3D Performance Analysis of Golf Swing Variability

[Big Lex]

Our discussion in another thread about the consistency of all golfers' swings was very interesting, and enlightening I think for most of us.

In the discussion, we agreed that if golf swings were essentially repeating and consistent for all golfers irrespective of skill level, but that results obviously varied significantly across skill levels, there must be some variability somewhere in the poor golf swings that isn't immediately recognizable on standard video. We used the terms "macro" and "micro" to illustrate this; we said all of our swings are consistent on the "macro" level, but that the worse you are at golf, the more inconsistent your swing is at the "micro" level.

This got me thinking about whether anyone has studied this, and of course, someone has. And as you might imagine, the studies come from our friends in the UK, the home of golf....appropriately so, I think. This paper is a collaboration of scientists from Leeds University in England and University of Limerick in Ireland.

It's not what you might expect from an article on the golf swing....it's highly technical and it isn't easy at first to see how it applies to how we play golf or how golf is taught, but I think it's very interesting reading. I put it here instead of the swing or instruction forums because it's really more a geek thing than a golf thing. But I think this kind of research will someday help us a great deal in showing us how to go about teaching and learning golf. 

The study looked at the variability from swing to swing for expert golfers (hcp </= 5), and correlated that variability with the variability of outcome of the flight of the ball. For this study, they used ball velocity as the outcome measure. The measure of variability was extremely high tech; a three dimentional motion analysis of multiple points on the body, including two on the head, and several on the upper torso and several more on the legs and feet. The study found that there was no correlation between the variability of the golfers' swings and the variability of their outcomes. In other words, even though each golfer's swing varied from swing to swing with regard to how each body part moved, the variability of these movements did not predict variability of the outcome.  

Stated yet another way....they weren't able to determine that consistency or lack of it among any specific body part predicts anything about the result of a shot.

Read! Enjoy! Talk amongst yuh-selves.   

Tucker et al (2013) Is outcome variability related to movement variability in golf.pdf

Edited March 26, 2016 by Big Lex

[natureboy]

32 minutes ago, Big Lex said:

The study looked at the variability from swing to swing for expert golfers (hcp </= 5), and correlated that variability with the variability of outcome of the flight of the ball. For this study, they used ball velocity as the outcome measure. The measure of variability was extremely high tech; a three dimentional motion analysis of multiple points on the body, including two on the head, and several on the upper torso and several more on the legs and feet. The study found that there was no correlation between the variability of the golfers' swings and the variability of their outcomes. In other words, even though each golfer's swing varied from swing to swing with regard to how each body part moved, the variability of these movements did not predict variability of the outcome.  

Stated yet another way....they weren't able to determine that consistency or lack of it among any specific body part predicts anything about the result of a shot.

Read! Enjoy! Talk amongst yuh-selves.   

Tucker et al (2013) Is outcome variability related to movement variability in golf.pdf

Non-golfers analyzing golf? Measuring ball velocity as the only outcome measure seems pretty myopic for what happens at impact. Initial direction, spin magnitude, and spin axis are all very important to golf.

I'll have to read the article, though.

[Big Lex]

11 hours ago, natureboy said:

Non-golfers analyzing golf? Measuring ball velocity as the only outcome measure seems pretty myopic for what happens at impact. Initial direction, spin magnitude, and spin axis are all very important to golf.

I'll have to read the article, though.

Yeah I thought the same thing. But ball speed is actually a very good, rich piece of data when you think about it. Just about any error in striking, whether it's an off-center hit or an open or closed face will change the ball speed. 

Since they used a flightscope, they certainly have all that other data you mention. I wonder if they just weren't sure how to analyze it, or maybe are saving it for another study. And the authors do mention that these other things you mention are important and need to be studied as well.  I think you'd be surprised how well these research scientists actually do understand the sports they study. But since they are conducting true, rigorous research, where they have to really control the environment and sources of unwanted variation, often the data they collect seems weird or limited. 

[iacas]

13 hours ago, Big Lex said:

Since they used a flightscope…

FWIW, they didn't use a FlightScope. For ball speed what they used is probably fine, but I wouldn't trust the other data from it.

I've only skimmed but they didn't even share the ball speed data did they? And 10 were women? 5 handicaps?

