## I created a stat – VEDS (Volume, Efficiency, Defense, Spacing)

*This stat was created by Julien Rodger*

I made a statistic this summer called VEDS, short for Volume, Efficiency, Defense, Spacing. My philosophy the last few years is acquiring the most defenders who space the floor is a key to getting ahead in the NBA. At least until everyone catches up. The Warriors last year were the greatest spacing and defense masterpiece we’ve seen to date. But how are players like Draymond Green or DeMarre Carroll quantified compared to the ones who fill the statsheet?

My stat intends to include how to value players in these four categories. I have a formula for each of the four, then add them together. Here’s my explanation for each:

**Part 1: Volume (V)**

Volume scoring is the skill becoming less in vogue by the year. The old school mentality is high point per game players create their own shot to avoid stagnant defenses. That go-to scorers help you win in the 4th quarter.

The analytics movement has drawn the focus away from high volume scorers and towards the context of efficiency. However volume is far from meaningless. Many players in the NBA such as shooters or finishers at the rim rely on open shots to be most productive. Stars help draw double teams and shift the help defense to get their teammates open.

To use an example, the S in my stat for Spacing is designed to value long range, off the ball shooters who draw defenders away from the rim. When floating that spacing concept, one forum poster brought up the subject of Dwight Howard in Orlando. His argument was although Dwight plays exclusively near the rim and thus performing poorly in my “Spacing” section, Howard was critical to creating space for his teammates on the Orlando Magic. The defenders he soaked in off the pick and roll or in the post opened up 3 point shots for the litany of Magic shooters like Jameer Nelson, Rashard Lewis, J.J. Redick, Jason Richardson, etc. Yet Dwight is rated as a poor floor spacer. The Volume category however, is how this stat evaluates Dwight situations by giving him credit for the volume of possessions he uses. In a way “Volume” is as much a floor spacing statistic as “Spacing”. Volume scorers open space for teammates, just while playing on the ball, not off the ball.

How I calculate Volume is simple. First, I use this to evaluate possessions:

FGA + (FTA*.44) + TOV

Using the 0.44 coefficient for free throws instead of 0.5, is the common way to adjust for the effect of and-1s.

For example Russell Westbrook who led the league in Volume per game last year averaged 22 FGA a game, 9.8 FTA a game and 4.4 TOV a game, or roughly 30.7 possessions per game.

To calculate Volume, I divide the player’s possessions by 25, then multiply by 10. Westbrook’s 30.7 possessions, when divided by 25 and multiplied by 10 equals 12.3.

Why divide by 25? Because using exactly 25 possessions a game would give a player a rating of 10 in Volume. Last year only Westbrook, Kobe Bryant in limited games, James Harden, DeMarcus Cousins and Lebron James used more than 25 possessions a game, with Carmelo Anthony fractionally under. In all my categories I tried to make breaking 10 as the threshold of an “elite”, special performance in it. 25 possessions a game felt as good a number as any to represent a 10.

Here is the top 20 players in the league in Volume per game, rounded. Note that since possessions per game is included, having a higher minutes per game affects these rankings:

Russell Westbrook – 12.3

Kobe Bryant – 10.9

James Harden – 10.6

DeMarcus Cousins – 10.6

Lebron James – 10.3

Carmelo Anthony – 10.0

LaMarcus Aldridge – 9.6

Dwyane Wade – 9.4

Kevin Durant – 9.1

Blake Griffin – 8.9

Anthony Davis – 8.8

DeMar Derozan – 8.8

Stephen Curry – 8.7

Rudy Gay – 8.7

Damian Lillard – 8.6

Chris Bosh – 8.6

Derrick Rose – 8.5

Kyrie Irving – 8.5

Monta Ellis – 8.4

Michael Carter-Williams – 8.4

**Part 2: Efficiency (E)**

This is the most complicated of my 4 formulas. How I would describe it “How productively the player used his possessions” vs an average replacement. For example when evaluating James Harden’s impact on the game, his value is not only drawing attention to himself as a volume scorer, floor spacing or defense, but because he uses X amount of possessions at Y greater levels than the league average in efficiency, this alone has a significant mathematical impact on his team’s offense.

Once again my possession stat of FGA + 0.44*FTA + TOV plays an important role in this calculation. Another key is Dean Oliver’s individual ORTG from basketball-reference.com which goes beyond just capturing shooting efficiency like TS%, but also turnovers, assists and offensive rebounding.

Here is how I calculate it:

Step 1: I start by taking the player’s individual ORTG and subtract the league average ORTG for that season.

