Strokes Gained (Tee, Fairway, Rough, Sand, and Green)

I'm looking for input on a spreadsheet that broadens an earlier attempt to analyze strokes gained putting. Some may have seen it, but here's a copy (sorry, Mac only--> https://www.dropbox.com/s/cemid8atb4b1srz/strokes%20gained%20putting.numbers?dl=0. Excel version someday later hopefully).

First here are some background links:

Broadie paper " Assessing Golfer Performance on the PGA TOUR" (April 8, 2011), Columbia University Graduate School of Business--> http://www.columbia.edu/~mnb2/broadie/Assets/strokes_gained_pga_broadie_20110408.pdf

Discussion of a new app (scroll down for detailed discussion) based on Every Shot Counts where we should be able to analyze strokes gained for all aspects of our game --> http://everyshotcounts.com/248-2/

Example data from a guy named "dornstar" on golfwrx forum--> http://pic100.picturetrail.com/VOL875/3165781/24393663/410160915.jpg

I haven't verified all of the data, but what I did check definitely matches the graphs in Broadie's paper above. The guy did a nice job on the chart.

I've found many people talking about creating their own spreadsheet to analyze their own games, but I haven't found any samples. Broadie mentions his team working on an app that I haven't found released yet, and it would certainly make a spreadsheet obsolete, but then again, they'll probably charge, and at least this "poor man's" attempt would be free.

Anyway, here is what I had in mind, and I was looking for ideas to get this to work better. Interface ideas. Other things to incorporate and calculate, etc.

As it stands now, you simply input as follows:

You fill out every shot with the yardage to the pin with a code after it (t- tee, f- fairway, r-rough, p- penalty, x- recovery shot, s- sand, g- green). Apps like Game Golf will be able to get this from GPS. You also need to enter the par for each hole. But that's it.

From there, I'm working to get this to work (formulas not written- it's tricky). An instant breakdown about where you lost strokes compared with PGA players:

Here are the details how it would calculate (just showing the first few holes so it fits the width):

Basically, it adds a column next to your input column, and it looks up from the starting distance and lie how many shots a PGA player would take to hole out from that position.  For example, on the first hole, the data shows a PGA player takes 3.99 shots from 400yds off the tee. From 200yds in the rough, the PGA player takes 3.42 strokes.  For that hole, this player took 3 shots, but only got 0.57 strokes "closer," so the player gets "-2.43" for that drive in strokes gained (0.57 - 3).  Obviously, that "p" hurt- probably an OB tee there.

The lower section shows for that hole how the player "gained" or "lost" strokes for various types of shots.  The colors match the shots above, so you can visualize which shots resulted in those "strokes gained."

Other things that most apps already give:

Fairways Hit

GIR

nGIR (if you specify distance from hole is "near")

All the same putting stats in the spreadsheet above (proximity to hole, strokes gained for various ranges on the green, etc)

Sand saves

None of those stats are as key as strokes gained to me, however. It can show you where you need your practice. Driver, irons, putting, etc.

So my questions are:

Anyone know of anything that already does all this? It sounds simple, but the spreadsheet calculations are somewhat tricky. I enjoy the challenge though- so I might try to do it anyway.

Broadie seems to want to calculate strokes gained for recovery shots separately, but I prefer to lump recovery shots with the bad shot that preceded it. On hole #2 above, for example, it makes sense to say that because of his lousy driving, it took 2 shots to get to 200yds in the fairway. That should all count against your tee shot Strokes Gained, since that was the ultimate cause. Maybe that's a logic error, but it makes sense. Ditto for penalties- lump that in with the preceding shot that caused the penalty, as you see on Hole #1. The penalty is charged to the tee shot. Does that make sense?

Any other cool ideas with this? Other links? Are there apps that calculate strokes gained yet? Broadie was mentioning in "Every Shot Counts" link above that a great app was forthcoming. But I've seen nothing.

All ideas and feedback appreciated.

Good info- your link to the real source was helpful. I don't recall if he was claiming credit for it, but I was unsure about its source from the way his post was written. Makes sense now, because it was lining up exactly with the graphs in the Broadie paper, even though his paper did not specifically include that data chart that I could find.

