GAME GOLF Inaccurate Strokes Gained Putting?

[kpaulhus]

I created the table below based on how the PGA Tour calculates Strokes Gained putting, using my most recent round. 

For some reason, GameGolf is telling my my Strokes Gained (or lost) was -4.62 strokes for this round against a scratch golfer. Seems quite a bit off considering the actual calculation is -3.02 against the PGA Tour Average. That's over a 1.5 stroke difference. I don't know that I would take the time to do this for every round I play, but it seems that GG is quite a bit off when I spot checked a few more rounds.

 

 

[bweiss711]

It's cool to see you actually put some numbers to the inaccuracies.  I believe the issue is that GG only uses its users' data to calculate strokes gained/lost.  And I assume that most golfers do not edit their round afterwards and move the pin placement on each hole.  I imagine this leads to the issues with putting you demonstrated here.  Not sure what could be done about it.

[iacas]

Just now, bweiss711 said:

It's cool to see you actually put some numbers to the inaccuracies.  I believe the issue is that GG only uses its users' data to calculate strokes gained/lost.  And I assume that most golfers do not edit their round afterwards and move the pin placement on each hole.  I imagine this leads to the issues with putting you demonstrated here.  Not sure what could be done about it.

That's precisely the reason, as I understand things.

@RandallT and I have been working with them on this, but progress is… slow.

SG Putting numbers seem to be off compared to PGA Tour players by 1 to 2 strokes per round. That's substantial, particularly when you consider that it's PGA Tour vs. Scratch.

[kpaulhus]

If the GG calculation was correct, it wouldn't matter if people edit their rounds. The formula is calculating how far they are from the pin, and even when I edit my rounds correctly (I do 95% of the time) to show distance its still a bit off from what GG states vs the real strokes gained data provided on tour. 

We all know from several discussions on TST that the biggest skill gaps between a PGA Tour pro and the common amateur isn't putting, its the long game.

Thats why its weird that I would have "lost" 4.6 strokes to a scratch golfer but only 3 to a PGA tour player. If anything my off the tee or approach should have been much higher than my putting difference.

[No Mulligans]

2 hours ago, kpaulhus said:

If the GG calculation was correct, it wouldn't matter if people edit their rounds. The formula is calculating how far they are from the pin, and even when I edit my rounds correctly (I do 95% of the time) to show distance its still a bit off from what GG states vs the real strokes gained data provided on tour. 

We all know from several discussions on TST that the biggest skill gaps between a PGA Tour pro and the common amateur isn't putting, its the long game.

Thats why its weird that I would have "lost" 4.6 strokes to a scratch golfer but only 3 to a PGA tour player. If anything my off the tee or approach should have been much higher than my putting difference.

My understanding is Game Golf calculates the stats based on their database of Game Golf players (not based on PGA stats).

The problem is that if everyone doesn't edit their putting for all their rounds then the database is off.  I'd guess that this tends to show that on average players are making longer putts than actual.  For example, when I edit my round I'll change a displayed 20 foot made putt to an the actual distance of 1 foot, to many are leaving that as a 20 foot made putt.

I think the newer feature on Game Golf live of being able to edit the hole location which then keeps that hole location for others should clean up the data to some extent (if I'm understanding that feature correctly).  At some point they need to delete the old data to improve the database.  They could also run a program to exclude the wrong data, for example if someone often has a second putt that is much longer than their first putt, then the putting data from that round/player should be excluded.

[iacas]

3 hours ago, kpaulhus said:

If the GG calculation was correct, it wouldn't matter if people edit their rounds. The formula is calculating how far they are from the pin, and even when I edit my rounds correctly (I do 95% of the time) to show distance its still a bit off from what GG states vs the real strokes gained data provided on tour. 

I think you missed the point of what he was saying… @No Mulligans clarified:

1 hour ago, No Mulligans said:

The problem is that if everyone doesn't edit their putting for all their rounds then the database is off.  I'd guess that this tends to show that on average players are making longer putts than actual.  For example, when I edit my round I'll change a displayed 20 foot made putt to an the actual distance of 1 foot, to many are leaving that as a 20 foot made putt.

That's it.

In talking to @GAME GOLF, they've said they only add rounds to be included when they see that the person edited the round, but even still, either people grossly mis-estimate how long the putts they make are, or something.

I think @RandallT has found that per the database of stats they're using, @GAME GOLF users are slightly worse than PGA Tour players from 10 feet and in, but quite a bit better from longer ranges.

If you chip to five feet and hole your putt, but GG (due to GPS inaccuracy, a wrong hole location, etc.) believes you've chipped to 18 feet and holed that putt, it's going to skew both the short game stats and your putting stats (the wrong hole location could also mess with your approach stats, but minimally - you've still missed the green, you just might be 25 yards from the hole instead of 21 yards, or something).

Still, it seems to me that if the stats they're giving you are so far out of range for a PGA Tour player, they should almost cap them at that. I doubt scratch GGers are putting so much better than PGA Tour averages… so just cap them at PGA Tour averages if any particular stat is saying "we're better at this than they are." It's highly, highly unlikely to be true for any average of a large enough group.

