PDA

View Full Version : Understanding USG% (Because it's obvious many of you don't)



scott
07-06-2024, 12:34 PM
I wanted to make this post because I see a lot of misusage (heh) of the USG% stat in rhetorical arguments. A fine example:


Yeah that player's scoring average went up, but it's only because their USG% went up. Once their USG% goes down, so will their average. We shouldn't want that player.

This statement sounds like someone who has some deep understanding of advanced stats and has uncovered some unique and meaningful insight into a player's game, when in reality they're just telling you what would be an obvious relationship between a basic counting stat (points per game) and an advanced stat (USG%).

What does USG% tell us? Basketball-reference.com and NBA.com use slightly different formulas to calculate USG% (which you can find here (https://www.basketball-reference.com/about/glossary.html#:~:text=Usg%25%20%2D%20Usage%20Perce ntage%20(available,he%20was%20on%20the%20floor.) and here (https://www.nba.com/stats/help/glossary#usgpct)), which spit out slightly different results, but they are both telling you the same thing: the percentage of a team's plays that end with an individual player while he is on the court. A play can only end in a finite number of ways: a made shot, a missed shot, or a turnover. That's it. There is no other way for a play to end except for the game clock expiring, which counts as nothing. There is no other possible way for a play to end.

USG% does not measure: how ball dominant a player is, or how much they need the ball in their hands, or how much of a possession a player takes up, or whether they set up their teammates, etc. It only tells us the percentage of a team's plays that end with an individual player while he is on the court.

A very exaggerated example of how USG% can be deceiving:

Imagine a game of 2-on-2 with a 24 second shot clock and 100 plays per team and every shot is worth 2pts.

Player A and Player B are teammates. Player A brings the ball down the court and on EVERY possession, dribbles the ball around the court for 23 seconds before passing it to Player B, who is forced to throw up a quick shot or get caught with the ball in his hands as the shot clock expires (a turnover). Let's say that 50% of the time he gets the shot off, and 20% of the time he makes it. The other 50% of the time he commits a turnover. Let's also assume every missed shot ends in a defensive rebound.

In this scenario, Player A will have a USG% of 0, and Player B will have a USG% of 100. Player A will average 10 assists/gm, Player B will average 20 ppg on 20% shooting and 50 TOV/gm.

This is a purposely exaggerated example to show you that USG% isn't really all that insightful at all. You might look at the stats and conclude that Player B is a ball dominant, inefficient chucker who is sloppy with the ball, when that's not really the case at all.

I'll come back to this example a little later.

Because scoring is the result of shooting the basketball, there is an extremely strong correlation between USG% and scoring average. So strong, in fact, that among the Top 40 scorers in the NBA last year, 36 of them were in the Top 40 in USG%.

Here is a scatter plot of the PPG average of the Top 40 scorers last year against their USG%.

https://i.ibb.co/cNNVLjp/USG-v-PPG.png

You'll notice a pretty clear correlation here, which makes sense. To score, you must shoot, and by shooting more, your USG% increases. Pretty straightforward, right?

There is a wrinkle, of course - and that is in a player's TOV rate and their shooting efficiency.

If we go back to our exaggerated example, if Player B increases his shooting % from 20% to 30% (remember, these are all tough shots he's taking, thanks to his teammate), then his scoring average will increase from 20ppg to 30ppg while his USG% will remain unchanged. Similarly, if he maintains his shooting % of 20%, but he reduces his turnovers from 50/gm to 40/gm, and those 10 possessions generate 10 extra shots, his scoring average will increase from 20ppg to 24ppg.

So, whenever we see a player's PPG increase, then there can only be a handful of explanations why: his USG% went up, his shooting efficiency went up, or his turnovers were replaced with more shots (if his turnovers are reduced but they are not replaced by shots, then his USG% will just go down without a corresponding increase in PPG).

So, whenever you see someone say this:


HiS sCoRiNg WeNt Up OnLy BeCaUsE hIs UsG% WeNt Up

You can respond with "yeah, no shit."

The idea that the Spurs will add a #2 scorer whom they'll presumably pay the max at some point, and then turn around and reduce their usage considerably is a weird concept. Why would you acquire that player just to do that? Likelywise, if that player is more efficient at scoring, why would you want to reduce their usage for someone who is less efficient? The only explanation would be that you expect that player's efficiency to increase dramatically. This would be the situation with Wemby. He's actually one of the less efficient high usage scorers, ranking 39th in scoring average but #8 in USG%. But he was a rookie, and we expect that to continue to improve dramatically as his shooting improves and (maybe more importantly) his turnovers are reduced.

