Introducing Simple Shot Quality Model (SSQM) v1.0

Sravan February 17, 2024 [NBA] #Shooting #Shot Quality #SSQM

My first attempt at creating a shot quality model using publicly available NBA data

Introduction

While I was digging into adjusted net ratings (you can find my article here), I found a version of luck adjusted team ratings by Krishna Narsu using qSQ (Quantified Shot Quality) and qSI(Quantified Shooter Impact). I have been interested in shot quality ever since. The disadvantage of these metrics is that they are from Second Spectrum and not public. So wanted to see if I can develop a model on my own using publicly available shooting data.

Methodology

Getting Started

Before starting from scratch I look at existing work from other people. Ryan Davis and Andrew Patton have good tutorials for NBA Analytics with working code and those are always good place to start when working on a new analytics project. I found Andrew Patton's Shot Quality Model Tutorial interesting. Andrew shows the approach of how to model shot quality and the data processing framework necessary for creating a shot quality metric. After trying my own variations of Andrew's demo shot quality models, I wanted to try something different while utilizing some of the concepts he introduced in the tutorial.

Approach: Using League Average Shooting

I have been making NBA Hex Plots for more than 2 years now.

The different colors in hex plots shows how efficient a player is wrt league average. I accomplish it by binning the shots into different zones and calculating the FG% of all shots in the zone attempted by the player. I then compare it to the league average FG% of all shots in the zone attempted by all players in that season. Here are the shot zones and the league averages shown in Steph Curry's shot-chart for the season:

My idea is that the approach I adopted for the hex plots is a good framework for a Simple Shot Quality Model and thus name SSQM (sounds like a good name for a model which is very simple/basic).

Developing the Shot Quality Model Metrics

I will be using the league average FG% as expected FG% in my model and will refer it as xFG from now now.
So the expected points scored are:

$xPTS=\begin{cases} 2\times FGA \times xFG &, FGA \text{ is 2 pointer} \\ 3\times FGA \times xFG &, FGA \text{ is 3 pointer} \\ \end{cases}$

Effective field goal percentage (eFG) is calculated as:

$$ eFG = \frac{PTS}{2\times FGA} $$

Similarly expected eFG (xeFG) is calculated by replacing PTS with xPTS.

$$ xeFG = \frac{xPTS}{2\times FGA} $$

Shot Making is defined as:

$$ \text{Shot Making} = \frac{PTS - xPTS}{FGA} $$

which is nothing but points per point (PPS) above league average.

Points Added is defined as:

$$ \text{Points Added} = PTS - xPTS $$

Another way of calculating Points Added:

$$ \text{Points Added} = \text{Shot Making}\times FGA $$

Results

Now that I have a Shot Quality Model based on shot location (grouped in zones), I ran it for the 2023-24 NBA season to generate the shot quality metrics for all players. An issue here is that there are players who have not attempted many shots but made most of them. We need to filter those players out, which can be accomplished by removing players who have scored less than 100 points.

Best Shot Makers

Here are the best shot makers in the league i.e. the players having the highest Shot Making values:

	Player	Team	FGA	FGM	eFG	xeFG	PTS	xPTS	Shot Making	Points Added
1	Jalen Smith	Indiana Pacers	253	160	0.719	0.577	364	291.9	0.28	70.84
2	Amir Coffey	LA Clippers	162	91	0.673	0.55	218	178.2	0.25	40.5
3	Dereck Lively II	Dallas Mavericks	208	153	0.736	0.619	306	257.6	0.23	47.84
4	Nicolas Batum	Philadelphia 76ers	135	70	0.681	0.577	184	155.7	0.21	28.35
5	Garrison Mathews	Atlanta Hawks	99	45	0.657	0.554	130	109.7	0.21	20.79
6	Kevin Durant	Phoenix Suns	912	491	0.598	0.497	1090	907.2	0.2	182.4
7	Matt Ryan	New Orleans Pelicans	85	38	0.635	0.536	108	91.1	0.2	17
8	Jakob Poeltl	Toronto Raptors	320	216	0.675	0.58	432	371.4	0.19	60.8
9	Aaron Wiggins	Oklahoma City Thunder	211	123	0.68	0.587	287	247.6	0.19	40.09
10	Grayson Allen	Phoenix Suns	414	211	0.658	0.566	545	468.9	0.18	74.52

The players here include centers like Jalen Smith (who also shoots 3s), Lively and Poeltl and three point shooters like Grayson Allen, Batum. The only super star in here is Kevin Durant, who is have a ridiculous shooting season.

Worst Shot Makers

Here are the worst shot makers in the league i.e. the players having the lowest Shot Making values:

	Player	Team	FGA	FGM	eFG	xeFG	PTS	xPTS	Shot Making	Points Added
1	JT Thor	Charlotte Hornets	139	51	0.435	0.573	121	159.3	-0.28	-38.92
2	Xavier Tillman	Boston Celtics	213	87	0.437	0.567	186	241.7	-0.26	-55.38
3	Isaiah Livers	Washington Wizards	116	40	0.44	0.557	102	129.1	-0.23	-26.68
4	Scoot Henderson	Portland Trail Blazers	506	190	0.427	0.544	432	550.2	-0.23	-116.38
5	Josh Okogie	Phoenix Suns	183	76	0.481	0.591	176	216.1	-0.22	-40.26
6	Kris Murray	Portland Trail Blazers	118	48	0.487	0.591	115	139.5	-0.21	-24.78
7	David Roddy	Phoenix Suns	389	157	0.465	0.568	362	441.9	-0.21	-81.69
8	Mike Muscala	Detroit Pistons	134	48	0.459	0.552	123	148	-0.19	-25.46
9	Cam Reddish	Los Angeles Lakers	205	83	0.483	0.577	198	236.7	-0.19	-38.95
10	Chris Duarte	Sacramento Kings	129	47	0.453	0.551	117	142.1	-0.19	-24.51