Yah, I used to participate with "Swing by Swing" as a beta user in their development process, and I discussed with them a ton of ideas about user interface with their data collection. "Strokes gained" was not then anywhere on their radar, as I recall.  I was just starting out with golf again, and I loved the idea of tracking all the stats possible through these GPS apps. Club tracking was the big thing then. Over time, I learned that for me, golf was actually more fun when I was unencumbered by a smartphone. I actually knew how I was playing from my own gut, and the stats didn't often reveal much I hadn't known already intuitively. It turns out I also knew my club distances pretty much already!

It took me a while to figure out I should stop using the apps, because you probably tell, I'm a bit of a nut for data analysis. But I was glad I spent the time diving into the data collection and analysis- as I learned much about what is possible and feasible with collection and what is not. Last season, I went to paper and pencil, and it was total freedom. Golf is a bit more relaxing when you forget about the tools that can intrude.

As for why I do all the spreadsheet stuff, I think it's mainly for the academic exercise. I've always been a numbers guy. Report generation. Data analysis. I enjoy massive amounts of data and poring over it to spot trends. I agree with the limits on the usefulness that you mention, of course. Golf will always be more art than just pure adherence to some mathematical model. What "strokes gained" does for me, however, is provide some glimpse of a ballpark measurement against where different aspects of our games compare with the gold standard. If it helps someone target a weakness, that's pretty cool. Bottom line: we should not be slaves to analysis- just do enough to get a glimpse of where we stand and fix the weakness.

Cool! I had fun with that one. I'll tweak it however you think it might best fit in with the LSW branding.

That putting spreadsheet is pretty straightforward data entry (a minute of typing maybe?) that someone can do for a recent round to get a rough benchmark on their putting. If they choose to analyze multiple rounds, that's possible too, but the eye-opener to me was how simple it is get good putting stats with just typing in the distances of your putts for the day. I'm not aware of too many of the apps today that track putting distance (I think GolfLogix does let you track first putt), and if they do, they certainly don't do all the cool things that spreadsheet does (mainly strokes gained). Over time, the apps will likely all do that stuff though. It's not that hard.

Well, I went ahead and implemented something with strokes gained for all aspects, not just putting. It works basically as described above, but I learned a lot about ways to make it look better for data input, and thought through some of the outputs better.

Here's a sample round from yesterday Deutsche Bank. This was Phil Mickelson's round (shown because he has penalties,  "recovery" shots, bunker shots- all of the different things, typical Phil):

This is the scoring system. You simply fill out each shot with a code, and the rest computes. The code is a combination of distance to hole and the lie you have (or a penalty).

Every shot has a value from the chart in the first post. For the first hole, 365t is about 3.93, and the tee shot went to 82r (82 yards in the rough). That's about 2.97 in the chart above. So Phil took one shot to improve 0.96, so he lost 0.04 shots on that drive.

Every shot can be computed that way.

If you tally it all up, you get this:

Phil lost 2.3 strokes to the "field," but the field was not THIS tournament, it was the combined stats of an entire season from PGA shot tracker. His +3 round, so it appears this course on Sunday played slightly (0.7 strokes) tougher than a typical course.

Since we are tracking ranges of every shot, you can do some fancy things like calculate how well you score from each range, what your proximity to the pin is from various ranges, etc.  Here's Phil yesterday:

As seen above, Phil likely felt that his irons were just a bit off (his blue circles are larger than the green PGA averages). His short game was actually decent, but he missed 3/4 putts after chipping from inside 20yds.

Phil putted pretty well, as shown below. He his a couple putts from outside 15ft that helped. But even though he was 8/13 in the 3-15ft range, he lost half a stroke in that range. Still good putting to gain 1.1 strokes on the field.

He sank 4/9 of his GIR putts. That was better than he did when he was just nGIR or not even nGIR! (1.56 putts/hole GIR, but was 1.61 putts overall- that's rare to see inverted!)

Anyway, I just wanted to show that the project is making headway, and feel free to drop a note if you're interested in more.

I've tested the spreadsheet with several people here on TST, and it does give seemingly worthwhile information on our games... although I'm still looking to make sure the analysis is sound. There could be flaws I haven't thought through.

For interest, I'll post Rickie Fowler's winning round in a hidden thing below (if you want to compare his -3 round with Phil's +3 round).

Rickie excelled at driving (gained 2), approaches (gained over 1), and putting (gained over 1) yesterday.

Rickie's short game was off, but small sample size. He only had 6 opportunities inside 60yds!