Coincidently, among the Top 40 scoring average leaders last year, two of the two most efficient were Demar Derozan, who ranked #24 in scoring but #44 in USG%, and Lauri Markkanen, who ranked #26 in scoring but #51 in USG%. In the case of Lauri, that's because he's a highly efficient scorer (indeed, he's a 50/40/90 guy who doesn't commit a lot of turnovers). So it's odd to hear people say they'd want him to be our number 3 option, or his usage would get significantly reduced if we signed him... why would you do that? Why would you prefer less efficient Devin Vassell shots or Keldon Johnson shots to highly efficient Lauri Markkanen shots?

Yes, this has been a secret Lauri post. But hopefully you learned something in the meantime, and folks can cut back on their utterly braindead misusage (heh) of USG%.

Mr. Body
07-06-2024, 12:39 PM
Yeah, when Markkanen had to share the ball he was a 14ppg 5 rpg player. When he got many more shots on a bad team he turned into a 23ppg.

That's not real. That's called inflation.

When he plays for a good team, he will be back to a 14ppg player.

He's a pretty good player. Is he worth a max? Fuck no.

I'm glad you're pointing this out.

scott
07-06-2024, 12:40 PM
I knew I could count on Mr. Body doubling down on stupid.

scott
07-06-2024, 12:42 PM
If we sign Shai and greatly reduce his usage, he'll also see a significant reduction in his scoring average.

Shai isn't worth the max.

Go KiLl YoUrSeLf

Dejounte
07-06-2024, 12:44 PM
I knew I could count on Mr. Body doubling down on stupid.

One of the other terrible takes he keeps spouting is that it’s not worth trading for players that other teams want to trade away. I guess the Spurs should never try to make trades then. You can’t fix stupid.

scott
07-06-2024, 12:49 PM
One of the other terrible takes he keeps spouting is that it’s not worth trading for players that other teams want to trade away. I guess the Spurs should never try to make trades then. You can’t fix stupid.

Spurs FO: Hi, Mr. Opposing GM, I am interested in [PLAYER], would you like to trade with us?

Opposing GM: Hi, Mr. Spurs FO, I would be willing to trade [PLAYER] for the right offer, what do you suggest?

Spurs FO: NEVERMIND, HE SUCKS.

SpursBills
07-06-2024, 12:49 PM
I think for many it's very easy to confuse "usage" with "time of possession", and while they are sometimes correlated, they definitely do not measure the same thing. The discrepancy is most evident in high efficiency play finishers both perimeter (eg Klay, Lauri Bird) and paint based (eg Amare Stoudemire) who may not initiate or connect a ton but are very efficient at taking the ball from someone else and getting it into the basket.

R. DeMurre
07-06-2024, 12:51 PM
If anything, the discussion should be that Markkanen went from a situation where he wasn't even a top 3 option on offense (In Cleveland, where he was #6 on the team in FGA per game) to a situation where he was the #1 option and therefore much more the focus of opposing team defenses... If his improvement wasn't legit, his efficiency would've gone down, not up, and then the argument that his increased scoring was only due to getting more shots might make sense. But with more pressure and more attention, his efficiency only improved.

LeBowen
07-06-2024, 12:52 PM
I knew I could count on Mr. Body doubling down on stupid.

This is the first time ever I've seen someone criticize a player for raising his ppg by 10 while keeping the same efficency.
He just has zero self awareness.

R. DeMurre
07-06-2024, 12:56 PM
This is the first time ever I've seen someone criticize a player for raising his ppg by 10 while keeping the same efficency.
He just has zero self awareness.


And he didn't keep the same efficiency, he increased it. In Cleveland his TS% was .582, and in his two years in Utah it was .640 and .631, the two best of his career.

scott
07-06-2024, 01:02 PM
And he didn't keep the same efficiency, he increased it. In Cleveland his TS% was .582, and in his two years in Utah it was .640 and .631.

Sounds like exactly the kind of player who usage should be immediately reduced on another team!

024
07-06-2024, 02:20 PM
I knew I could count on Mr. Body doubling down on stupid.
:lol great post. Would be nice to see more legitimate posts on analytics and advanced stats. Not as the sole points of discussion of course but as well-reasoned supporting facts.

But most people form a position after just "feeling" a certain way and pull in some lazy analysis that support their world view without bothering to break down the layers and examine if it really does. Then they get angry and double down on their argument whenever challenged with opposing facts instead of considering they should re-evaluate their position.

