When will data science finally reach the world of golf?

[Palmer4Ever 1]

First of all: I'm not a golf expert by any means, if anything, I'm a newbie. So, when I make the assessment that data science has not reached the golf industry, I might not have all the facts, I'm just reproducing what it looks to be the pattern from the outside and as someone who is familiar with the data revolution in baseball, basketball, and football.

I also just finished the book: "Everybody Lies Big Data, New Data and What the Internet Can Tell Us About Who We Really Are", a great book on data science in general, not only in sports, so the topic is very vivid on my mind. 

How much of the Moneyball revolution has transpired to the world of golf? Are we valuing players and predicting tour results better in the face of big data?

If you are into this, I would love to learn more and know your opinion on this matter. Also, if I am completely wrong, I am ready to learn about the advanced quantitative methods used in golf today.

Thanks!

 

[saevel25 1,620]

41 minutes ago, Palmer4Ever said:

How much of the Moneyball revolution has transpired to the world of golf? Are we valuing players and predicting tour results better in the face of big data?

It would still be extremely tough to accurately predict the winner. These advance stats tell us that the winners on the PGA tour are near the top in ball striking (strokes gained tee-to-green) in combination with having a hot putting week. 

It's impossible to predict if someone will have a hot putting week. 

With regard to a golfers value, in terms of golfing ability to dollars, is pretty much spot on. Maybe not for endorsements, but for winnings it is. Golfers only make that money for how well they play. It's not like baseball where there are contracts and you can overpay for a player.

[nevets88 1,680]

Moneyball in golf has been huge for awhile, see all the stats available, the consultants players hire to optimize how to practice and play and in analyzing the swing with data from radar, pressure plates, 3D, putting trackers, etc... Mark Broadie wrote his book in 2014, Trackman was being used for golf, I dunno, well before 2010. If you ask me, it reached the world of golf a long long time ago, a tiny sliver of it is only being presented in the national broadcasts. The radar numbers you see on the telecast is just 1/1,000th of what is available.

[Wally Fairway 340]

@Palmer4Ever you should read Lowest Score Wins

As far as predicting tournament winners week-to-week, that would be very tough - look this week and last; this week the #1 ranked player in the world has had his struggles is likely to miss the cut (although the cut keeps moving higher and higher), last week Spieth finished 2nd after missing the cut the prior week.

[iacas 7,959]

1 hour ago, Palmer4Ever said:

How much of the Moneyball revolution has transpired to the world of golf? Are we valuing players and predicting tour results better in the face of big data?

Predicting winners each week is still fraught with the obvious difficulties, but the data-driven looks at golf HAVE come to golf.

Primarily through two books:

• Every Shot Counts
• Lowest Score Wins

I'm one of the members of a small but growing group of people who are pushing this, and who are experts in this field.

So yeah, it's here. It's come to golf. And it's shaken things up quite a bit… but there are still a lot of people who think putting matters a lot, unfortunately.

[406pat 13]

31 minutes ago, nevets88 said:

Moneyball in golf has been huge for awhile, see all the stats available, the consultants players hire to optimize how to practice and play and in analyzing the swing with data from radar, pressure plates, 3D, putting trackers, etc... Mark Broadie wrote his book in 2014, Trackman was being used for golf, I dunno, well before 2010. If you ask me, it reached the world of golf a long long time ago.

I would say data science has been huge in golf, but not Moneyball.

I don't think there will ever be a specific moneyball revolution to impact golf... at least in the way it did/has for other sports. The reason I say that is that the real innovation wasn't the data, it was challenging the assumptions behind assembling a team. Pre-moneyball teams sought the best overall player at every position based on offensive and defensive play with defensive play often being the deciding factor. Moneyball used data to show that the old assumptions were flawed and teams could benefit by breaking from that way of thinking. Baseball has a long season with a lot of games so there is enough comparable data to extract reasonable predictions. Golf is largely an individual sport at the pro level and infrequency of team events renders the statistics insignificant. Too many variables change and the data sample is too small. 

I could maybe see sponsors using a more data driven approach to figuring out who to sign and for how much or something like that. I still think there's too much variability in an individual game that can't be accounted for in past performance data to spark a revolution though.

 

[nevets88 1,680]

I'm interpreting moneyball differently, not predicting winners, but to win. For baseball, moneyball, at least from my pov is to win. To win is to buy runs. To buy runs is to find guys with the highest OBA or whatever for the best price, I'm not a baseball stats guy.

For golf, it's to win, to win is to minimize the number of strokes taken over 4 rounds. To do that is to analyze where you miss, to find tendencies that only appear to you when you look at lots of historical data, the inadequacies of your swing or your putting and that's where stats, radar, high speed video (I think the info you get from video is moneyball), etc... That's moneyball applied to golf to me. To predict the winner, that's more fantasy sports, no?