117ft of putts made. 2/7 from long range (>25ft): a 34footer and a 38footer. Make those and you get some cushion in the shorter ranges to give up a stroke or two, as he did.

As always, feedback welcome.

Randy did 3 rounds of my data and I found the resulting info interesting and useful.  Ever since I started playing golf, I tracked some stats in spreadsheet - GIR, total putts, FIR, # of pars, birdies, and blown up holes.  I also track ball striking accuracy for wedges, irons, hybrid, wood and driver from range practice.  I used the stat to focus on weak or regressed areas and they served me well.   Randy's tool complements my stat ... and so professional looking, with colorful graphs and all .

One thing I learned more about by using Randy's tool is my putting game.  By recording exact distance from hole,  I've learned most of my 1st putts end within a foot from hole, many just a few inches from it.   I.e, I don't need to practice lag putts so much.  I also missed my share of short putts which I am now focusing more of.

It is difficult to record shot by shot info for an entire round.  I could only do that when it was just my wife and I playing.  When playing four-some on weekends, it proved too much to record all data.  Some holes, I forgot b/c of everything that was going on.  But Randy may have some idea on what to do with skipped holes.   He and I discussed it when he worked on my data.

I hope to use Randy's tool periodically.

Thanks for the comments! I got a lot out of that exchange we had.

I think that the way you might use it is likely the most effective. A spot check from time to time. Do a few rounds that you feel are somewhat representative of how you're playing (I'd not want to analyze or skew that analysis with an outlier round).  Just get an overall feel of a snapshot of where various aspects of your game stand. No need for tracking every shot, round after round after round.

I agree that shot by shot data is somewhat difficult to record, but my thought there is that it doesn't really need to be precise. Typically I can remember for a day each of my shots within some vague tolerance (+/- 10/15yds?), and I'll not record anything on the course.  I've gone tool-free on the course this year. No phone. No GPS. Sometimes I'll jot down my score on a scorecard, but mostly I'm just enjoying the day, and focusing on just the shot ahead. It's liberating!!!

You also gave me a GREAT idea to improve the driving stats, and I'll try to implement that. The hardest part is looking back to see how I coded everything! I can remember shots somewhat after a day or so, but I can't remember spreadsheet code after several months!

For everyone's info, the tool gives a very general idea of your drive length (subtracting distances left to hole), but there may be better ways to distinguish which drives to count from the ones that shouldn't.  I've seen many in Game Golf get frustrated with this same concept too. I've only provided a small snapshot of the reports the tool generates- perhaps soon I'll post all of the output.

On a slightly different topic to help validate this spreadsheet (or invalidate it!): Here's the PGA analysis of the same winning round of Rickie Fowler from Sunday (my spreadsheet calculations of his round in RED )

So my overall take is that no matter how detailed we track our shot data, this tool is extremely approximate. The numbers I entered in for Rickie for this round were the exact PGA shot tracker numbers , but the tool's strokes gained was off by a stroke or so for each area (but we both arrived at 3.7 gained!) when compared to how the PGA is calculating it for that specific day.

We both agree that his short game was the weakest area of his final (and winning!) round. (I've double-checked Rickie's round against Mark Broadie's data shown in the first post of this thread).

So there are caveats with how precise any of it is (and ultimately how useful), but the broad view is that it can give you some sense of where you stand.

So if someone is shooting +17 and is loses 20 strokes according to the tool (say: Drives 7, Approaches 6, Short Game 2, Putting 5)... I'd say that those numbers are probably +/- a couple strokes of reality. That still helps to know (or confirm), doesn't it?

For me, it served as confirmation for what I thought of my game.  For bogey golfer, I thought I had a relatively stronger short game than other bogey golfers.  Conversely, my full swing game is weaker.   The data confirmed both.   I also suspected that my hybrids are more accurate than 6i & 5i.  In fact, my range practice data supports that.  By recording shot by shot data, I've confirmed it, too.   A mild surprises were my lag putt data.   I didn't mention in the previous post but I need to adjust my putt to go a few inches longer.   Too many of my putts just die near the hole and I want it consistently go past 12 - 16 inches the hole when I miss.

To be honest, it is also fun to evaluate and pore over the stat.  That comes from my engineering background, I suppose.

EXACTLY! Now you've hit on why we really do this analysis. The game improvement is secondary, and we are simply geeks.

