Feel like strokes gained (or maybe just Arccos?) isn't helpful

[Jtgavigan]

Ok, before I go off on this, a little background on me:

 

1. Electrical engineer by degree, so math is not an issue. 

2. I own a company that does data analysis for industrial companies and am very good at looking and gleaning information from data. 

 

So, not that I am a sports statistics nerd, but I feel like there are issues with the SG metrics, or maybe just the Arccos app. For the PGA and other professional tours, SG makes perfect sense. Everyone plays the same course under essentially the same conditions. 

 

For us, we rely on an app that measures us against a mean of some handicap level. This is includes all shots on all courses under all conditions everywhere. This just seems flawed, but I guess a case can be made that if there is enough data, anomalies will filter out and everything will tend toward the mean.

 

I am playing to around a 3 handicap or slightly better at the moment and measure myself against scratch. I like tracking the stats because I can at least see some places I need to work on (but not sure I need SG to tell me where). So, not that i find the Arccos stats completely worthless, just misleading. 

 

Enter today as my case in point. Cool, breezy day here in NE Florida. I go place a course I play reasonably often. Tight track. Scorable if you are hitting it straight off the tee. Nightmare if not. Decent day. Shoot 75. Lipped two 3 footers for par I should have made. I doubled a par 5 after a bad drive and an approach into the water after punching out. I also had a bad drive on 18 that led to a bogey. 

 

I hit 8 fairways and 11 greens. I also was on 3 fringes. Didn't hit many super close, but nothing too far away and felt really good about all but a few shots. 

 

Go look at SG approach against a scratch and I am losing 3.4 shots. Hmmm....

 

Go dig a little deeper...

 

See these....

 

 

interesting that I DO have a 50 yard approach that I stuffed to 3 feet. I also had a 72 yarder downwind I hit long. Had maybe 40 feet left. Chipped on and missed the 3 footer. Maybe Arccos just doesn't recognize the 50 yarder? But, on average, I would say that is a wash, but I am losing 0.7 shots to a scratch from those shots. Hmmm.. well, originally the 50 yarder was marked at 42 yards on the app and I changed it afterwards to reflect what the rangefinder told me. 

 

Looking at a few other stats, I am closer on average in every category. I know the penalty I took for the water ball (3w, 244, wind L to R and slightly hurting hurts), but give most scratch players that shot and it is probably 3.75 shots on average. I took 5, so I took a bath there. I get it. I offset that with a 230 yard one into the wind where I got down in 3 when I stuffed a 30 yard approach to a foot. So, maybe a slight loss there. I also made birdie from just over 200 on another par 5 with a long putt. Those approaches were both reasonable in the conditions today. 

 

So, other than the one egregious water ball, my approaches were pretty good today. 

 

 

Walking away from this round, I am thinking about these things to improve on:

 

1. Still need to tighten my 120 and in game a little, which matches to a degree, but don't think I lost 2 strokes there today. Maybe 1...if that. I need to give myself more reasonable birdie chances.

2. Had two drives get away from me the last few holes. One a pure lapse in focus (on 18).

3. Still need to work on 3 to 10 foot putts. That also matches. 

 

So, trying to understand if this is an Arccos software issue or the math itself. The quick overall math isn't mathing in my head. I have played at scratch before and played with some scratch to plus 2's and just don't see where I would have lost 3 and a half strokes approach today. 

 

Any comments? I had some thoughts of where I needed work today and overall, approach play was way down that list. 

 

 

 

I actually think this metric below is dumb. I hit 3w several times and hybrid 4 times off the tee today. Distance across all drives is a ridiculous metric. This really should be broken down by club or something. I could go to a different course with hard fairways and hit driver 14 times and average way more than this. Had to punch out twice today, so I am OK with the accuracy number below.

 

 

 

No issues here:

 

 

 

This is funny as short game hasn't been a strength but today was a good day. Not sure this good. To be honest, my approach play was good and had 3 or 4 putts from just off the green that could be counted as "chips" so again, I feel like I got rewarded and penalized in the wrong places. Approach play was better than above and the short game wasn't THAT good honestly. 

