Putt Probabilities

Like many others, I was incredibly jazzed by Tiger’s victory on Monday.

But I was frustrated that the commentators routinely failed to mention the putting distance to the hole. It would be nice to know, “It looks like Rocco has a 25-foot putt.” But I, for one, would like commentators to go further and routinely give us information about the probable outcome of the next stroke. If the golfer is putting, I want to know the probability that the put will go in.

Analogous information is an old hat in basketball. When Kevin Garnett goes to the free-throw line, we are told what his FT percentage has been this season and in the playoffs (and in this very game).

Putting is not as standardized as shooting a free throw. But it would be useful to know that pros generally only make 60 percent of five puts. At least that’s what’s shown in this article (based on pre 1995 data):

Golf Chart

It shouldn’t be insurmountable to give player-specific probabilities conditional on distance to the hole or even real time predictions that take into account distance, player, and even place on the green. You can double check after the fact that your predictions are unbiased — simply by looking at the times when you predicted a 60 percent make probability and seeing if the ball sank 60 percent of the time.

Real time predictions could also be made about the probability of hitting the fairway on a drive or probability of hitting a green. Letting viewers know the probability that an ordinary pro (or that a particular player) will make a putt doesn’t take the drama out of the game. Just the opposite. They let us know when players make a truly improbable shot or blow what should be a gimme.

For example, I bet few in the viewing public would have been able to formulate a very accurate assessment of the odds that Tiger would have made the last putt on Sunday.

After the fact, it’s too easy to say, like Mediate, “I knew he was going to make it.”


it's easy:

3ft - basically guart'd
5ft - almost guart'd
10ft - it should go in
15ft - it would be nice if it went in
20ft - not guart'd
25+ft - its ok to take a break from watching


That graph is a nice exponential decay, beloved of scientists everywhere! I agree with the others that putts are so different and distance is only one variable.

To make it accurate, you'd have to factor in the incline, amount of break and time of day, e.g. give stats for, say, uphill slight left to right 15 footers with lots of spike marks.

I don't think anyone wants to go down to that level of granularity though!


Putts of the same length can have very different levels of difficulty depending on the slope - up hill vs down hill, break, wind, and speed among other factors. Therefore its very difficult if not impossible to create putting statistics which are meaningful. Thanks.


Is the Freakonomics crowd scared to know how poor of golfers they really are?

Yes, all putts are different, they move the holes, they roll best in the morning, yada, yada yada. So Kobe's field goal % is based on the same guy guarding him 1 on 1? Nope. Is it good to know? Yep.

So maybe you have the putting data by player, by course, by hole, by hour, by relative score, by round, by w or w/o Tiger in your group, etc. One thing for sure, it's interesting and would bring a whole new level of understanding, respect, and passion to the game.


From the looks of this graph, 3 foot puts have about a 17% MISS rate? Is this among pros? That's astounding, and tough to believe. Even with the bias of being more likely to remember the misses as opposed to the routine makes, that's hard to believe.


I too heard Johnny Miller more than a few times on Sunday give probablilities to certain putts and shots. I heard him say it was 17 to 1 that Tiger could make a shot onto the green from where he was.

What I'd like to see is how the field is doing on a given putt. I think it would be easy to plot out throughout the day then by the time the leaders came through at the end you'd have a green with over a hundred dots on it and the %ages from each spot which would account for the differences in difficulty of each point on the green. Now we have a low-tech version of that with a different commentator at each green and they usually can tell the viewing audience how the field has done from that spot on the green. You'll hear Roger Maltby say something like, "I've only seen one putt made from this spot today. Everyone's been reading it as a left-right, but it never breaks."


While a golfer might have a better feel for the green on Sunday compared to Friday, as it was stated above, the hole locations vary from day to day. Also, the time of day and how many people have been on the course ahead of you make a large difference on medium and long distance putts. Footprints that have not had time to fully rise create very bumpy conditions that cause a putt to go off line (watch the ultra close up of Tiger's final Sunday putt and you can see it bounce like 15 times on the way to the hole).

Also, golf is extremely mental (especially putting) so the pressure of making that putt on Sunday could be far greater than any putt you have ever hit in the past which means that the sample you used to generate your percentages may have little bearing on the putt at hand.


@Gary- the only issue with your theory is that they move hole locations every day. on many greens, that means the look of the putts is VERY different. plus, at the US Open, they even moved tee locations on some holes to further throw the players off their game.

A good point as well. What I really had in mind was the speed of the greens which (neglecting weather) seems to be pretty consistent throughout the weekend. The greens at the Open this weekend made it look like they were putting on cement. You can read the green perfectly, but if your speed is off by just a touch on a green like that, you're sailing the ball past the hole.


I've often wondered why golf doesn't place concentric circles around the whole giving ranges of probibility for a given distance. This would be most effective when the approach shot is hit, then you could see as the ball is rolling, what might happen. This would be similar to the computer generated colors on the field that are displayed just prior to a field goal kicker in football making his attempt. As for the degree of difficulty. How much effort would be needed to 'grade' each putting green? There are everyday givens like the slopes of the green. And variable like wind and firmness. This isn't rocket science. Anyways its just for estimating purposes anyways. If we can accept 7 day weather forcasts as plausible, when we can accept this.