Tbf this lesson is needed everywhere in society.

Seventyniner
07-06-2024, 03:37 PM
This is the first time ever I've seen someone criticize a player for raising his ppg by 10 while keeping the same efficency.
He just has zero self awareness.

Just keeping efficiency the same while increasing scoring average is really hard. Most players who see an uptick in usage are less efficient iirc.

scott
07-07-2024, 07:38 PM
I thought a fun side metric to run would be PPG/USG%, or, in other words, how many points does a player get for each unit of usage.

What I found was that, of the Top 40 Scorers in the league, here were the Top-5 most efficient (at turning usage into points)

1. Luka
2. KD
3. Demar
4. Lauri
5. AD

The bottom 5:

36. JJJ
37. Cam Thomas
38. Cade
39. LaMelo
40. Wemby

I suspect we'll see Wemby take a huge jump in efficiency this year as he matures and as we get improved play around him

CGD
07-07-2024, 07:43 PM
Sweet lord, another backdoor Markenen thread?? How many of these do we really need guys.

MultiTroll
07-07-2024, 08:32 PM
40. Wemby

I suspect we'll see Wemby take a huge jump in efficiency this year as he matures and as we get improved play around him
Absolutely. Wemby was given the ball in such shitty position by the Freeze Out Krew and also horked a lot of shots early in the shot clock early in the season. Knowing it was the only way he was going to get a shot.

Crusty Clifford Paul will be an improvement. For that matter out grandmothers would be an improvement at pg vs last years Popped offensive offensive strategy.
So CPO has an extremely low bar to clear.

scott
07-07-2024, 09:32 PM
Absolutely. Wemby was given the ball in such shitty position by the Freeze Out Krew and also horked a lot of shots early in the shot clock early in the season. Knowing it was the only way he was going to get a shot.

Crusty Clifford Paul will be an improvement. For that matter out grandmothers would be an improvement at pg vs last years Popped offensive offensive strategy.
So CPO has an extremely low bar to clear.

I'm going to keep track of this PPG/USG% stat throughout the year, I think it's a pretty nifty one. The trouble is that I don't have access to any kind of database where I can easily pull PPG and USG from the same table - I had to do some manual coding for it this time. Only took about 10 minutes, but it still not just something I can easily look up.

For those who want to follow along, Wemby's PPG/USG% was 0.665, Luka's was 0.942. The simple average amongst the Top40 was 0.841 with a median of 0.848.

Here's what it looks like for the top 40 if you lay them all out from best (Luka) to worst (Wemby)

https://i.ibb.co/F7gQbJq/PPG-per-Usage.png

And here is the data table for anyone interested:




Player
PPG
Scoring Rank
USG%
USG% Rank
PPG/USG
PPG/USG Rank


Joel Embiid
34.7
1
39.6
1
0.876
16


Luka Doncic
33.9
2
36
2
0.942
1


Giannis Antetokounmpo
30.4
3
33
4
0.921
7


Shai Gilgeous-Alexander
30.1
4
32.8
5
0.918
8


Jalen Brunson
28.7
5
32.5
6
0.883
14


Devin Booker
27.1
6
29.8
19
0.909
10


Kevin Durant
27.1
6
29
25
0.934
2


Jayson Tatum
26.9
8
30.2
17
0.891
13


Donovan Mitchell
26.6
9
31.4
9
0.847
21


De'Aaron Fox
26.6
9
31
11
0.858
18


Stephen Curry
26.4
11
31.3
10
0.843
23


Nikola Jokic
26.4
11
29.3
23
0.901
11


Anthony Edwards
25.9
13
32.3
7
0.802
29


Tyrese Maxey
25.9
13
28
30
0.925
6


Trae Young
25.7
15
30.5
13
0.843
24


LeBron James
25.7
15
29.2
24
0.880
15


Kyrie Irving
25.6
17
28.1
29
0.911
9


Ja Morant
25.1
18
30.4
15
0.826
26


Anthony Davis
24.7
19
26.7
34
0.925
5


Damian Lillard
24.3
20
28.4
28
0.856
19


Julius Randle
24
21
29.9
18
0.803
28


DeMar DeRozan
24
21
25.8
38
0.930
3


LaMelo Ball
23.9
23
34.3
3
0.697
39


Desmond Bane
23.7
24
28.5
27
0.832
25


Kawhi Leonard
23.7
24
26.5
37
0.894
12


Lauri Markkanen
23.2
26
25
40
0.928
4


Jaylen Brown
23
27
28.9
26
0.796
30


Zion Williamson
22.9
28
29.7
20
0.771
32


Cade Cunningham
22.7
29
30.8
12
0.737
38


Paolo Banchero
22.6
30
29.7
20
0.761
33


Anfernee Simons
22.6
30
28
30
0.807
27


Paul George
22.6
30
26.6
35
0.850
20


Cam Thomas
22.5
33
30.5
13
0.738
37


Jaren Jackson Jr.
22.5
33
30.4
15
0.740
36


Dejounte Murray
22.5
33
26.6
35
0.846
22


Kyle Kuzma
22.2
36
29.7
20
0.747
35


Karl-Anthony Towns
21.8
37
27.4
33
0.796
31


Pascal Siakam
21.7
38
25.2
39
0.861
17


Victor Wembanyama
21.4
39
32.2
8
0.665
40


Jamal Murray
21.2
40
27.9
32
0.760
34

scott
07-07-2024, 09:37 PM
:lol great post. Would be nice to see more legitimate posts on analytics and advanced stats. Not as the sole points of discussion of course but as well-reasoned supporting facts.

But most people form a position after just "feeling" a certain way and pull in some lazy analysis that support their world view without bothering to break down the layers and examine if it really does. Then they get angry and double down on their argument whenever challenged with opposing facts instead of considering they should re-evaluate their position.

Tbf this lesson is needed everywhere in society.

Been thinking about this a little bit and aside from my slipping in my hidden Lauri agenda, I think some educational posts like these on advanced metrics would be really informative and helpful for this board. I admit, there are a lot of metrics that I don't fully grasp and would love to understand better.

SpursBills (https://www.spurstalk.com/forums/member.php?u=18470) is one of the best posters around when it comes to advanced metrics, so maybe I can recruit him (and others) into this effort. Maybe timvp (https://www.spurstalk.com/forums/member.php?u=8) could pay us in Chuck E Cheese coupons to use these for his website if he wanted :lol

PennSpur
07-07-2024, 10:32 PM
You need to do pts/36 or something similar. Raw PPG will be biased toward players who play big minutes.

Edit: In response to your post comparing PPG/USG above.

scott
07-08-2024, 12:10 AM
You need to do pts/36 or something similar. Raw PPG will be biased toward players who play big minutes.

Edit: In response to your post comparing PPG/USG above.

Good call. I'll do that if I have time later this week. It's mostly the same guys in the top PPG scores, with a couple of exceptions, but the order is obviously shuffled around a bit. Victor, for example, shoots up from 39th in PPG to 15th in Pts/36.

scott
07-08-2024, 07:32 PM
I went ahead and calculated Pts/36 per USG% based on the suggestion above and was saw a little bit of movement. When I started this, I did not have this outcome in mind... but LOL, Lauri is b far the most efficient Top 40 scorer in the league, and the ONLY player in the Top 40 with a Pts/36 per USG% of greater than 1.

Top 5:

1. Lauri
2. Shai
3. Giannis
4. Kawhi
5. Kyrie

Bottom 5:

36. Randle
37. Wemby
38. Cade
39. Paolo
40. LaMelo

Full data table, ranked by Pts/36 per USG



Player
PPG
Scoring Rank
USG%
USG% Rank
PPG/USG
PPG/USG Rank
Pts/36
Pts/36 Rank
Pts/36 per USG
Pts/36 per USG Rank