 

 

 

So, ultimately,  is this a SG issue or Arccos Software issue? 

 

I just know I was disappointed when I opened the Arccos app and saw the SG Approach loss.

 

Curious as to others' experiences. 

[Jtgavigan]

Here was a day that I really sucked on approach. Lost 3.3 strokes. I hit 7 greens and only had one legit birdie look. Just a bad day. 

 

Really good driving day. Hit 3w quite s bit but man I drove it on a string...

[Jtgavigan]

Here is one where I played the RTJ course in Auburn,  AL. Lots of downhill tee shots, firm fairways, hit driver more. Reasonably wide fairways. 

 

The greens were VERY fast (12 to 13 on the stimp if I had to guess) and undulating. Miss in the wrong spot, it is a 3 putt. Hit a great wedge to 12 feet and 3 putted. Had almost no chance. 

 

The stats told me: "World beater driver, crappy putter." But, man, I had some really tough putts not knowing the course. 

 

Just apples and oranges to the round today. Soft fairways, more wind, and slower greens (maybe an 8).

 

If I played the RTJ course all the time, the data would bias toward me being a good driver/bad putter. If I played the course I did today all the time, then I wouldn't be a great driver, ok apprpach, decent short game and putting.

 

I wish we could measure against the same ability player at the same course. 

 

Edited December 27, 2024 by Jtgavigan

[betarhoalphadelta]

Also an electrical engineer, so likewise math isn't an issue...

 

A few ideas:

 

1. You say you're playing off a 3 right now, and you lose 3.4 strokes to a scratch... Given course ratings, isn't that about normal for a 3? 
2. It seems like you're looking at your analysis of a single round (or two) and questioning Arccos. As someone who does data analytics for a living, you have to realize that sample size matters, and 1 round is going to have various outliers that smooth out over time. 
3. As someone who does data analytics for a living, you have to realize that the data, or even SG analysis, is a starting point. Teasing out the relevant information out of that data is something that you have to do, even if SG analysis is trying to do it for you. 

Ultimately SG analysis is but one prism to look at your golf game amongst many. It's not an authority. It's one tool in the toolbox. 

 

If you're a EE with a career in data analytics, you should be more than equipped to figure out what things SG tells you that are relevant and what aren't.

[Birdman03]

I think you have to look at these numbers over a long time window such as 10 or 20 rounds.  On a given day your putting may be off because you are not hitting it close on approach shots, or you just don't have a feel for the speed.

Edited December 27, 2024 by Birdman03
Words

[Pnwpingi210]

100-125 approaches….did you have any big misses?  Did you stick any tight?

 

your average was 24 feet which is similar to the target for scratch.  
 

with stokes gained there’s a big benefit to hitting it inside 6-8 feet, and then diminishing returns outside of that very quickly up to like 25-30 ft, and then it starts to get more penalizing again. 
 
usually when I have a round like that where arrcos tells me I lost strokes in a category when I think I played well it’s one or two shots that bad (land on green or are 40 ft from pin) that do it.

 

penalty strokes always seem to cost you 2-3 stokes gained if you are low single digits.

 

as others have said, it’s a starting point and is only as good as the data set that gets input into the system.  I think it continues to improve.  You also have to ask yourself, do you every question where it says you gained strokes?  Or just where you are losing them?

Edited December 27, 2024 by Pnwpingi210

[Red4282]

I cant speak for arcoss or how it works, but I do all my stuff manually. I try to log all my shots-the best i can then get a strokes gained chart and fill out a spreadsheet. Not sure if this is correct or not but say Im 150 with a 20 mph headwind and its playing 180. I log it as a 180 shot. Same for downwind. Like someone else pointed out, if you lost 3 strokes to a scratch as a 3 cap, Id say good round. You also need to take single round results with a grain of salt. Sometimes its telling you something, sometimes its not. But after enough rounds you should see a decent trend or consistency develop. 

[Jtgavigan]

2 hours ago, betarhoalphadelta said:

Also an electrical engineer, so likewise math isn't an issue...

