SoTD: Net DRTG

[supersub15]

Obviously, there is no one-size-fits-all measurement for individual defensive prowess. There are many ways to do it, but each has its warts. One of these ways (and, of course, it has its own warts too) to measure a player's defensive impact is to see how the team performs defensively with the player on the court, and with the player off the court.

What I did was calculate team DRTG with the player on the court and the team DRTG with the player off the court, then deduct one from the other to see if the team becomes better or worse when the player is off the court. The criteria I set is simple, the player has to have played at least 750 minutes of NBA ball to matter (that represents about 9 minutes of court time over an 82-game season). However, I did take into account injured players (like Joel Przybilla), who normally play more than 9 minutes per game.

Top-10 in defensive impact on team:

Code: Select all

Player Name      Min   DRTG ON  DRTG OFF  Net
Amir Johnson    1457   107.92   116.17   -8.25
Eric Maynor      904    97.74   105.94   -8.19
Ben Wallace     1974   107.36   115.19   -7.83
Nick Collison   1560    99.50   107.06   -7.56
Joel Przybilla   683   100.88   108.26   -7.38
Rodney Stuckey  2501   108.71   115.75   -7.04
Keyon Dooling    970   105.28   111.64   -6.36
Stephen Graham   804    98.30   104.65   -6.34
Jared Jeffries  1464   107.21   113.50   -6.29
Chris Andersen  1696   104.34   110.55   -6.21

Worst-10 in defensive impact on team:

Code: Select all

Player Name     Min   DRTG ON   DRTG OFF  Net
Brook Lopez     3025   111.53   105.49   6.04
David West      2948   111.63   105.56   6.06
C. Villanueva   1849   114.61   108.42   6.19
Will Bynum      1669   115.14   108.53   6.61
Jonny Flynn     2341   114.41   107.33   7.08
Devin Harris    2218   113.29   105.95   7.34
J.J. Hickson    1692   108.64   101.20   7.44
M. Speights     1017   116.45   108.05   8.40
Andrea Bargnani 2801   115.87   106.77   9.10
Jeff Green      3041   106.55    95.94  10.61

These lists doe’t tell us who are the best defenders. They only tell us who should have played more when the team was having defensive problems. For instance, Amir’s 107.92 ON the court isn’t stellar on its own. That doesn’t mean he’s a bad defender either (it’s all about the context). As an example, Amir’s last year with Detroit:

Code: Select all

Player Name   Min   DRTG ON   DRTG OFF   Net
Amir Johnson  908   100.49    109.42   -8.93

Also, there are 2 main factors that can influence the numbers: Who's on the court with the player, who's on the court for the opponent. So, this is not the end-all, be-all stat. For instance, Bargnani was +2.31 and +0.74 in his first 2 seasons under Mitchell (and with Moon/Parker at the wings), but +7.41 and +9.10 in his last 2 seasons under Triano (and with Marion/Parker and Turk/DeRozan at the wings). There's no denying that he's a bad defender, but maybe the defensive philosophy has an impact (Triano!) and he needs to play with capable defenders in front of him. Also, watch out for something else: the Net DRTG doesn’t tell us that Green was still a decent 106.55 when on the court. It’s just that his replacement was even better.

However, over thousands of minutes of play, you can determine who's having a defensive impact on his team by his presence on the court, and who's not. And the 2 lists pass the smell test.

Raptors (750+ mins):

Code: Select all

Player Name     Min   DRTG ON   DRTG OFF   Net
Amir Johnson    1457   107.92   116.17   -8.25
Marco Belinelli 1120   110.09   114.37   -4.28
Antoine Wright  1390   110.53   114.60   -4.07
Chris Bosh      2527   112.42   114.46   -2.04
Sonny Weems     1365   111.85   113.86   -2.02
Jarrett Jack    2245   113.70   112.42    1.28
Jose Calderon   1817   114.69   111.87    2.83
DeMar DeRozan   1667   115.24   111.63    3.62
Andrea Bargnani 2801   115.87   106.77    9.10

New Raptors:

Code: Select all

Player Name      Min   DRTG ON  DRTG OFF   Net
Leandro Barbosa  783   107.77   110.56   -2.79
David Andersen   893   106.26   108.79   -2.53
Julian Wright    873   109.26   110.34   -1.08
Linas Kleiza    1821   107.70   105.60    2.10


[5DOM]

I thought Jose Calderon was by far the worst defender in the league.
Not even in the bottom 2.