Yah, those "1g" entries make you wonder- was it a "perfect" miss that went just past, or were you a bonehead and left it short. The PGA listed several putts for Rickie that were under 1ft.  I felt guilty changing his "2 inch" long putt to "1g" (they tell you how long each shot was compared to how far he started, so you can tell if it was long or short).

I struggled with data entry of things like, did you hit the shot left or right of your intended target? Were you long/short of your target? I decided simple = good. For now.

I particularly like the geeky charts on proximity and scoring number by range (has nothing to do with strokes gained).

Glad to hear those confirmed your gut on hybrid/5/6iron performance. Often our gut is right on these things, but sometimes we are surprised with an analysis of the data and we actually learn something counter-intuitive to what we feel. So we geek out looking for something like that! Maybe you found a little surprise in the putting analysis so that's good. You got some value from the work in plotting the data, at least!

So far, my biggest surprise is the small number of strokes that bogey (and bogey plus) golfers seem to be losing in their "Short Game" (/emoticons/tongue.png"/>

Overall, I'm happy that we have a "unit of measure" to track that is meaningful: STROKES.  I like that better than % this or % that. Or #GIR, #putts, etc. Our scores are in "strokes," so let's measure each area of our game in strokes, and this does that- even if lots of caveats and grains of salt in the final number.

One definite reason for the mis-match is that on the long shots (drive & approach), the lateral dispersion from the 'optimal target' is also measured by Broadie's algorithm and can result in significant fractional gains or losses. I see Rickie's htting some good bombs but slightly offline accounting for the stroke difference in SGD and his irons have typically been stronger / morre consistently accurate than his driver. I would bet if you estimated the degrees offline (using center of fairway as target) you would be a lot closer. The shot value per degree offline is in ESC.

Other than drives,the biggest discrepancy was short game. One of the issues with Broadie's SG table is the 'chunkiness' of the data. He is working off a continuous function. How would the SG numbers work using the smoother approximation I shared with you? I'd be curious to see if it was more or less on target than the interpolated 'chunky' values. Also, did you remember to put the shot values in the appropriate Broadie categories (short = < 100 yards)? With shots around the green, I would think the strokes gained value of being on or off the target line is less than with the long game since the pros are generally looking at a putting radius around the hole after a short game shot, though if they flub into a collection area or something, I would expect that location would have significantly higher expected strokes than a putt from the same distance.

Hi,

Just spotted your old post about your spreadsheet for strokes gained. 

Impressive stuff! 

I also use SwingbySwing and been beta testing for them for a while. They started development again (after an ownership change) and said they'd look at my feature request of adding Strokes Gained but I doubt that will happen anytime soon.

So that leads me to your spreadsheet... Do you still have it and if so, can I possibly use it for my own use? Or even a shared google calc (think free online excel) if you want to remain in control of your "algo". 

I'm used to inputting everything in an app so for me the extra work is not a big deal and doesn't slow me down. So hopefully one day SbS integrates it...

Thanks in advance,

Stephane

On 3/1/2016 at 9:38 AM, VoodooZ said:

Hi,

Just spotted your old post about your spreadsheet for strokes gained. 

Impressive stuff! 

I also use SwingbySwing and been beta testing for them for a while. They started development again (after an ownership change) and said they'd look at my feature request of adding Strokes Gained but I doubt that will happen anytime soon.

So that leads me to your spreadsheet... Do you still have it and if so, can I possibly use it for my own use? Or even a shared google calc (think free online excel) if you want to remain in control of your "algo". 

I'm used to inputting everything in an app so for me the extra work is not a big deal and doesn't slow me down. So hopefully one day SbS integrates it...

 

Thanks in advance,

Stephane

Welcome to TST. Hope you stick around and join in. Lots of good info here.

FYI- I'm working on a stripped down spreadsheet on Google Docs that has any proprietary info stripped out of it. Much of my spreadsheet I've worked on was tightly integrated with the ideas put forth in the book, Lowest Score Wins.

Once I do remove any of the info that ties together with the book, I'll put something up on Google Docs, and I'll make the link available to folks on TST who are interested and contribute here regularly. I get occasional requests for a copy of it, but I'm just not comfortable sharing it until it's in proper shape.

Plus, there are other places on the internet where folks are giving away their copies of Strokes Gained spreadsheets, if you search hard enough (google "strokes gained spreadsheet golf forum" and you'll likely see). I just did this one for fun, and I'm not much interested in being a clearinghouse of ideas to expand/enhance/distribute changes, etc. I did have a blast testing it out with a bunch of folks here to see how their golf games were analyzed by the tool. I learned a lot from them, and made some tweaks for the better. I still have a list of things to fix and improve that I wonder if I'll ever get to!