 

A few ideas:

 

1. You say you're playing off a 3 right now, and you lose 3.4 strokes to a scratch... Given course ratings, isn't that about normal for a 3? 
2. It seems like you're looking at your analysis of a single round (or two) and questioning Arccos. As someone who does data analytics for a living, you have to realize that sample size matters, and 1 round is going to have various outliers that smooth out over time. 
3. As someone who does data analytics for a living, you have to realize that the data, or even SG analysis, is a starting point. Teasing out the relevant information out of that data is something that you have to do, even if SG analysis is trying to do it for you. 

Ultimately SG analysis is but one prism to look at your golf game amongst many. It's not an authority. It's one tool in the toolbox. 

 

If you're a EE with a career in data analytics, you should be more than equipped to figure out what things SG tells you that are relevant and what aren't.

 

I lost 1.3 overall today. The course was 72.3/136, so yeah, reasonable.

 

However, I gained a lot on short game and lost a lot on approach, just because of how the numbers are calculated. I guess over time it will work out better, but my other two posts point out that depending on the course I play, how much I hit driver, or even how tight the course is, it skews the numbers. 

 

I am not saying that it doesn't have some value, as it does offer relative weaknesses,  but I could "game" the numbers to a certain extent that won't accurately reflect what I am actually struggling with. I could play on very fast and very undulating greens and may be a good putter but lose strokes gained against a scratch handicap playing slower, less undulating greens, and then I could beat myself up that I am not a good putter, which is not true. Or, I could play a tight course where driver is almost never the play and my driving game would look bad.

 

To your point, it is just one tool and I will need to learn how to use it effectively. 

[Jtgavigan]

2 hours ago, Pnwpingi210 said:

100-125 approaches….did you have any big misses?  Did you stick any tight?

 

your average was 24 feet which is similar to the target for scratch.  
 

with stokes gained there’s a big benefit to hitting it inside 6-8 feet, and then diminishing returns outside of that very quickly up to like 25-30 ft, and then it starts to get more penalizing again. 
 
usually when I have a round like that where arrcos tells me I lost strokes in a category when I think I played well it’s one or two shots that bad (land on green or are 40 ft from pin) that do it.

 

penalty strokes always seem to cost you 2-3 stokes gained if you are low single digits.

 

as others have said, it’s a starting point and is only as good as the data set that gets input into the system.  I think it continues to improve.  You also have to ask yourself, do you every question where it says you gained strokes?  Or just where you are losing them?

On the 100 to 125, neither. I didn't stick any or have any major misses.

 

Penalty strokes are certainly a killer and I have been really working in eliminating them. I didn't need Arccos for that 🤣, but it made it even more painfully obvious how much I lose.

 

Yes, I questioned today that my short game gained me 3.5 strokes while my approach lost 3.4. It was only because I putted off the fringe 4 times and hit at least one nice chip. So, my short game was skewed positive. 

[Pnwpingi210]

17 minutes ago, Jtgavigan said:

On the 100 to 125, neither. I didn't stick any or have any major misses.

 

Penalty strokes are certainly a killer and I have been really working in eliminating them. I didn't need Arccos for that 🤣, but it made it even more painfully obvious how much I lose.

 

Yes, I questioned today that my short game gained me 3.5 strokes while my approach lost 3.4. It was only because I putted off the fringe 4 times and hit at least one nice chip. So, my short game was skewed positive. 

Interesting on the 100-125.  I know they added some advanced analytics on the web based dashboard vs the app.  It’s possible you can get some more detail there.  
 

but more to your last post, looking at trends over the last 5-10 rounds is probably a better way to use the data vs 1 round

 

 

 

[DLiver]

As someone else mentioned, SG isn't very helpful when comparing a single round to the SG averages. You really need to be looking at 5 rounds minimum (10 is better) to get a clear picture of your game. As you've noted, a single exceptional (or exceptionally poor) shot can produce odd results when looking at a single round.

 

Look at it this way: you are comparing yourself to SG of millions of golfers playing 100s of millions of rounds. All of their anomalies, like hitting a 150 yard shot to 2 feet, have been smoothed out. However when you compare a single round of yours to these averages, your anomalous shots will produce SG data that is not indicative of your skill level over time.