Lauri Markkanen
23.2
26
25
40
0.928
4
25.3
21
1.012
1


Shai Gilgeous-Alexander
30.1
4
32.8
5
0.918
8
31.8
3
0.970
2


Giannis Antetokounmpo
30.4
3
33
4
0.921
7
31.2
4
0.945
3


Kawhi Leonard
23.7
24
26.5
37
0.894
12
24.9
25
0.940
4


Kyrie Irving
25.6
17
28.1
29
0.911
9
26.4
14
0.940
5


Joel Embiid
34.7
1
39.6
1
0.876
16
37.2
1
0.939
6


Anthony Davis
24.7
19
26.7
34
0.925
5
25
24
0.936
7


Nikola Jokic
26.4
11
29.3
23
0.901
11
27.4
7
0.935
8


Pascal Siakam
21.7
38
25.2
39
0.861
17
23.5
37
0.933
9


Stephen Curry
26.4
11
31.3
10
0.843
23
29.1
6
0.930
10


Devin Booker
27.1
6
29.8
19
0.909
10
27.1
8
0.909
11


Kevin Durant
27.1
6
29
25
0.934
2
26.2
15
0.903
12


Luka Doncic
33.9
2
36
2
0.942
1
32.5
2
0.903
13


Paul George
22.6
30
26.6
35
0.850
20
24
34
0.902
14


Jalen Brunson
28.7
5
32.5
6
0.883
14
29.2
5
0.898
15


LeBron James
25.7
15
29.2
24
0.880
15
26.2
15
0.897
16


Jayson Tatum
26.9
8
30.2
17
0.891
13
27
10
0.894
17


Tyrese Maxey
25.9
13
28
30
0.925
6
24.9
25
0.889
18


DeMar DeRozan
24
21
25.8
38
0.930
3
22.8
39
0.884
19


Zion Williamson
22.9
28
29.7
20
0.771
32
26.1
17
0.879
20


Karl-Anthony Towns
21.8
37
27.4
33
0.796
31
24
34
0.876
21


Desmond Bane
23.7
24
28.5
27
0.832
25
24.9
25
0.874
22


Damian Lillard
24.3
20
28.4
28
0.856
19
24.8
28
0.873
23


Jamal Murray
21.2
40
27.9
32
0.760
34
24.2
33
0.867
24


Donovan Mitchell
26.6
9
31.4
9
0.847
21
27.1
8
0.863
25


De'Aaron Fox
26.6
9
31
11
0.858
18
26.6
11
0.858
26


Jaylen Brown
23
27
28.9
26
0.796
30
24.7
29
0.855
27


Dejounte Murray
22.5
33
26.6
35
0.846
22
22.7
40
0.853
28


Anfernee Simons
22.6
30
28
30
0.807
27
23.6
36
0.843
29


Trae Young
25.7
15
30.5
13
0.843
24
25.7
19
0.843
30


Cam Thomas
22.5
33
30.5
13
0.738
37
25.7
19
0.843
30


Jaren Jackson Jr.
22.5
33
30.4
15
0.740
36
25.2
22
0.829
32


Ja Morant
25.1
18
30.4
15
0.826
26
25.1
23
0.826
33


Kyle Kuzma
22.2
36
29.7
20
0.747
35
24.5
30
0.825
34


Anthony Edwards
25.9
13
32.3
7
0.802
29
26.6
11
0.824
35


Julius Randle
24
21
29.9
18
0.803
28
24.3
32
0.813
36


Victor Wembanyama
21.4
39
32.2
8
0.665
40
26
18
0.807
37


Cade Cunningham
22.7
29
30.8
12
0.737
38
24.4
31
0.792
38


Paolo Banchero
22.6
30
29.7
20
0.761
33
23.2
38
0.781
39


LaMelo Ball
23.9
23
34.3
3
0.697
39
26.6
11
0.776
40



And here is a visual representation for which is good for seeing how Lauri and Shai really stand out from the pack.

https://i.ibb.co/svYBQtL/Pts-per-36-per-USG.png

spurraider21
07-08-2024, 07:39 PM
you son of a bitch

Arguendo
07-08-2024, 07:46 PM
TIL Assists, as step1 in a 2-step terminal play, are not included in usage. Preesh. But also FTs aren’t included either??

scott
07-08-2024, 07:52 PM
TIL Assists, as step1 in a 2-step terminal play, are not included in usage. Preesh. But also FTs aren’t included either??

FTs are included. FGA, FTA and TOV are the things that count towards usage. Basically, all the ways that a team's play can end.

ambchang
07-09-2024, 07:04 AM
Basketball reference:

Usg% - Usage Percentage (available since the 1977-78 season in the NBA); the formula is 100 * ((FGA + 0.44 * FTA + TOV) * (Tm MP / 5)) / (MP * (Tm FGA + 0.44 * Tm FTA + Tm TOV)). Usage percentage is an estimate of the percentage of team plays used by a player while he was on the floor.

MannyIsGod
07-09-2024, 09:53 AM
Great thread, Scott. Especially the efficiency portion. Obviously usage has to go up for scoring to increase but how efficient the new touches are in generating points is a really important metric to think about it.

scott
07-09-2024, 01:17 PM
Great thread, Scott. Especially the efficiency portion. Obviously usage has to go up for scoring to increase but how efficient the new touches are in generating points is a really important metric to think about it.

Manny, as one of our fellow math nerds - maybe you can take a stab at doing a write up on some advanced analytics. Not to toot my own horn, but I do think posts like these could help people understand advanced metrics a lot better - because people tend to throw them around incorrectly which leads to some bizarre basketball takes.