The other strokes gained spreadsheet I'm thinking of on the internet (and with whom I've spoken a bit in the past about its development) is much more complete than mine. We took very different approaches from the start, but it has a lot of advantages. I took a bunch of design "shortcuts".

For example, once a round is entered, I simply turn that round into metadata, and you can't enter that round data anymore.  Theirs is one long list of every single shot, so you can edit every shot you've ever taken. Theirs is much easier, as a consequence, but it recalculates a bit more slowly since it must refresh the entire list of shots. Mine is kind of kludgy to enter a round into your list of rounds to do summary calculations, because I couldn't figure out a graceful way to do that. But since it was mainly for my own consumption, no big deal.

I think my stripped down version will be simply a strokes gained calculation of one round only. No multi-round summaries. Nothing fancy. Just the bare bones of how well you played that round broken down by drives/approaches/short game/putting. If anyone wants to build more from there, they'd be welcome to it. Hopefully I can finish that soon, but I keep getting distracted.

On 10/27/2015 at 0:03 PM, natureboy said:

Other than drives,the biggest discrepancy was short game. One of the issues with Broadie's SG table is the 'chunkiness' of the data. He is working off a continuous function. How would the SG numbers work using the smoother approximation I shared with you? I'd be curious to see if it was more or less on target than the interpolated 'chunky' values. 

Good stuff. Obviously, I got sidetracked and put this mostly off to the side! I did put your Short Game info into the tables, and I believe that it didn't dramatically change many of the numbers by too much for subsequent rounds. I didn't do a thorough analysis, I must admit. But it wasn't a huge change from my anecdotal rounds I compared. If I get back into it soon, I'll double-check that.

On 3/2/2016 at 5:01 PM, RandallT said:

Good stuff. Obviously, I got sidetracked and put this mostly off to the side! I did put your Short Game info into the tables, and I believe that it didn't dramatically change many of the numbers by too much for subsequent rounds. I didn't do a thorough analysis, I must admit. But it wasn't a huge change from my anecdotal rounds I compared. If I get back into it soon, I'll double-check that.

Interesting. I wonder if Broadie published an older table and is working off different / more accurate baseline #'s now. One thing for sure is he uses a multi-year baseline method that I'm pretty sure always includes the most recent years. As soon as ESC was published, his baseline numbers were going to change from using recent real-world data while the table in the book stayed static. I'm surprised a few years of extra data would shift it the numbers that much, though. Could still also be some contribution from cumulative small differences from the degrees offline component on all shots too. Do you use a different short game distance than Broadie (75 vs 100 yards) - that's a diff of ~ .75 shots / rnd. I would expect the putting numbers to be closer, but Broadie may apply a course-specific adjustment. 

Thanks RandallT. 

I found a few ones and learned a lot in the process.

I made the request to SwingbySwing developers to have such a feature added as it already has GPS and course mapping so adding a simple quick GUI for tracking distance, lie and club wouldn't take much added effort and would make data entry much faster than paper for me. Their app is dirt cheap so it beats having to pay for yet another app/subscription.

Cheers,

Steph

17 minutes ago, VoodooZ said:

Thanks RandallT. 

I found a few ones and learned a lot in the process.

I made the request to SwingbySwing developers to have such a feature added as it already has GPS and course mapping so adding a simple quick GUI for tracking distance, lie and club wouldn't take much added effort and would make data entry much faster than paper for me. Their app is dirt cheap so it beats having to pay for yet another app/subscription.

 

Cheers,

Steph

Hi Steph! Glad to hear it. I used to use SbS, and I enjoyed it. I signed up for their beta dev team, so you might ask to see if you can join that. You get an inside track into the wish lists. Plus you get the latest trial versions of their phone app and give them feedback before the rest of the people get it. I'm not positive it still works the same way, but it was quite fun at the time.

Yep. It's still like that. I've been on the list for a few years now too. I'm a big fan of open source so I like the openness, especially for a closed software. 

After the take over by the Back9 network they had to pause development but they apparently back at it so I'm not holding my breadth but considering I've invested years in the project and all my stats are there I'd rather not switch to another app just for SG stats.

Cheers,