only 4 within that range. Now there is 12 in the Total Driving Accuracy metric. But, we still see the same tendency in that most of the players miss bias is to the left.

Here’s the bottom-25 in Total Driving Accuracy and their miss bias:

Phil Mickelson 58.9% R
Gary Woodland 58.1% R
Michael Bradley 57.2% R
Troy Kelly 55.9% R
Martin Flores 55.4% R
Joe Ogilvie 54.8% R
John Daly 52.7% R
Jhonattan Vegas 52.3% R
Matt Bettencourt 51.2% R
Derek Lamely 51.1% R
Jason Day 49.1% L
J.B. Holmes 48.2% L
Danny Lee 47.9% L
Andres Romero 47.8% L
Jason Kokrak 47.5% L
Stewart Cink 46.7% L
Aaron Baddeley 46.7% L
Troy Matteson 46.7% L
James Driscoll 45.3% L
Jimmy Walker 44.9% L
Angel Cabrera 43.5% L
Mark Anderson 41.2% L
Daniel Chopra 40.3% L
Charles Howell III 38% L
Stephen Gangluff 36.7% L


Again, it’s not exactly crystal clear, but the trends start to point to the more inaccurate golfers off the tee tend to have a more extreme miss bias.


***


I think it’s safe to say that using Driving Accuracy and trying to find a trend will not provide us with very clear data to interpret. However, if we use driving distance, we start to see a clearer picture.

The data shows that the longer hitters are much closer to 50/50 miss tendency than the shorter hitters. I think that is due to the longer hitters being less sure of where a ball goes whereas the shorter hitters like a Mark Wilson have a much better idea of where their misses will end up; in part due to not being able to bomb one further than they expected.

I then decided to combine both Driving Effectiveness along with the golfer’s distance off the tee. I wanted to look at the long hitting drivers on Tour that were effective off tee and their miss percentage. Then I wanted to do the same with short hitting players on Tour who were effective off the tee.

First, I looked at a combination of:

Top-30 in Driving Distance + Top-50 in Driving Effectiveness

There are currently 17 players in this category. Their miss percentages looked like this:

52.5% L
52.2% L
51.4% L
51.1% L
51.2% R
51.8% R
52.5% R
52.7% R
53.1% R
54.6% R
55.2% R
55.8% R
57.1% R
58.3% R
59.1% R
60.1% R
60.6% R



While these players were all very effective off the tee as they are all in the top-50 in Driving Effectiveness, the players that had a greater miss bias were lower ranked than the players with a miss bias closer to 50/50. And again, notice how the tendency is to miss right of the fairway.

Here’s a look at those players in the top-50 in Driving Effectiveness, but a shorter hitter on Tour. For the purposes of this post, I will use any player that is ranked 115th to 179th in driving distance on Tour.

53.4% L
51.4% L
51.1% L
51% L
50.6% L
50.5% L
50.3% R
50.8% R
52% R
52.5% R
56% R
59.5% R


Here we see a very tight aim bias, with most of the players being very close to 50/50.


***


[image]

So, what can we draw from this?

The more accurate and effective drivers on Tour tend to have more of a right miss bias.

Analysis: This may be due to the fact that since most of the players on Tour are right handed, that the swing mechanics and impact conditions that tend to cause a miss left like:

a) Closed Clubface
b) Toe-hit
c) Outside-to-in path

Are more conducive to a player who will hit poor driver shots than a player who misses the fairway to the right, which is most likely a pushed shot where the mechanics and face angle are only slightly off.


Generally speaking, the extreme miss bias golfers (8% or more) tend to be less effective off the tee.

Analysis: I think that this again shows that a 1-way miss is not exactly what it is cracked up to be because if a golfer has an extreme miss bias, chances are it is a bad shot miss that they cannot control.


There is not a lot of concrete analysis to draw upon this data.

Analysis: It appears that in order to be more effective off the tee, the miss bias needs to be closer to 50/50. But, there are plenty of exceptions to the rule. I think one could say that the shorter hitting golfer on Tour needs to be close to 50/50 in their miss bias, but that does not explain the above average to average to below average distance golfers on Tour.

There is also some trend that the longest golfers need to be closer to 50/50 in their miss bias, but it the evidence is not strong enough to draw a confident conclusion to.

The flaw in trying to judge golfer’s games by their miss bias is that if a golfer misses left or right of the fairway, it may be right in line with where they wanted their misses to go. If there’s water hugging the left side of the fairway and a golfer misses the fairway right, that may be perfectly fine to miss the shot right of the fairway.

I think Driving Effectiveness is probably more reliant on the golfer getting the ball to curve 1-way, either left-to-right or right-to-left. I would generally think the consistency of getting the ball to curve 1-way, unless the golfer is purposely trying to work the ball differently, generally makes for a more effective driver of the ball.

Having a '1-way' curve of the golf ball has often been referred to as a 'shot cone' by some golf instructors.

I would generally recommend that golfers on Tour keep the miss bias to under 5%. However, I think we would need more data to tie in with aim bias to draw more concrete conclusions. Otherwise, I think the main thing we can tell is that missing 1 way too much of the time is likely detrimental to the golfer. But anything in between is too difficult to draw a meaningful conclusion.

I will address metrics like this in the e-book, 2012 Pro Golf Synopsis, which will be coming out in December.






3JACK

No comments: