THE BOOK--Playing The Percentages In Baseball

I have not RTFA yet, but there are several reasons why this may be the case other than a different “approach” by the golfer.  One reason is that par putts are more often made after a chip from off the green whereas birdie putts are more often made from an “approach” shot (usually 100 or more yards away).  When a PGA golfer is chipping from just off the green he is going to put the ball in an area on the green relative to the hole which is easiest to make (for example a flat or uphill putt).  So a 6 foot putt, for example, is going to be an easier putt on the average when you just chipped the ball as compared to when you just hit a 130 yard approach shot.

Interesting, so you are suggesting that one of the things that they did not control for is the slope of the green relative to the position of the ball.

So, what they must do then is to not only control for distance of ball, but the distance/lie of the PREVIOUS shot that got them onto the green.  Yowza, that’ll kill their sample size.

Also, I’d like to know the score of the NEXT hole to see if there’s an effect to bogeying / par the hole in question.  If golfers really think about this stuff as much as they do, there might be a short-term effect in play.

I don’t know that that is the case, but I am speculating that it is part of the equation.  I doubt that the researchers controlled for the break of the putt.

On the other hand, I don’t doubt that their conclusions are true.  Being an avid golfer, myself I have little doubt that all golfers, even professional ones, are more aggressive on par putts for the reasons stated in the article above.

But, as I said, I think that you will find par putts easier than birdie or eagle putts from the same distance for the reason I mentioned in my first post.

Actually, I’ll give you another reason which is probably the primary one:

Like the short chips I was talking about, a lot of those par putts are when the birdie putt goes past the hole.  When a PGA golfer hits a putt past the hole, he knows exactly what the putt is going to do, break-wise, coming back.  IOW, a 4 or 5 foot putt “coming back” after you missed a long putt is MUCH easier than a 4 or 5 foot putt for birdie when it is your first putt.  Even on a putt which is short, it is much easier to sink the next putt because you know the speed of the green near the hole and you have a better idea of the break on the next putt.

If the researchers did not control for this, their study is pretty worthless to be honest.

Now, maybe they did control for these things, but if they did not, shame on the researchers.  Are there any peer reviews of these sports papers?  We seem to find one bad paper after another, usually because the researchers are not familiar enough with the sport.

The professors found that birdie putts were made about 3 percent less often than otherwise identical putts for par.

That to me says they controlled for what MGL says in comment #1. Unless you’re talking about slope of the green (which I think you might be), in which case you have a good point.

Right, he’s talking about the slope.

From the paper:

Position on the green. Although we account for distance, it is possible that birdie putts start from a more precarious position on the green than equally-distanced par putts. For example, approach shots to the green may be shorter for par putts than they are for birdie putts, and as a result, players putting for par may be able to avoid difficult spots on the green (e.g., sections of the green with awkward slopes).
The level of detail in our dataset enables us to address this concern. We use the x, y, z coordinates to control for putts taken from different positions on the green. To do this, we first divide the area around each hole in each round in each tournament into four quadrants or ―pie pieces.

We characterize each putt attempt by quadrant, and we report regression results with dummy variable controls for each quadrant-hole-round-tournament in Column (5) of Table 3. These fixed effects control for location-specific differences.

I’m not even talking about controlling for the slope of the green in my second post. If you have a 4 foot putt with a certain slope after your “approach” shot (say, an iron from 130 yards) and you have another 4 foot putt with the exact same slope after you just putted (or chipped) the ball 4 feet past the hole, you will hole the latter MUCH more often (10% more if I had to guess) than the former.  And, as I said, most par putts are the latter and most birdie (or eagle) putts are the former.

Now, to be fair to the authors, assuming that they did not control for these things, my scenario would rarely happen in 10 or 15 (or more) foot putts, so you would likely see the effect on putts from, say, 3 to 6 feet.  But, then again, I am not sure how many 15 or 20 foot (or more) par putts there would be in their sample.  That is a fairly rare thing for a PGA golfer.  They would have to have hit a bad bunker or chip shot, or taken a penalty stroke prior to the shot onto the green, or something like that.

I’d have to read the actual paper.  It does not look like it is available.

There’s a link to it in the article. I didn’t have a problem downloading it. They didn’t even charge for it, a rare thing.

Oh, OK.  I went to that link and I thought it was just an abstract.  I’ll check it out.

Okay, it looks as if they have considered at least some of my concerns above:

We consider and rule out several competing explanations for this finding. First, prior to hitting a par putt, players may have learned something about the green (by having already attempted a birdie putt). Second, birdie putts may start from a more precarious position on the green than par putts due to a longer approach shot.... For example, we can match par and birdie putts attempted within 1 inch of each other on the exact same hole in the same tournament.

I have to read the rest of the paper.  If they just matched locations on the green, that will not help them with the problem of a player getting a “read” on a putt after missing a putt.  IOW, they would be properly addressing my first concern, but that would not be addressing the second, and most important, concern.

We’ll see if they address that.  The only way I can think of to address that concern is to compare “first putts” only, from the same locations on the green.  I hope they did that.

It appears as if they controlled for lots of other variables, including the number of prior putts taken.  For example:

Learning. Alternatively, players may learn about the condition of the green from earlier putts. After putting once, golfers may learn important information about the slope or conditions on the green. Compared to birdie putts, par putts are more likely to be the second putt attempted by a golfer on the green. As a result, par putts may be more accurate than birdie putts, because golfers have learned important information. In fact, it is possible that players may learn important information from watching their partners putt on the green.
To control for learning effects, we include separate dummy variables for the number of putts already attempted on the green by the player and the player’s partner.6 The results from this specification suggest that learning is important. As we report in Column (3) of Table 3, golfers are significantly more likely to make the second and third putts on the green than they are to make otherwise similar first putts on the green.7
By including controls for prior putts on the green, the point estimates for birdie and eagle putts are reduced by 20 to 30%. However, the remaining differences attributed to birdie or eagle putts relative to par putts remains highly significant. In all future specifications, we include these controls.

This appears to be a very good study. As I said, I am not too surprised at their results.

The fascinating thing is the magnitude of the bias.  An extra stroke per 72 holes is enormous as indicated by the 1.2 million dollar difference.  Any professional golfer who reads this study or is made aware of it should take heed!

It feels like in sports studies like this, you find the players or others, arguing against the study and finding every possible way to discredit it with “what ifs” and “you didn’t consider this or that.”

However, if the NY Times report is right, it seems that the players are agreeing with the study and they aren’t surprised by their own irrational results.  That doesn’t mean the authors of the paper are right, but I do think it means there is a better chance they are.

King Yao, I agree.  This is one of those rare instances where, as an avid golfer, and one who follows the PGA tour closely, I was not too surprised at the results.  And I certainly jumped the gun in suggesting that they had not considered alternative explanations.  They did a very good job of that.

I don’t think you are criticizing anyone for critiquing these studies, but if anyone is, keep in mind that that is what is supposed to be done.  It doesn’t do anyone much good to pat a researcher on the back and say, “Nice job.” It is imperative that when someone does a study, that it get “ripped apart” (not literally of course) by anyone who is qualified to do so.  That is the proper vetting process to protect the integrity of the research and especially the conclusions.  There is just too much bad research out there that slips by.  Of course, before it gets mainstream publicity, it should already be vetted, but that does not always happen.

Anyway, not that I understand all the (statistical) intricacies of this study, but this paper seems to be a very good one with valid conclusions.

And, as I like to point out a lot, there is a strong Bayesian a priori probability that strengthens the significance of the results. Basically the a priori probability is the hypothesis in this study - that there is risk aversion behavior even in the highest level of a professional sport.  That should not be too surprising to us, as we see ridiculously copious amounts of suboptimal, risk-aversive behaviors and strategies in baseball, by players, coaches, and managers.  Why not a little bit in golf.

Although this is not really pure “risk aversion” like we see in other sports and behaviors.  I am not sure we can put a word or words to it, but it is more like, “I would like to make the birdie or eagle, but if I don’t, I am happy with the next best thing, and in fact, I don’t want to blow the next best thing (that is the risk-averse part) by making too aggressive a putt.  But if I am putting for par or bogie, I will be aggressive because if I don’t make it, **ck it!  I don’t care if I get double or triple bogie.  If I don’t make par, I’m screwed anyway.”

As the researchers point out, that is not good strategy unless you are leading a tournament by 1 or more strokes with only a few holes to go, or something like that.

Given that the golfers interviewed agreed with the conclusions of the researchers, I wonder how many of them really understand how much their suboptimal, risk-averse behavior costs them, and how many of them would be willing to change. They certainly could if they wanted to, although it would take practice as it is likely a subtle and partially subconscious dynamic.

16: “Given that the golfers interviewed agreed with the conclusions of the researchers, I wonder how many of them really understand how much their suboptimal, risk-averse behavior costs them, and how many of them would be willing to change. “

My gut guess is very few will even think about it again, and even fewer will even try to change.  And of those that do try to change a little, a good portion of them will revert back to the old ways the first time it backfires even a little bit.  I’m not saying anything bad about golfers, I just think that’s human nature.

King, I agree, mainly because it is indeed human nature to be risk aversive, even among otherwise smart and rational people.  And in this case, I think the difference is subtle even though it apparently (according to the researchers - I have not verified their “math") translates to a very significant difference in earnings per year.  And for the record, golfers are an extremely intelligent and educated breed, as compared to the other major sports (I think and I assume).  That is probably to be expected given the average socioeconomic class of the PGA golfer’s family.

Did they investigate how often the more aggressive par putts led to extra future putts?  Does the average risk-averse birdie avoid three-putts more than the aggressive par putt (assuming the par putt is the first putt)?  If not, then the golfers aren’t even lowering their risk with their suboptimal behavior.

Any time you make an “aggressive” putt, you increase your chances of a 3 putt.  That goes without saying.  On almost any “makeable” putt, a PGA pro is aiming for about 1-2 feet past the hole, in order to reduce the chances of coming up short.  An “aggressive” putt is one in which the golfer aims for maybe another half a foot past the hole, especially on putts that are not that short.  By definition, they will make slightly more putts, by virtue of almost never leaving it short AND by virtue of taking out some of the break (the faster the putt, the less the break), with the penalty being slightly longer putts when missing the cup.

In addition, an aggressive putt can also mean leaving your next putt in a less easier position to make, although not necessarily longer.  For example, if I am putting with a sidehill lie and I am putting NOT aggressively (say I need to 2 putt or less in order to win the tournamant), I might putt in such a way as to make sure that my second putt, if I don’t make the first, is an uphill or a reasonably straight putt.  However, if I am putting “aggressively”, I am trying to maximize my chances of making the first putt at the expense of possibly leaving my second putt in a position such that it is not that easy (downhill, sidehill, etc.).

So, again, by definition, an aggressive putt is one that maximizes my chances of making it, with little or no regard for the second putt.  So it pretty much goes without saying that the second putt after an “aggressive” putt is going to be holed less frequently than after a conservative (lag) putt, since by definition, a conservative putt is one in which I maximize my chances of making the second putt after a miss.