I will read with more depth eventually but it didn't seem like a well done study. What actual variability were they testing? How the right elbow and 10 other spots varied from one swing to another (or, really, across all 10 swings)?

Because even if one of them is different by a little the others might differ as the compensation. Millimeters aren't large.

[Big Lex]

7 hours ago, iacas said:

FWIW, they didn't use a FlightScope. For ball speed what they used is probably fine, but I wouldn't trust the other data from it.

I've only skimmed but they didn't even share the ball speed data did they? And 10 were women? 5 handicaps?

I will read with more depth eventually but it didn't seem like a well done study. What actual variability were they testing? How the right elbow and 10 other spots varied from one swing to another (or, really, across all 10 swings)?

Because even if one of them is different by a little the others might differ as the compensation. Millimeters aren't large.

Regarding Flightscope, I must have confused this with another study I read that day. 

  The variability number was a complicated calculation. It would take me forever to write it here so you'll have to read the paper. But essentially they had average positions for each body part across all swings, then calculated the variation of each individual swing against the average. They looked to see if there was a correlation between movement variance and ball speed variance. 

It was a within-subjects design. This means they look to see if golfer A's movement variation is predictive of golfer A's ball speed variation. 

[iacas]

22 minutes ago, Big Lex said:

The variability number was a complicated calculation. It would take me forever to write it here so you'll have to read the paper. But essentially they had average positions for each body part across all swings, then calculated the variation of each individual swing against the average. They looked to see if there was a correlation between movement variance and ball speed variance. 

It was a within-subjects design. This means they look to see if golfer A's movement variation is predictive of golfer A's ball speed variation. 

I've now read it. I don't care to figure out what mm³/mm on a logarithmic scale meant in terms of what you'd actually see or measure.

I thought the study was pretty flawed, to be honest:

• 10 women? Meh.
• No display of the actual ball speed numbers they got?
• Why not clubhead speed? Ball speed depends on contact, etc.
• If the right shoulder is in one place but the elbow compensates and is in a slightly different location too, maybe the player hits the ball a similar speed, but they see little correlation? They seemed to try to track too many variables.
• The shirts can move a little from swing to swing. Less so the ankle and other ones, but the headband ones can, too. I'd be curious to know what the address variability was (maybe it was noted and I'm forgetting).
• What were the contact locations like?

[Big Lex]

They went to great lengths to make sure the marker placement was as accurate and repeatable as possible, and yes they did acknowledge the potential issue with address position variance, and they accounted for this as well.

Like any system you design and measure, you have to make some decisions and some compromises or you don't get anything done. I think in a way ball speed was a clever thing to measure for precisely the reasons you identify. Measuring ball speed would "capture" errors or variance in club head speed, face contact location, and face angle.

11 hours ago, iacas said:

10 women? Meh

Why the problem with studying women's swings?

11 hours ago, iacas said:

If the right shoulder is in one place but the elbow compensates and is in a slightly different location too, maybe the player hits the ball a similar speed, but they see little correlation? They seemed to try to track too many variables.

I'm not sure what you're getting at in the first sentence. They ran correlations between the ball speed and a number of the body measurements. They didn't correlate one body part to another....which might have been an interesting thing to look at in itself....I had a similar thought about the number of variables they chose to study.

I guess I can see how an instructor might find this study of little practical use. I think it's the equivalent of a "basic science" experiment on golf...trying to get at primary principles. The sorts of studies that you did in LSW and that we see in other books studying the golf swing from a teaching perspective are more like "clinical trials."

I don't agree that it's a flawed study; the design is excellent. From a science perspective, it's excellent. Good controls, a clearly defined null hypothesis, rigorous statistics, etc. It may not have immediate practical utility for golfers, though.

[iacas]

Busy all day but I disagree it's a good study. I've quickly listed the reasons. The shirts move slightly too. You can say "we tried" and still not have the cleanest results.

Plus, again, how much variation did they actually see? What were the actual ball speed numbers?

And spare me the "golf instructor" stuff. I've got degrees in the hard sciences. I know a thing or two about studies. This one was weak IMO. Too many variables. A bad comparative (ball speed). Women aren't very good. Etc.