James Harden’s ORTG in 2014-2015 was 118, while league average ORTG was 105.6. The difference is +12.4.

Step 2: Next, I divide this number by 100 and add to 1 to get a number I can multiply with.

12.4 / 100, + 1 = 1.124.

Step 3: I multiply Step 2’s result with the player’s number of possessions, then subtract the number of possessions. This gives me the difference between how efficiently the player used his possessions compared to the league average.

James Harden uses roughly 26.6 possessions a game last year. 26.6 multipled by 1.124 is 29.9, or +3.3 compared to if 26.6 had been multiplied by 1. Effectively what I calculated or close enough to it, is that Harden created 3.3 more points with his 26.6 possessions than a league average player would.

Step 4: I multiply 3.3 * 3 to get a number more representive of a scale of 10.

James Harden’s +3.3 translates to a rating of 9.9 in Volume. When I first created this part of the stat I wanted to multiply by 2.7 which is an estimate of how many wins a point is worth in the NBA. Multiplying by 2.7 leads to one player in the 2014-2015 in Chris Paul crossing the threshold of 10 which I established in the “Volume” category as representing an elite number. Multiplying by 3 includes 3 others in Anthony Davis, Stephen Curry and Kevin Durant. I chose the latter to match it up with the “Volume” category’s scale slightly more.

Here is the top 20 players in the Efficiency per game:

Chris Paul – 11.2

Anthony Davis – 10.8

Stephen Curry – 10.7

Kevin Durant – 10.5

James Harden – 9.9

Jimmy Butler – 9.1

Enes Kanter (OKC) – 8.5

Brandan Wright (DAL) – 8

Kyrie Irving – 7.2

Tyson Chandler – 7.2

Deandre Jordan – 6.3

J.J. Redick – 5.3

George Hill – 5.3

Anthony Morrow – 5.1

Jonas Valanciunas – 5.1

Russell Westbrook – 5.0

Kyle Korver – 5.0

Lebron James – 5.0

Blake Griffin – 4.9

Isaiah Thomas (BOS) – 4.8

The list combines players who are lower volume but so efficient they make the best use of their value to be high enough to make this list such as Brandan Wright, Tyson Chandler and Anthony Morrow types, and players who use many more possessions at a less efficient but still above average rate such as a Russell Westbrook or Kyrie Irving. Chris Paul proves to have the highest rating combination of elite efficiency at a quality amount of possessions.

As an aside (skip this part to get to the rest of the formula if you like), the origins of this part this “Efficiency” stat goes back several years for me. There was a time when I thought this formula alone could be enough to rate NBA players. One difference is I compared the player’s possession efficiency to the team’s on-court defense with the player on the court. Therefore it would capture how much more efficient player’s possessions are vs *his opponent*. My thought process at the time is this could explain all wins. Using possessions more efficiently than your opponent adds to the team’s chances of winning, using it less efficiently than them decreases their chance of winning. This is the essence of the difference between winning and losing in the NBA.

But while my statistic at the time did an excellent job explaining teams wins the season before, it’s predictive power proved poor. To explain why I’ll call it the “Kosta Koufos problem”. When creating the stat it was before the 2013-2014 season. The Grizzlies the season before had traded Rudy Gay, won 56 games and made the Western Conference Finals. With John Hollinger’s savvy on their side in the summer they acquired Kosta Koufos, Ed Davis and Mike Miller all of whom looked excellent in my stat. Kosta Koufos rated as one of the top 50 players in the league and a coup for the Grizzlies. In Denver he had used possessions at a great 122 ORTG and his team played quality (103.7) defense with him on the court. Ed Davis in his first half season with the Grizzlies had been at 113 ORTG with his team at 98.6 DRTG on the court and he projected to get an uptick in minutes by the Grizzlies. They got Mike Miller at 117 ORTG with his team at 105.9 ORTG his last season in Miami. On paper Hollinger had set up the Grizzlies to be the best team in the league. They had taken a 56 win, already perhaps better than that by subtracting Rudy Gay, and complimented their elite defense with the efficiency possession users to make them strong on offense. I predicted them to finish 1st in the West.

In reality the Grizzlies went 50-32 and squeaked into the playoffs. Marc Gasol’s injury didn’t help, but they had only started 7-5 up to that point regardless. Kosta Koufos efficiency fell from 122 to 106 in the Grizzlies system, making him not near the star upgrade as projected. Koufos was an efficient scorer in Denver, but as with Ed Davis didn’t space the floor and took shots at a low volume. Understanding the need to value factors such as volume scoring creating space for opponents, spacing the floor and better ways to value defense in the case of Koufos type players, is one of the reasons I created VEDS instead.