And the new ideas derived from data collected and the smartest guys analyzing it have shaken up traditional beliefs taken as fact. The folks with the biggest megaphone, your national broadcasters, imho are of the traditional set and slow to adapt new findings from big data.

If you look at moneyball from the player's perspective, it has expanded, big time, I would guess, starting around 10 years ago.

[iacas 7,959]

27 minutes ago, 406pat said:

I don't think there will ever be a specific moneyball revolution to impact golf... at least in the way it did/has for other sports.

I think we had it. It's from ESC/LSW as I noted above. We flipped the idea of how golfers score on its head. We showed how "Drive for show, putt for dough" was almost as backward as it could get.

27 minutes ago, 406pat said:

Baseball has a long season with a lot of games so there is enough comparable data to extract reasonable predictions. Golf is largely an individual sport at the pro level and infrequency of team events renders the statistics insignificant. Too many variables change and the data sample is too small.

Golf's "moneyball" is where we put together a good season, or consistently good tournaments, or consistently good rounds of golf. We don't have positions (like shortstop, catcher, etc.), but we have different types of shots (drivers, irons, bunker shots, etc.), we have different strategies available to us (like the shifts in baseball, or what pitches to throw when, or when to switch pitchers, we have the choice of where to aim, when to be aggressive, what clubs to hit, how penal the rough is for a certain shot versus flirting with a hazard).

[jkelley9 70]

I think it falls apart with the fact that it's one person. It'd still be reaching even for college teams and match play. There's also a lot of overlap. Strokes gained was an amazing insight, but it's not Moneyball caliber (in terms of utility). 

There's no "reset" to a normal condition in golf. Unlike in baseball, where every play resets to the exact same position but with different players in the field (but not all the time) - same with football, every play resets to the line of scrimmage. Each hole in golf is only played 4 times, and it's a small fraction of the total 72 holes of a tournament. Makes it tough.

[iacas 7,959]

13 hours ago, jkelley9 said:

There's no "reset" to a normal condition in golf. Unlike in baseball, where every play resets to the exact same position but with different players in the field (but not all the time) - same with football, every play resets to the line of scrimmage. Each hole in golf is only played 4 times, and it's a small fraction of the total 72 holes of a tournament. Makes it tough.

I disagree. Everyone tees off from the same spot. In baseball pitchers, weather, the count, the game situation is different all the time. Players employ shifts on different teams differently, etc.

[Palmer4Ever 1]

On 6/2/2017 at 0:46 PM, nevets88 said:

Moneyball in golf has been huge for awhile, see all the stats available, the consultants players hire to optimize how to practice and play and in analyzing the swing with data from radar, pressure plates, 3D, putting trackers, etc... Mark Broadie wrote his book in 2014, Trackman was being used for golf, I dunno, well before 2010. If you ask me, it reached the world of golf a long long time ago, a tiny sliver of it is only being presented in the national broadcasts. The radar numbers you see on the telecast is just 1/1,000th of what is available.

 

Thanks for the book recommendation, I'll read it for sure! I think it's true for most sports that, even when the data science behind them has advanced quite a bit, the mainstream perception and tv announcers seem to be on a different level.

On 6/4/2017 at 8:00 AM, iacas said:

I disagree. Everyone tees off from the same spot. In baseball pitchers, weather, the count, the game situation is different all the time. Players employ shifts on different teams differently, etc.

Yeah, this is an interesting perspective. Definitely worth exploring.

On 6/2/2017 at 0:53 PM, Wally Fairway said:

@Palmer4Ever you should read Lowest Score Wins

As far as predicting tournament winners week-to-week, that would be very tough - look this week and last; this week the #1 ranked player in the world has had his struggles is likely to miss the cut (although the cut keeps moving higher and higher), last week Spieth finished 2nd after missing the cut the prior week.

Has anyone tried, though?

[SLAMGolf 4]

People are doing data science with golf.  For example:

http://www.golfdigest.com/story/the-golf-digest-tournament-predictor-for-the-2017-us-open

Predictors appear to be previous tournament success, scoring average, driving distance to course length, etc.  Probably every stat you can grab from the pga tour website, tournament results for the players, plug it into a gradient boosting tree regression model and see what it spits out.  My guess is, the RMSE would be unsurprising.  e.g. Brian Harman on Erin Hills !?

[SandraH155 0]

It's a great topic for research. I'm pretty sure we can find some great examples of data science in golf. Wanna ask important site to do that research and I think it will be a great paper about this.