This list is made mostly of young players who are trying to find their place in the league including rookies like Scoot Henderson. It is very hard to adjust to the NBA and it shows up in their shooting numbers. The other players here are Mike Muscala, who are aging centers who are older than 30 years. Loss of athleticism has an adverse impact on shot making.

Most Points Added: Best Volume Shot Makers

Just looking at Shot Making doesn't tell the whole story of best shot makers in the league. The Shot Making value is actualized only if the players are taking lots of the shots. So the best shot makers in the league are who make lots of shots at high volume. To differentiate these players (who are all stars) from best players in Shot Making metric, I call them Volume Shot Makers. These players are players who have the highest Points Added Metric:

	Player	Team	FGA	FGM	eFG	xeFG	PTS	xPTS	Shot Making	Points Added
1	Kevin Durant	Phoenix Suns	912	491	0.598	0.497	1090	907.2	0.2	182.4
2	Nikola Jokic	Denver Nuggets	937	541	0.607	0.529	1138	991.7	0.16	149.92
3	Kawhi Leonard	LA Clippers	820	432	0.595	0.516	976	846.1	0.16	131.2
4	Stephen Curry	Golden State Warriors	989	457	0.589	0.524	1166	1036.8	0.13	128.57
5	Luka Doncic	Dallas Mavericks	1102	542	0.574	0.518	1265	1141.8	0.11	121.22
6	Shai Gilgeous-Alexander	Oklahoma City Thunder	1063	580	0.576	0.527	1225	1119.4	0.1	106.3
7	Devin Booker	Phoenix Suns	864	432	0.557	0.5	963	864.7	0.11	95.04
8	Giannis Antetokounmpo	Milwaukee Bucks	1017	626	0.627	0.584	1276	1186.9	0.09	91.53
9	Jalen Williams	Oklahoma City Thunder	662	360	0.598	0.535	792	708	0.13	86.06
10	Domantas Sabonis	Sacramento Kings	706	439	0.64	0.581	904	821	0.12	84.72

This list mostly consists of super stars and all-stars except for Jalen Williams who is having a hyper efficient season fo far. The list being comprised of stars is expected since taking and making tough shots is what makes them stars.

Most Points Removed: Worst Volume Shot Makers

Now let's look at the worst volume shot makers, these are players with the lowest Points Added metric:

	Player	Team	FGA	FGM	eFG	xeFG	PTS	xPTS	Shot Making	Points Added
1	Scoot Henderson	Portland Trail Blazers	506	190	0.427	0.544	432	550.2	-0.23	-116.38
2	Jalen Green	Houston Rockets	815	335	0.477	0.545	778	889.1	-0.14	-114.1
3	Jordan Poole	Washington Wizards	730	292	0.469	0.533	685	778.8	-0.13	-94.9
4	Russell Westbrook	LA Clippers	523	242	0.497	0.575	520	601.6	-0.16	-83.68
5	Saddiq Bey	Atlanta Hawks	585	247	0.504	0.576	590	674.1	-0.14	-81.9
6	David Roddy	Phoenix Suns	389	157	0.465	0.568	362	441.9	-0.21	-81.69
7	Jeremy Sochan	San Antonio Spurs	557	246	0.488	0.557	544	620.8	-0.14	-77.98
8	Keldon Johnson	San Antonio Spurs	668	299	0.513	0.563	685	752	-0.1	-66.8
9	Josh Giddey	Oklahoma City Thunder	550	246	0.494	0.554	543	609.9	-0.12	-66
10	Damian Lillard	Milwaukee Bucks	887	375	0.506	0.541	898	959.1	-0.07	-62.09

This list consists of young players and players on bad teams like Jordan Poole and Keldon Johnson. An interesting observation is that guards comprise of most of this list. I was expecting to find Russell Westbrook in here but was shocked to find Damian Lillard. Dame is having a horrible season so far and Bucks have no shot to compete for a NBA championship unless Dame shoots better. Both Russ and Dame are in a long line of aging superstar guards who drop off in mid 30s. While for Russ it has been a gradual process, for Dame it has been a sudden drop off this year.

Conclusions

I've developed a simple shot quality model SSQM v1.0 based on publicly available shooting data. The model is based purely on shot location (shot zones to be more specific). Then we can sort the players by the shot quality metrics like Shot Making and Points Added to find best and worst shot makers and volume shot makers.

While working on this model, it came to my mind that only shot location is not a holistic way of looking at shot quality and there should be other factors like shot type, defender distance, touch time, dribble etc. which need to be incorporated in the model. I'm working on SSQM v2.0 which does exactly that and I will release a follow up to this blog describing that model.

Thank you for reading, and any feedback is appreciated. You can reach me on Twitter at @SravanNBA.

Resources

Andrew Patton's Shot Quality Model Tutorial