 

I have found SG to be very helpful in highlighting areas of my game that needed improvement. I discovered my iron play was exceptional for my level (2-3 index, but my iron play was like that of a +3-+4), and my putting was terrible (putting like a 10 HC). Both of these were quite surprising to me. I thought I was a mediocre iron player and an above average putter. SG is an interesting metric, but as many have pointed out, it isn't perfect.

[RCGA]

I think tracking strokes gained is a waste of time unless you can practice a ton (like someone in high school or retired). 

 

Let me guess, every area of your game needs work. I'd say you could also identify your weaknesses without tracking SG.  

Edited December 27, 2024 by RCGA

[iacas]

19 minutes ago, RCGA said:

I think tracking strokes gained is a waste of time unless you can practice a ton (like someone in high school or retired).

 

I think tracking strokes gained is great for people who don't have a ton of time to practice, because you can focus your practice more precisely on the areas that need the most help.

[Golfonthemind]

I sometimes use a strokes gained app and it seems really good for just about everything besides short game. I am guessing that’s just because the app doesn’t have a ton of data from my course. I will be 10 feet below a green short sided with a bunker between me and hit a phenomenal flop shot to 15 feet and the app says that shot lost .8 or something to a scratch player. 

[Ty_Webb]

15 minutes ago, iacas said:

 

I think tracking strokes gained is great for people who don't have a ton of time to practice, because you can focus your practice more precisely on the areas that need the most help.

 

This - if you have a ton of time to practice, you can practice everything and it won't really matter all that much where your weaknesses are. If you don't have a lot of time to practice, you'd be crazy to spend all the time practising your driver if your driver is already good, but your approach play is poor. Much better to practice your approach play with your limited time.

[Jtgavigan]

30 minutes ago, RCGA said:

I think tracking strokes gained is a waste of time unless you can practice a ton (like someone in high school or retired). 

 

Let me guess, every area of your game needs work. I'd say you could also identify your weaknesses without tracking SG.  

Yeah, without SG, I would say putting hasnt been as solid as it needs to be and keeping the ball in play has been an issue. I also don't give myself enough good looks from 100 yards and in. 

 

So, have focused a lot on swing technique that helps me keep the ball in play. Have been working much more on wedges during practice time. Putting will be next, as I still have some good days there. I just know I get more consistent with practice. 

 

SG reflects almost all of that. 

 

As someone mentioned above, there is a dashboard on the website also. That is more helpful than SG, even though I don't think all the data is right. Some of the visual also aren't very helpful. I am hoping they have an API that i can use to get my data. 

 

I also knew my mid irons were hurting me too. I miss left a lot with 8i, which is odd and need to keep track ofMy only saw that on the web dashboards. My shot pattern is skewed too far left at the moment for sure. 

 

 

[Jtgavigan]

3 minutes ago, Golfonthemind said:

I sometimes use a strokes gained app and it seems really good for just about everything besides short game. I am guessing that’s just because the app doesn’t have a ton of data from my course. I will be 10 feet below a green short sided with a bunker between me and hit a phenomenal flop shot to 15 feet and the app says that shot lost .8 or something to a scratch player. 

That is exactly my point. 

 

I get that it is trying to quantify EVERYTHING into some algorithm,  and it is just impossible. 

 

I might go out and shoot 75 on a tough track in cool and windy conditions and it will tell me how I lost strokes to a scratch player.

 

[LongNwrong27]

10 minutes ago, Jtgavigan said:

That is exactly my point. 

 

I get that it is trying to quantify EVERYTHING into some algorithm,  and it is just impossible. 

 

I might go out and shoot 75 on a tough track in cool and windy conditions and it will tell me how I lost strokes to a scratch player.

 

That’s one round though and noise in grand scheme of things if you do this for a couple months or a season. If you feel the conditions are significantly different than normal I would even propose not including that round in your stat tracking purposes. Obviously hitting a 150 yd approach in 25 mph wind is going to be different than on a normally calm ish day. 
 

It’s not going to tell 100% of the story but it can generally tell 90-95 and you can figure out the rest. 