---

As I mentioned, @david_wedzik and I, along with @RandallT a bit, are working with them on this and other things.

[Dave2512]

I've encountered a few GG users since I've had one and all were mostly interested in "how far", club tracking stats. Those that see me using it always ask the same thing, does it track drives. I'd be surprised if many to most signed rounds regarding putts are only accurate to the number of putts rather than where from. Is it even possible to correct that? And even then if it was accurate to dead on does it matter as a game evaluation tool the answer is going to be improve ball striking for all but a few. Nothing about GG would change my putting practice.

 

[Jeremie Boop]

7 minutes ago, Dave2512 said:

I've encountered a few GG users since I've had one and all were mostly interested in "how far", club tracking stats. Those that see me using it always ask the same thing, does it track drives. I'd be surprised if many to most signed rounds regarding putts are only accurate to the number of putts rather than where from. Is it even possible to correct that? And even then if it was accurate to dead on does it matter as a game evaluation tool the answer is going to be improve ball striking for all but a few. Nothing about GG would change my putting practice.

 

For awhile I was very diligent in editing my rounds and making sure my putting distances were as close as I could get them to be. In the end I saw pretty much no impact in my sv on putting so I kind of just let it go. As long as the number of putts was correct and reflected on/off the green as required I didn't see the point in the extra work if the system didn't work for the sv.

 

**Basically I got lazy**

[No Mulligans]

20 minutes ago, Jeremie Boop said:

For awhile I was very diligent in editing my rounds and making sure my putting distances were as close as I could get them to be. In the end I saw pretty much no impact in my sv on putting so I kind of just let it go. As long as the number of putts was correct and reflected on/off the green as required I didn't see the point in the extra work if the system didn't work for the sv.

 

**Basically I got lazy**

It's kind of like getting the flu shot...  If enough of us do it the herd becomes protected/benefits.

[Dave2512]

25 minutes ago, Jeremie Boop said:

For awhile I was very diligent in editing my rounds and making sure my putting distances were as close as I could get them to be. In the end I saw pretty much no impact in my sv on putting so I kind of just let it go. As long as the number of putts was correct and reflected on/off the green as required I didn't see the point in the extra work if the system didn't work for the sv.

 

**Basically I got lazy**

I solved it by not using GG anymore. Through 2 seasons it didn't tell me anything I didn't know, didn't change the way I practiced.

[kpaulhus]

1 minute ago, Dave2512 said:

I solved it by not using GG anymore. Through 2 seasons it didn't tell me anything I didn't know, didn't change the way I practiced.

I have had GG for 2.5 years now, and while it isn't super beneficial to me right now (because it has taught me what to practice) I still use if for 95% of my rounds. Its cool info and I like to see how far drives are and whatnot. Its relatively close to the stats I keep on my scorecard. We have started using the live scoring feature within my groups on the weekend as 3 other guys have GG as well.

[iacas]

2 hours ago, Dave2512 said:

And even then if it was accurate to dead on does it matter as a game evaluation tool the answer is going to be improve ball striking for all but a few. Nothing about GG would change my putting practice.

Yeah, it matters. If you're three-putting, is it because of poor distance control? Bad reads? Or missing more short putts? Do you get enough putts to the hole, particularly from inside 15 feet? Or do you come up short too often?

There's a lot of data saved but not necessarily exposed right now. Entering in the data now just stores it away for the future. And with the ideas @david_wedzik and I have, and the ideas they have in the pipeline, you might regret not keeping up.

2 hours ago, Jeremie Boop said:

For awhile I was very diligent in editing my rounds and making sure my putting distances were as close as I could get them to be. In the end I saw pretty much no impact in my sv on putting so I kind of just let it go. As long as the number of putts was correct and reflected on/off the green as required I didn't see the point in the extra work if the system didn't work for the sv.

They still have the stats, and as they further develop the features, the potential exists to see more and more fine-grained stats. Maybe you'll find that you putt better at one course than another, or you're pretty good until you get to 25+ footers. Or you're more likely to three-putt in the morning when there's still dew on the greens (maybe).

Plus, @Jeremie Boop, editing the rounds shouldn't take you very long. If you get good at it, it takes five minutes to adjust everything. Five minutes where you get to re-consider your round, too. Time well spent.

[Hardspoon]

1 hour ago, iacas said:

There's a lot of data saved but not necessarily exposed right now. Entering in the data now just stores it away for the future. And with the ideas @david_wedzik and I have, and the ideas they have in the pipeline, you might regret not keeping up.

So...this is the biggest reason that I religiously edit each round.  There's no reason not to, and I'm excited to see what they can do with it some day (I'd love it if they would open-source the raw data, but that's probably never happening unless someone hacks their GG unit).

Spoiler

OT: In life in general, I'm basically the opposite of a pack-rat; I almost religiously discard things without immediately obvious utility.

But...when it comes to data, I'm a collector.  I will save spreadsheets of random information and then feel like I have to keep it organized and up-to-date.  Right now, on my phone, I have a list of the exact time that I caught the trolley to work (and home) each day...