**Part 3: Defense (D)**

The need to evaluate defense needs little explanation. The question that may be asked is since defense is half the game, is it fair that defense only accounts for 1/4th of this statistic while Volume, Efficiency and Spacing are the remaining 3/4ths?

This is a fair point and it’s entirely possible I’m wrong for giving defense this low a weight. My justification is calling defense more of a team activity than offense. On defense one man can’t be everywhere but have the entire offense built around him. I am not alone in believing this. For example Andrew Bogut is considered an elite defensive player with non-existent offensive impact and James Harden is an elite offensive player with non-existent defensive impact. Harden is rated as superstar as his defense can be covered up by other players, while his offense is invaluable.

To evaluate defense, there is no perfect stat, I can only use the best available option. I decided ESPN’s defensive real plus minus (DRPM) which uses adjusted +/- and boxscore information, had acceptable enough results to use.

My goal with this stat was to have it on a scale comparable to my Volume and Efficiency categories. Last year Draymond Green led the league at 5.23 DRPM. My first thought was to multiply DRPM by 2.5, and then divide their minutes by 36 to account for minutes played. Minutes played would be taking into account because categories like Volume and Efficiency are affected by having the minutes to use more possessions per game. This gave me 4 players over 10 in DeMarcus Cousins, Draymond Green, Anthony Davis and Kawhi Leonard, with Cousins taking over the top spot because of more minutes played than Green.

However I had a problem with how this affected the players at the bottom of DRPM. By multiplying with DRPM, the players who had some of the poorer DRPMs in the league ended up with a negative score far beyond the worst rating players in my other categories. For example Enes Kanter’s -3.87 DRPM translates to -9.7 when multiplied by 2.5. This would make his lack of value on defense as negative as a superstar offensive players in a category like volume or efficiency. Even players with a more acceptable score like -2 on DRPM would take a major hit of -5 in their overall VEDS threatening to wipe out most of the positive value in other categories.

The formula I settled on is taking the player’s DRPM, multiplying it by 1.5, and then adding 5. Followed by dividing their minutes by 36. For example Draymond Green’s DRPM is 5.23, multiplied by 1.5 and adding 5 gets to roughly +12.8. Adjusting for Green 31.5 minutes per game scales his rating back to +11.2, once again sliding behind Cousins for the top spot.

This gives me the same top 4 players who break the 10 threshold in Cousins, Green, Davis and Leonard. The impact on the negative players is not as destructive however. A rating of -2 in DRPM would translate to a rating of 2 in Defense, more analogous to what a poor possession user rates in the Volume category. Kanter comes out at -0.6.

Here are the top 20 players in Defense per game:

DeMarcus Cousins – 11.4

Draymond Green – 11.4

Anthony Davis – 11.3

Kawhi Leonard – 10.5

Tim Duncan – 9.6

Serge Ibaka – 9.4

Khris Middleton – 9.3

Tony Allen – 9.0

Lebron James – 9.0

Tyson Chandler – 8.7

Nerlens Noel – 8.6

Andrew Bogut – 8.6

Trevor Ariza – 8.5

Michael Kidd-Gilchrist – 8.4

Markieff Morris – 8.4

Deandre Jordan – 8.3

Zach Randolph – 8.2

Wesley Matthews – 7.9

Rudy Gobert – 7.6

Paul Millsap – 7.6

With only DRPM and minutes per game leading to these numbers, I am fairly satisfied with that list rating the top 20 most impactful defenders per game in the league. The minutes help weed out some players who stick out on the DRPM list such as Jusuf Nurkic and Darrell Arthur who rated 9th and 10th in DRPM, but fell out of this top 20 due to minutes played.

**Part 4: Spacing (S)**

The obvious place to start with a spacing statistic is distance of shot from the rim, tracked on basketball-reference.com.

But using distance along brings various problems. A player who takes 2 3pt shots from 23 feet and one shot from 1 feet, rates as 15.7 feet from the rim on average for the 3 shots. How do you value this player’s floor spacing vs a midrange shooter who takes 3 shots from 16 feet?

What I stumbled into as a powerful way to account for this is assisted %. I believe this is because 3 point shots are assisted at a much higher rate than midrange jumpers. By taking a 3 compared to a midrange shot not only is the player shooting a distance from the hoop but it is more likely his shot is assisted. So for example if Klay Thompson takes 2 3pt shots and one from the rim and DeMar Derozan takes 2 midrange shots, the extra liklehood Derozan’s shots are unassisted is a factor in evaluating how floor spacing-friendly each of those 3 shots are.