 

[LongNwrong27]

I guess I don’t really see the problem here. It seems like the data is consistently telling you that your approach play is the main area you lose strokes to scratch players. Do you think that is true or not? If not then where do the shots come from 

[Jtgavigan]

As several have stated, and I agree to an extent, that 5 to 10 round averages are more telling.

 

However, I don't think it is out of bounds to call out that single round stats should be better or they can be almost ignored. 

 

Let's say I have a customer who makes some product via a batch process. If I average out 10 batches, the data looks fine, like they are producing good product. However, the data swings wildly between batches because of crappy instrumentation, or fluctuations in their process. I need to help them pinpoint the anomalous batches and understand why that is happening. I have to have a good way to do that on a batch by batch basis and help them either fix their instrumentation or their process. They could send out a lot of bad product that were made in the bad batches, but if I tell them "it will all work out in the wash, your 10 batch averages all look fine," I have misguided them.

 

I consider rounds like my customers' batches. I should be able to have accurate data on a round by round basis. But as I see it today, that isn't possible unless a bunch of scratch and low handicap players go play the same course I am playing on the same day and we all measure everything. Then, the metrics will make sense. Otherwise, I have to take them with a grain of salt every time. The algorithm can't sense the difficulty of a shot from the same distance.  You could have a shortsided shot like @Golfonthemind stated on a really quick green vs a 15 yard shot that is straight forward on a moderate speed green.

 

If I consistently play course with hard and fast greens and difficult rough and short side myself a lot, it might tell me I have a terrible short game, when in reality, it is my approach play leaving me in a pickle all the time. My proximity may be fine, but the reality is i am leaving myself no chance at par a lot.

 

I compared one of my "best" ball striking rounds to yesterday's. I had a few more close shots with irons, but on avearge, I wasn't really that much closer. There was almost a 7 SG difference approach between rounds. I shot 74 on my best approach round and 75 yesterday,  yet felt I had more control over the ball yesterday. I had a really "bad" putting day on my "best" approach day, but the greens there are very quick and undulating. So, 3 putts can be expected if you hit it in the wrong places. 

 

I think I am finding the most benefit in the online dashboards where I can see better how individual clubs are performing. Missing left so much with an 8i is a stat to keep an eye on. I have also been missing left with 6i too. In all, my shot pattern is skewed too far left on approach which is valuable info.

 

Edited December 27, 2024 by Jtgavigan

[Jtgavigan]

13 minutes ago, doctor220 said:

I guess I don’t really see the problem here. It seems like the data is consistently telling you that your approach play is the main area you lose strokes to scratch players. Do you think that is true or not? If not then where do the shots come from 

Historically,  not keeping the ball in play off the tee has been the big one. Putting can be inconsistent (speed control) and I can be an atrocious chipper at times. 

 

Approach play is somewhat meh, but not as big of a priority. 

 

I had more of an issue with certain rounds, like yesterday,  telling me I was losing or gaining too much.

 

Overall, when I average it out, the numbers seem reasonable I guess. I posted above as to what some of the issues are.

 

I also have been following more of the DECADE approach with my on course strategy. I fire at fewer pins and leave myself easier up and downs these days. I bet my approach SG would have been better a few years ago, yet I often short sided myself and left very difficult up and down opportunities. 

Edited December 27, 2024 by Jtgavigan

[Jtgavigan]

This discussion has been very helpful,  and I appreciate all who has and will post. 

 

Maybe I am just a looney tune, maybe I just have more to learn, but all has been helpful.

[Pnwpingi210]

2 minutes ago, Jtgavigan said:

This discussion has been very helpful,  and I appreciate all who has and will post. 

 

Maybe I am just a looney tune, maybe I just have more to learn, but all has been helpful.

It’s a good discussion and I think as someone that has an engineering background and some quantitative data and process analysis background it’s understandable why you expectations are very high.

 

I think the simple answer is while the arrcos data set is large, it’s really just in its infancy of being collected and algorithm refined.  It’s not mature enough yet to be able to do the batch analysis with the specificity you’re expecting.  I think it will get there as more resources are invested.

[RayPlan]

28 minutes ago, Jtgavigan said:

That is exactly my point. 

 

I get that it is trying to quantify EVERYTHING into some algorithm,  and it is just impossible. 