Like Gary said, putt difficulty is much more than just distance to the hole. Tiger would probably be pretty good at sinking 40 foot puts on ground that has absolutely no left or right break. He's pretty good at sending the ball exactly in the direction he wants it to go. If he wants to aim 2 and a half ball lengths to the left of the hole, he can probably do it fine from 40 feet.

But an alternating left-right 10 footer with a sharp downhill beyond the hole into the water would be a much tougher putt. The read of the ground and having to judge how the speed of the ball will affect the amount of break is likely more difficult.

I suppose commentators can be consistent with describing each putt as fully as possible. Though they seem to fairly often describe what the read looks to be, estimate the distance, and mention how others had fared with similar looking putts.


Please...let's not suggest anything Miller says is in any way scientific. Any kind of "17 to 1" sort of ratios he's mentioning are based off what he thinks a great player (i.e., himself in 1965) would be capable of.

Ken D.

I agree with with the post, but there was actually a rare exception on this telecast. An announcer, I think Johnny Miller, referred to a 20 foot putt as being about a one in five chance; it caught my attention because reference to such probabilities is indeed unusual. It was on one of the crucial concluding holes, but I am not sure which one.

Charles D

I would love to see this. Guys always need numbers to talk about sports, it gives us something to do and argue about. They don't have to be a good predictor since puts can be so different, but it would be a starting point.

Statistics have come a long way though and I could see golfers really benefiting from it. I'd like to see some specific statistics like they did in the movie Major League.


I thought this article was about Put probabilities. What was funnier though is that even when I realized what it was about, it had a graph that looked like the volatility smirk. Interesting coincidence.


What they could do is the kind of real-time tracking that MLB offers, where you can see the actual pitch trajectories. They could note each player's putts and then put that info up: here's how the putts from this area break, how many have gone in, where they go if they're short, where they go if they're long. You could then generate probabilities for each golfer based on his history (and then you could see the hot streaks and bet on them in real-time).

This info could be made available in more than one form. MLB gives some away but charges for more data. That could then make the webcasts more valuable, especially given the combination of subscription and ad revenue.


The PGA tour now uses Shotlink, a GPS based system, to record every shot hit on the PGA tour including putts. NBC could easily announce Tiger's 5-10 ft percentage of made putts over the last 3 years if they wanted to.

Generally, the variability of putts gets worse later in the day as more foot traffic tramples the area near the hole. Its called the "lumpy donut" by Dave Pelz and is why even when using a machine, beyond 6 feet from the hole, the percentage of made putts drops below 50% and dramatically declines.

Currently, shot link does not record whether the putt is downhill or uphill or left-to-right or right-to-left, so the difficulty of the putt is not measured, but is a significant factor.

The one stat I would love to see is Tiger's percentage on those dramatic putts that matter. It seems like he can muster amazing skill at the particular moment. I suspect the data reveals that he is a very good putter all the time, whether on the 4th or 18th hole and the reason it seems like they "matter" so much is because he is in contention so often, not because he has an extra gear. If anything, his consistency shows that his persistence is as significant as his dramatic flair in the moment of truth.



Yeah, I'm curious for that data on (what looks like) the 14 ft putts


Sometimes broadcasts will throw out some stats - you often get information on putting average, driving distance, fairways hit, greens in regulation. but as for putt/distance info - it's just because every putt is so unique, stats don't mean much. they WILL however - very often - mention when a player is doing very well or very poorly. I know they mentioned that Rocco hadn't hit a putt outside of 10 feet.

but the variability is a problem. maybe a player sinks 77% of his 10 foot putts - but this putt is 10 feet on a slope above the hole. its not very relevant that he hits 77% of his 10 foot putts - it's really only relevant the % of downhill putts he hits from 10 feet, when the greens have a stimp rating of -whatever this green is-.

either you have to be very general in the putting statistics, or you have to be so specific as to severely limit your sample size.

@Gary- the only issue with your theory is that they move hole locations every day. on many greens, that means the look of the putts is VERY different. plus, at the US Open, they even moved tee locations on some holes to further throw the players off their game.



The data for this is all widely available; the PGA tour sponsors a data collection service called ShotLink that records every single shot a player takes. On putts, the distance to the hole is measured to the inch using a laser system (Notice those green towers behind the greens? They are staffed by volunteers charting each shot).


The commentators frequently commented on how players had been hitting or missing putts from a particular region of the green earlier that day. They often predicted exactly how a player would miss the shot. From that it seems that the likelihood of a player making a shot has more to do with the hole location and green shape than the general accuracy of that player at a particular distance. A more useful predictive graphic would be to show the trajectory of the previous puts from that region of the green had travelled that day (possible only if you automatically record each putt on that hole).