[SandraH155 0]

22 hours ago, SandraH155 said:

It's a great topic for research. I'm pretty sure we can find some great examples of data science in golf. Wanna ask important site to do that research and I think it will be a great paper about this.

Any pieces of advice for me about this topic, because I'm pretty confident in doing that research. Maybe the help of your own experience will be of a great help.

Edited August 13, 2017 by RandallT
Removed suspicious link that frankly seemed spammy

[scotth 45]

On 6/2/2017 at 3:26 PM, saevel25 said:

It would still be extremely tough to accurately predict the winner. These advance stats tell us that the winners on the PGA tour are near the top in ball striking (strokes gained tee-to-green) in combination with having a hot putting week. 

....not to mention the fact that occasionally you have pro golfers do bonehead things nobody can explain on the golf course like Jason Day yesterday.

https://www.yahoo.com/sports/nick-faldo-jason-day-one-worst-decisions-ive-seen-professional-golfer-make-000417146.html

[Wally Fairway 340]

Here is a twitter account that gives weekly predictions for PGA events - not sure how well they have done with this

 

[Buckeyebowman 417]

On 8/13/2017 at 6:46 AM, scotth said:

....not to mention the fact that occasionally you have pro golfers do bonehead things nobody can explain on the golf course like Jason Day yesterday.

https://www.yahoo.com/sports/nick-faldo-jason-day-one-worst-decisions-ive-seen-professional-golfer-make-000417146.html

Or Phil on the 18th in the final round of the U.S. Open at Winged Foot! "Moneyball" came into vogue because of the Oakland A's and the movie! It's just analytics! Reducing player performance stats to a set of numbers that can be easily digested by someone familiar with that system.

But what do these numbers show, really? How are statistics generated? Aren't statistics merely the record of past performances? Yes. That's all well and good, but how can they predict the future? Fact is they can only predict odds.And sometime not very well.

After all, how many World Series did the A's win during their "moneyball" era? I think it might have been one! When they managed to assemble a "lights out" pitching rotation. 

More recently, I, as a long suffering Cleveland Browns fan, have seen "analytics" introduced into the Browns front office. How has that worked out for them? 1-22 anybody?

[billchao 1,781]

11 hours ago, Buckeyebowman said:

Or Phil on the 18th in the final round of the U.S. Open at Winged Foot! "Moneyball" came into vogue because of the Oakland A's and the movie! It's just analytics! Reducing player performance stats to a set of numbers that can be easily digested by someone familiar with that system.

Advanced statistics and analysis in baseball is a lot more prevalent than just the Oakland A's and  the movie, but yea, the book is how the public became aware of it for the most part. Every front office in every major sport uses analysis whether they're public about it or not, most with their own proprietary systems.

11 hours ago, Buckeyebowman said:

But what do these numbers show, really? How are statistics generated? Aren't statistics merely the record of past performances? Yes. That's all well and good, but how can they predict the future? Fact is they can only predict odds.And sometime not very well.

Judging by your comments, I'm guessing you never studied statistical analysis or forecasting. No, you can't predict the future, but you can make decisions based on trends that are most likely indicative of future performance. It's a lot more effective than simply guessing or "going with your gut."

For example, if a guy misses almost every cut in a tournament with bad weather and his next one has rain in the forecast, it's a pretty safe bet he's going to miss the cut. That doesn't mean he definitely will, but his past performance indicates that is the most likely outcome.

I watched a program that talked about those daily fantasy gambling sites once and the top winners all use very complicated analysis tools to predict performance of athletes. They got into some really minute details like how an athlete performs under certain weather conditions, start times, etc. They're not just picking guys blind.

11 hours ago, Buckeyebowman said:

After all, how many World Series did the A's win during their "moneyball" era? I think it might have been one! When they managed to assemble a "lights out" pitching rotation. 

Zero. Billy Beane himself said the playoffs are based on luck. Baseball has a playoff format that is different than regular season play so you can build a team that can win 100 games but isn't necessarily the strongest for a best-of-five or seven series. Even if it is, you can have a couple of guys hit a slump during the playoffs and lose anyway.

The announcers made a big deal in the broadcast about how great Corey Kluber was in the month of September but then he faltered in the playoff series in October against the Yankees. They just assumed he was going to continue to be lights out, but he regressed to the mean. He's still one of the best pitchers in baseball, but if you combine his performances in September and October together, you're going to get a better picture of how he is as a pitcher than if you only evaluate his performance in the arbitrary month of September (or October, for that matter).

12 hours ago, Buckeyebowman said:

More recently, I, as a long suffering Cleveland Browns fan, have seen "analytics" introduced into the Browns front office. How has that worked out for them? 1-22 anybody?

How did they do before "analytics" was introduced? A poorly managed team is poorly managed, with or without additional stats to help make decisions.