 

I might go out and shoot 75 on a tough track in cool and windy conditions and it will tell me how I lost strokes to a scratch player.

 

But you did lose strokes, based on the information it had. You may have had an understandable reason for that (weather conditions), so that's where you can make your own subjective adjustment. No system is going to be perfect.

[iacas]

9 minutes ago, Jtgavigan said:

However, I don't think it is out of bounds to call out that single round stats should be better or they can be almost ignored.

 

It's not "out of bounds" but it's not realistic.

 

Let's pretend my overall SG:P relative to a scratch golfer is +2.5. It's not unusual for one round to be +6 and the next round, a day later, in the same conditions, to be -2. I think when Patrick Cantlay set the record for SG:P in a PGA Tour event… he had a day when he was negative (he lost strokes)!

 

Now, putting fluctuates more than the other SG categories, but still… conditions fluctuate. Your swing fluctuates. Your stats fluctuate.

 

And… as noted, though it comes out "in the wash," if you hit a tee shot into the "rough" but it's really a horrible lie, you could lose a shot or more on that one approach shot, when if the ball was a foot to the left (but still in the rough), it could be fine.

 

SG in a single round is a small sample size.

 

So, while it may not be "out of bounds," it isn't the best application of SG.

 

9 minutes ago, Jtgavigan said:

However, the data swings wildly between batches because of crappy instrumentation, or fluctuations in their process.

 

Your golf game is not automated machinery under controlled conditions.

 

9 minutes ago, Jtgavigan said:

I consider rounds like my customers' batches.

 

You're not Moe Norman.

 

9 minutes ago, Jtgavigan said:

But as I see it today, that isn't possible unless a bunch of scratch and low handicap players go play the same course I am playing on the same day and we all measure everything. Then, the metrics will make sense.

 

That'd help but it'd still be a small sample size when you're regarding YOUR stats for the day. Maybe you hit when the wind was more blustery than others. Or variance gave you better or worse lies than others. Or you play a higher spin ball, that held the firmer greens a bit better. Etc.

 

9 minutes ago, Jtgavigan said:

If I consistently play course with hard and fast greens and difficult rough and short side myself a lot, it might tell me I have a terrible short game, when in reality, it is my approach play leaving me in a pickle all the time. My proximity may be fine, but the reality is i am leaving myself no chance at par a lot.

 

SG isn't "everything." It doesn't mean you get to turn off your brain and stop thinking. That also doesn't mean it's worthless.

[Jtgavigan]

Fair enough @iacas, all good points. Wanted to make a wise crack about the Moe Norman comment, but everything I thought of was too easy to blow me up with. Lol. 

[betarhoalphadelta]

1 hour ago, Jtgavigan said:

As several have stated, and I agree to an extent, that 5 to 10 round averages are more telling.

 

However, I don't think it is out of bounds to call out that single round stats should be better or they can be almost ignored. 

 

Let's say I have a customer who makes some product via a batch process. If I average out 10 batches, the data looks fine, like they are producing good product. However, the data swings wildly between batches because of crappy instrumentation, or fluctuations in their process. I need to help them pinpoint the anomalous batches and understand why that is happening. I have to have a good way to do that on a batch by batch basis and help them either fix their instrumentation or their process. They could send out a lot of bad product that were made in the bad batches, but if I tell them "it will all work out in the wash, your 10 batch averages all look fine," I have misguided them.

 

I consider rounds like my customers' batches. I should be able to have accurate data on a round by round basis.

 

 

As someone who knows what it takes to get through an electrical engineering curriculum, I'm only going to say this once...

 

You're far too smart to say something this daft 😉 

 

What you're missing, with your final statement, is that you DO have accurate data on a round by round basis. But as you know from your day job, data is only one piece of the puzzle. 

 

I work in data storage. My employer exists to enable our customers to store large amounts of data. Everyone thinks their data is valuable, or they wouldn't store it. But I've got a saying... Data is worthless. Information is valuable. 

 

The difference is that you have to understand the data to turn it into information. 

 

You have accurate data on a round by round basis with Arccos. But that data on a round by round basis doesn't give you actionable information. And information is what you're looking for. You need a bigger sample size of data to get enough information to make decisions. 