Using assisted % punishes ball dominance. For example on the Houston Rockets last year Patrick Beverley and Trevor Ariza rate far greater in floor spacing than James Harden despite not a large difference in 3pt shooting skill. The logic behind this is that Harden taking 3s off the dribble does not “space the floor” like Ariza standing at the 3 point line off the ball. Ariza draws defenders away from on ball creators like Harden. If Harden takes a shot off the dribble, his 3pt shot didn’t draw defensive attention away from a teammates FGA attempt – because he took the shot!

As outlined before however, it doesn’t mean that Harden isn’t credited for “creating space” for his teammates by drawing double teams and collapsing the defense for penetration. It’s just Harden’s creating space impact is in the Volume category, Beverley and Ariza’s are in the Spacing category, which is really more meant for off the ball, spot up shooters who drag defenders away from the rim.

How I calculate this:

Step 1: I start by dividing the player’s distance from the rim by 23.

Kyle Korver’s shots last year were on average 22.2 feet away. Divided by 23 this is .965.

Step 2: I take the player’s assisted % on his FGs. To calculate this on basketball reference I multiplied the % of a player’s FGs that were from 2 by the % of them that were assisted, added to the % of a player’s FGs that were from 3 multiplied by the % of them that were assisted.

95.1% of Korver’s shots were assisted.

Step 3: I multiply the distance from the rim stat by the assisted FGs one.

Since Korver’s “Distance” number was .965 and his assisted FG % was .951, multiplied together this is .917.

Step 4: I multiply the above number by 15, followed by dividing the player’s minutes by 36. This is to scale it to 10 like the other categories.

Korver’s .917 times 15 is 13.7, or 12.3 after the minutes adjustment.

Korver’s 12.3 is the only player who rates above 10 in the category, with J.J. Redick coming closest at 9.7 and Trevor Ariza, Channing Frye and Mike Dunleavy rounding out the top 5. I didn’t scale the stat up slightly more to match the other categories, because the main reason for less elite performing players in the category is minutes. Redick (30.9 minutes), Frye (24.9 minutes), Dunleavy (29.2 minutes) and Anthony Tolliver (22.3 minutes) rated above 10 before the minutes adjustment. The best floor spacers in the league aren’t as likely to be valued enough to play 36 minutes per game as the highest volume possession users in the league, for example.

Here is the top 20 in spacing per game:

Kyle Korver – 12.3

J.J. Redick – 9.7

Trevor Ariza – 9.1

Channing Frye – 8.7

Mike Dunleavy – 8.5

J.R. Smith (CLE) – 8.4

Danny Green – 7.9

Matt Barnes – 7.8

Wesley Matthews – 7.7

Marvin Williams – 7.6

Kevin Love – 7.4

Avery Bradley – 7.3

Anthony Tolliver (DET) – 7.3

Nicolas Batum – 7.3

Ben McLemore – 7.2

C.J. Miles – 7.0

Jose Calderon – 7.0

Serge Ibaka – 7.0

Henry Walker – 6.9

DeMarre Carroll – 6.9

For a purely quantified way to judge spacing, I am more than happy with the above list. Sure it may not be perfect, but like defense, it’s purpose is to be close enough to work with to rate how much of an off the ball floor spacer a player is. There are no real frauds on the list and the presumed right players are at the top in Korver and Redick.

**Adding it all together**

The last step is simple. I add up the players score in all 4 categories to get one total stat – VEDS per game. Here is the top 50:

Anthony Davis – 35.4

Stephen Curry – 30.9

Kevin Durant – 30.4

James Harden – 28.2

Jimmy Butler – 28.1

Kyle Korver – 27.6

Lebron James – 27.2

Chris Paul – 26.2

J.J. Redick – 25.7

Wesley Matthews – 25.4

Draymond Green – 24.9

Kawhi Leonard – 24.8

Kevin Love – 24.7

Blake Griffin – 23.7

Serge Ibaka – 23.7

Trevor Ariza – 23.5

DeMarcus Cousins – 23.2

Russell Westbrook – 22.8

Klay Thompson – 22.4

Kyrie Irving – 22.3

Khris Middleton – 22.0

Gordon Hayward – 21.7

Pau Gasol – 21.5

Danny Green – 21.5

Tim Duncan – 21.2

Marc Gasol – 21.1

Mike Dunleavy – 21.0

Al Horford – 20.9

Lamarcus Aldridge – 20.8

Tyson Chandler – 20.7

Damian Lillard – 20.6

Kyle Lowry – 20.6

Paul Millsap – 20.2

J.R. Smith (CLE) – 20.0

Derrick Favors – 19.8

Chris Bosh – 19.6

DeMarre Carroll – 19.3

Deandre Jordan – 19.2

Chandler Parsons – 19.1

Anthony Morrow – 19.0

Matt Barnes – 18.9

Dirk Nowitzki – 18.7

Darren Collison – 18.2

Eric Bledsoe – 18.2

Zach Randolph – 18.0

John Wall – 17.9

George Hill – 17.8

Nikola Vucevic – 17.8

Markieff Morris – 17.5

Marcin Gortat – 17.4

The main difference is the list of the traditionally accepted star players, are invaded by the likes of Kyle Korver, J.J. Redick, Wesley Matthews, Draymond Green, Trevor Ariza, Khris Middleton, etc. The type of player who seems to suffer the most compared to consensus opinions are ball dominant guards who aren’t geared towards the 3 such as Russell Westbrook who rates as an excellent player but not an MVP caliber one, John Wall who sneaks into the top 50, and Dwyane Wade is unlisted but ranked out of the top 150.

Since the above stat takes into account both games and minutes played there are variations that could be made from it. Multiplying VEDS by games played/82, gives a games adjusted rating of value and would be a more fair way of judging the most valuable seasons in the league. After this games played adjustment Stephen Curry does indeed come out as the MVP by edging out Davis, followed by Harden, Paul and Korver in the top 5.

I also found it useful to create a per minute VEDs stat, which I calculated by dividing their VEDS by minutes per game. For example of the above top 10 in total VEDs, here is their VEDS in per minute form:

Anthony Davis – .98

Stephen Curry – .94

Kevin Durant – .90

James Harden – .77

Jimmy Butler – .73

Kyle Korver – .86

Lebron James – .75

Chris Paul – .75

J.J. Redick – .83

Wesley Matthews – .75

There were 163 players with a rating of .50 or better last year, although this includes players who barely got on the floor and players were counted twice if they played on multiple teams. When discounting those it’s clear being above .50 makes a player a quality contributor. The 409th highest per minute VEDS was .30, which is a fair estimate for replacement level. Therefore creating a VORP stat using .30 as replacement level would also work.

Per minute VEDS was a useful way to make projections for next season. For example in VEDS per minute the Spurs added Lamarcus Aldridge who rated 29th in VEDS and lost Tiago Splitter (156th), Marco Belinelli (205th), Cory Joseph (222nd), Aron Baynes (248th), Matt Bonner (334th). Looks great right? But minutes played per game has a major part of Aldridge rating so well compared to them. In VEDS per minute stats Aldridge rates at .59 compared to Splitter (.59), Baynes (.54), Joseph (.51), Bonner (.50), Belinelli (.45). From this perspective Aldridge may not produce more than the platoon of Splitter, Baynes and Bonner alone in the frontcourt, in addition to the Spurs losing two quality perimeter pieces in Joseph and Belinelli.

For the record using my projection system here is what I predicted for team wins this season. These predictions were posted in the APBR forum prediction contest before the season started:

East

1. Cleveland Cavaliers – 57 Ws

T-2. Atlanta – 54 Ws

T-2. Toronto – 54 Ws

4. Chicago – 47 Ws

5. Washington – 43 Ws

6. Boston – 41 Ws

7. Detroit – 39 Ws

8. Indiana – 38 Ws

9. Charlotte – 35 Ws

10. Miami – 34 Ws

11. Philadelphia – 33 Ws

12. Milwaukee – 30 Ws

13. Orlando – 26 Ws

14. Brooklyn – 20 Ws

15. New York – 14 Ws

West

1. Golden State – 66 Ws

2. Oklahoma City – 58 Ws

3. San Antonio Spurs – 56 Ws

T-4. Utah Jazz – 52 Ws

T-4. L.A. Clippers – 52 Ws

6. Memphis Grizzlies – 50 Ws

T-7. New Orleans Pelicans – 48 Ws

T-7. Houston Rockets – 48 Ws

9. Portland Trail Blazers – 43 Ws

10. Phoenix Suns – 42 Ws

11. Dallas Mavericks – 39 Ws

12. Sacramento Kings – 36 Ws

13. Denver Nuggets – 30 Ws

14. Minnesota – 25 Ws

15. L.A. Lakers – 20 Ws

I am interested to see how these predictions do and whether they can prove VEDS has predictive power.

## Leave a Reply