 

[glinks]

On 05/10/2016 at 8:16 PM, No Mulligans said:

My understanding is Game Golf calculates the stats based on their database of Game Golf players (not based on PGA stats).

do GG uses the data of its users to calculate SG vs 5 Hcp, 10 Hcp, etc ...?

[iacas]

3 hours ago, glinks said:

do GG uses the data of its users to calculate SG vs 5 Hcp, 10 Hcp, etc ...?

Currently, yes.

It's pretty accurate given how much of a course rating comes from the distance.

[RandallT]

On 10/5/2016 at 8:11 PM, Hardspoon said:

So...this is the biggest reason that I religiously edit each round.  There's no reason not to, and I'm excited to see what they can do with it some day (I'd love it if they would open-source the raw data, but that's probably never happening unless someone hacks their GG unit).

  Reveal hidden contents

OT: In life in general, I'm basically the opposite of a pack-rat; I almost religiously discard things without immediately obvious utility.

But...when it comes to data, I'm a collector.  I will save spreadsheets of random information and then feel like I have to keep it organized and up-to-date.  Right now, on my phone, I have a list of the exact time that I caught the trolley to work (and home) each day...

 

Agreed re: "no reason not to." Some day, it'll pay off. I think it'll be exciting to see what they can come up with.

Great thread! Just catching up things from some time away from thinking about golf.

 

On 10/5/2016 at 4:02 PM, iacas said:

I think @RandallT has found that per the database of stats they're using, @GAME GOLF users are slightly worse than PGA Tour players from 10 feet and in, but quite a bit better from longer ranges.

Minor tweak to this statement: it turned out that 5ft was the threshold. Like you say, one needed to putt a bit better than a PGA pro outside 5ft just to avoid losing strokes in GAME GOLF. Inside 5ft, their numbers seemed to be closer to what we'd expect: a tad worse than a PGA pro putts.

 

And for those new to this topic or new to the forum, lots of folks here worked with me last year to do our own little home-cooked strokes gained analysis across the different areas of the game. @kpaulhus's analysis in the OP aligns EXACTLY with what we saw time and time again- roughly 1.5 to 2.0 strokes difference between strokes lost putting (GG scratch v PGA). 

Interestingly, at the time we analyzed last year, it was at about the 10ft mark that GAME GOLF expected you to hit 50% of the putts to break even for a scratch player (the magical 1.50 strokes to hole out point). As we know for pros, that benchmark is typically 8ft. 

If anyone ever wants to confirm if they've tweaked their reference data set, just set 18 putts to 8ft. Make them all.  If you gain 9 strokes, then you've found the 50/50 point. If you gain fewer than 9 strokes when all 18 putts are 8 ft, then GAME GOLF expects you to make more than 50% from there. Try 9 ft for all 18 putts. Then try 10ft. When you get 9 strokes gained in the report for the round, you've found GAME GOLF's 1.50 expected hole out distance.  Oh, and then delete that round because you're corrupting the data sets, man!

For what it's worth, I used 3 different courses and found that on all 3 courses, the 10ft putt length was the 50/50 point- consistent across all 3. Probably overkill, but I wanted to make sure that they didn't use data specific to just one course, but sure enough, it seemed like they used the same reference data for all greens. No surprise, but it was worth the check since I was already geeking out. I thought maybe they had such a massive data set that they could've broken it down more! 

Note: I've signed a non-disclosure agreement with GAME GOLF (which I take seriously), but all of this info above is readily available from just using the system. I've not included anything above that I might've learned from any of our discussions where I mostly just provide a voice to show them how a stats geek experiences their product. 

 

 

 

[iacas]

I will add to @RandallT's post above that @GAME GOLF is aware of the discrepancy, and I expect that it will be improved shortly.

And retroactively.

[cutchemist42]

On 05/10/2016 at 5:57 PM, iacas said:

Yeah, it matters. If you're three-putting, is it because of poor distance control? Bad reads? Or missing more short putts? Do you get enough putts to the hole, particularly from inside 15 feet? Or do you come up short too often?

There's a lot of data saved but not necessarily exposed right now. Entering in the data now just stores it away for the future. And with the ideas @david_wedzik and I have, and the ideas they have in the pipeline, you might regret not keeping up.

They still have the stats, and as they further develop the features, the potential exists to see more and more fine-grained stats. Maybe you'll find that you putt better at one course than another, or you're pretty good until you get to 25+ footers. Or you're more likely to three-putt in the morning when there's still dew on the greens (maybe).

Plus, @Jeremie Boop, editing the rounds shouldn't take you very long. If you get good at it, it takes five minutes to adjust everything. Five minutes where you get to re-consider your round, too. Time well spent.

I sometimes check out their forum for user requests. I know I personally gave full votes to somehow integrating Stableford scoring. Saw lots of UK users who want it on that forum.

Another I had was being able to mark a shot as an attempt at a 1/2 or full PW/SW/etc, and see how consistent your yardages are when you attempt thoseshots. Basically just a different way of displaying the wedges instead of whats currently there, yard ranges from 5 yards to 120 yards for PW.

Im just curious what ideas you have for it?

Edited October 19, 2016 by cutchemist42