 

Per your customers, if every manufacturing batch varies wildly, looking at the data from one specific bad batch probably tells you nothing useful. You have to step outside the data and look at things from a wider perspective. The same thing is true here. You could go out and have a terrible approach day because you have a critical swing flaw requiring timing everything and you just couldn't time it up that day, or because you didn't sleep well the night before and aren't at your best, or because you're mentally distracted thinking about a customer problem, or (as iacas says) because you just got some bad luck with some terrible lies that day. One day tells you absolutely nothing useful. 

 

Arccos just gives you the data. You have to have the right approach (pun intended) to analyze it to make it useful information. 

[Jtgavigan]

25 minutes ago, betarhoalphadelta said:

 

As someone who knows what it takes to get through an electrical engineering curriculum, I'm only going to say this once...

 

You're far too smart to say something this daft 😉 

 

What you're missing, with your final statement, is that you DO have accurate data on a round by round basis. But as you know from your day job, data is only one piece of the puzzle. 

 

I work in data storage. My employer exists to enable our customers to store large amounts of data. Everyone thinks their data is valuable, or they wouldn't store it. But I've got a saying... Data is worthless. Information is valuable. 

 

The difference is that you have to understand the data to turn it into information. 

 

You have accurate data on a round by round basis with Arccos. But that data on a round by round basis doesn't give you actionable information. And information is what you're looking for. You need a bigger sample size of data to get enough information to make decisions. 

 

Per your customers, if every manufacturing batch varies wildly, looking at the data from one specific bad batch probably tells you nothing useful. You have to step outside the data and look at things from a wider perspective. The same thing is true here. You could go out and have a terrible approach day because you have a critical swing flaw requiring timing everything and you just couldn't time it up that day, or because you didn't sleep well the night before and aren't at your best, or because you're mentally distracted thinking about a customer problem, or (as iacas says) because you just got some bad luck with some terrible lies that day. One day tells you absolutely nothing useful. 

 

Arccos just gives you the data. You have to have the right approach (pun intended) to analyze it to make it useful information. 

I think that is my struggle - you nailed it. I am looking for valuable information,  and as I see it, with too high of expectations. 

 

We are taking a bunch, like millions, of bowls of various fruit and trying to compare them. We will always be limited in algorithmically comparing a game that has such wide variances in players, course, and conditions.

 

Also, per this discussion with @iacas, which also proves what I think I am trying to say and why my frustration is there:

 

Me:

 

  1 hour ago, Jtgavigan said:

But as I see it today, that isn't possible unless a bunch of scratch and low handicap players go play the same course I am playing on the same day and we all measure everything. Then, the metrics will make sense.

 

 

Iacas:

That'd help but it'd still be a small sample size when you're regarding YOUR stats for the day. Maybe you hit when the wind was more blustery than others. Or variance gave you better or worse lies than others. Or you play a higher spin ball, that held the firmer greens a bit better. Etc

 

That is what I am trying to say, it is difficult to categorize things in such a way that an algorithm like this can be super objective. It is just impossible as there are too many variables. As mentioned, shots a few feet away from each other can have drastically different difficulty levels.

 

Is SG useless or worthless? No. As valuable as I personally would like it to be? Also no. But that is on me and not the algorithm itself. 

 

My thing about yesterday (again, I know, you can't look at one round, I get that now) and why I thought the stats were misleading is that I hit the ball in places where I KNEW par was my worst score. I had two easy chips and 4 putts from the fringe, so the algorithm loved my short game and didnt like my approach game. I felt that my approach game was decent and my short game was pretty much as I would expect. Neither one spectacularly good or bad. 

 

My "chips" were easy up and downs because I picked a good target and missed just a little more than I really wanted to, but in general, I hit it very much where I intended all day. I played more club to short pins, less club to back pins, and always favored the safer side to miss the green and hit close to where I wanted.  It didn't reward me much on approach because I didn't hit it close or on the green in some cases. So be it. 

 

I have had so many days in the past where I pick a crappy target and try to stuff it on to a pin that my skill level doesn't warrant attacking. I leave myself a difficult short sided chip and try to get too cute with the pitch and leave it short again or blade it long and take double or worse. This particular algorithm has no way to really understand that maybe a 15 yard short sided chip over a bunker is worse than a 25 yard easy chip and run because it is trying to normalize everything.

 

I am just not sure that golf CAN be normalized. There are just too many variables and too much variance in courses, conditions, and golfers. I get why people are trying to. Makes a lot of sense actually.

 

However, I do get that this is the best we have and is better than nothing. It is trying to take the subjective and make it as objective as possible. I get that. Many other sports have done the same and old myths/thoughts have been busted. Likely some of my own thoughts about my own game will be busted through this one. 

 

It is a tool I will keep using for now, but this discussion has helped me look at it better and what it's place is. It should help me see patterns I may have missed just tracking fairways hit, greens in regulation, ups and downs, etc.

[Pnwpingi210]

 

13 minutes ago, Jtgavigan said:

I think that is my struggle - you nailed it. I am looking for valuable information,  and as I see it, with too high of expectations. 

 

We are taking a bunch, like millions, of bowls of various fruit and trying to compare them. We will always be limited in algorithmically comparing a game that has such wide variances in players, course, and conditions.

 

Also, per this discussion with @iacas, which also proves what I think I am trying to say and why my frustration is there:

 

Me:

 

  1 hour ago, Jtgavigan said:

But as I see it today, that isn't possible unless a bunch of scratch and low handicap players go play the same course I am playing on the same day and we all measure everything. Then, the metrics will make sense.

 

 

Iacas:

That'd help but it'd still be a small sample size when you're regarding YOUR stats for the day. Maybe you hit when the wind was more blustery than others. Or variance gave you better or worse lies than others. Or you play a higher spin ball, that held the firmer greens a bit better. Etc

 

That is what I am trying to say, it is difficult to categorize things in such a way that an algorithm like this can be super objective. It is just impossible as there are too many variables. As mentioned, shots a few feet away from each other can have drastically different difficulty levels.

 

Is SG useless or worthless? No. As valuable as I personally would like it to be? Also no. But that is on me and not the algorithm itself. 

 

My thing about yesterday (again, I know, you can't look at one round, I get that now) and why I thought the stats were misleading is that I hit the ball in places where I KNEW par was my worst score. I had two easy chips and 4 putts from the fringe, so the algorithm loved my short game and didnt like my approach game. I felt that my approach game was decent and my short game was pretty much as I would expect. Neither one spectacularly good or bad. 

 

My "chips" were easy up and downs because I picked a good target and missed just a little more than I really wanted to, but in general, I hit it very much where I intended all day. I played more club to short pins, less club to back pins, and always favored the safer side to miss the green and hit close to where I wanted.  It didn't reward me much on approach because I didn't hit it close or on the green in some cases. So be it. 

 

I have had so many days in the past where I pick a crappy target and try to stuff it on to a pin that my skill level doesn't warrant attacking. I leave myself a difficult short sided chip and try to get too cute with the pitch and leave it short again or blade it long and take double or worse. This particular algorithm has no way to really understand that maybe a 15 yard short sided chip over a bunker is worse than a 25 yard easy chip and run because it is trying to normalize everything.

 

I am just not sure that golf CAN be normalized. There are just too many variables and too much variance in courses, conditions, and golfers. I get why people are trying to. Makes a lot of sense actually.

 

However, I do get that this is the best we have and is better than nothing. It is trying to take the subjective and make it as objective as possible. I get that. Many other sports have done the same and old myths/thoughts have been busted. Likely some of my own thoughts about my own game will be busted through this one. 

 

It is a tool I will keep using for now, but this discussion has helped me look at it better and what it's place is. It should help me see patterns I may have missed just tracking fairways hit, greens in regulation, ups and downs, etc.

Good post.

 

in summary don't lose the forest for the trees. 🙂

[Jtgavigan]

20 minutes ago, Pnwpingi210 said:

 

Good post.

 

in summary don't lose the forest for the trees. 🙂

Thanks. Have definitely learned some things and also have a better idea of why I have some doubts.

 

Hope others have learned some things.