[Fairview4Life]

Seems about right. Is there any way to slot individuals in and out of 5 man units on the Raps, or are the sample sizes just way too small to matter? Like Amir vs. Bargs in various lineup combinations.

[supersub15]

I thought Jose Calderon was by far the worst defender in the league.
Not even in the bottom 2.

Again, this is not who the worst/best defenders are.

Who's replacing Jose in the lineup when Jose is OFF the court? Jack. What's Jack's Net? See what I mean? It's a whole team dynamic at play. Plus, you can't compare players across teams. You have to do it within the team.

[Spragga]

I thought Jose Calderon was by far the worst defender in the league.
Not even in the bottom 2.

this was on hoopshype this morning

JCalderon's D is indeed as bad as advertised. In 9th %ile of NBA players in overall D, ranks last among 166 players w/ min 600 D possessions

Calderon also ranks last among 65 players who defended the ballhandler on at least 150 possessions.

Last 2 tweets contain data from Synergy Sports Technology. Stats ranked by Points Per Possession.

[chyau.00]

I thought Jose Calderon was by far the worst defender in the league.
Not even in the bottom 2.

i think this stat depends on who replaces him .....and clearly jarrett jack isnt much better...

[derrek]

Very cool. I'm interested to see the difference after this year as a whole. I think our defense will be improved!

[lolwut]

IL MAGO is now IL MATADOR

[[SJJ]]

I thought Jose Calderon was by far the worst defender in the league.
Not even in the bottom 2.

this was on hoopshype this morning

JCalderon's D is indeed as bad as advertised. In 9th %ile of NBA players in overall D, ranks last among 166 players w/ min 600 D possessions

Calderon also ranks last among 65 players who defended the ballhandler on at least 150 possessions.

Last 2 tweets contain data from Synergy Sports Technology. Stats ranked by Points Per Possession.

Wow - I guess it just quantifies what we've been seeing over the past couple of seasons.

[Crazy-Canuck]

Anyone else surprised with how bad Green was last year defensively?

[supersub15]

Anyone else surprised with how bad Green was last year defensively?

If you read correctly, you would have figured that 106 ON isn't bad. His replacement was that much better (Collison is in the top 10, and probably played in combination with Ibaka who was -2.60)

[theonlyeastcoastrapsfan]

Anyone else surprised with how bad Green was last year defensively?

Was he real bad or was Ibaka and Colison real good?

Edit: Word up, Sub.

[Too Late Crew]

I thought Jose Cathe
lderon was by far the worst defender in the league.
Not even in the bottom 2.

You can't make that judgment from the numbers above. As the OP fairly stated [quote]Net DRTG doesn’t tell us that Green was still a decent 106.55 when on the court. It’s just that his replacement was even better.
{/quote]

Its not that Jose isn't terrible. When he plays on the floor the Raps are terrible on D. However when a unit without him plays they are also terrible on D. I take that to mean that the Raps are simply terrible on D no matter who plays.

Persoanlly I'd probaly raise the bar for minutes played from 750 to mayeb 1500. If you are not playing close to 20MPG I don't consider that sampel size all that meaningful. Too many garbage minutes.

Its also fair that the OP ackowldges you need to look at who they play with..but in additon who they do NOT play with.

Just for example. If Amir is subbing for Bargs who obviously terrible on D he lloks much better than if he Subs in for say Bosh who statitically is not as bad.

It does highlight one other thing. CACK is an Epic fail. Bargs is going to get his end of game minutes (BC will see to that) So does it really seem a good idea to put two of the other 3 guys on the team that have a negative defensive impact on the game out there with him?

[Courtside]

I thought Jose Calderon was by far the worst defender in the league.
Not even in the bottom 2.

Again, this is not who the worst/best defenders are.

Who's replacing Jose in the lineup when Jose is OFF the court? Jack. What's Jack's Net? See what I mean? It's a whole team dynamic at play. Plus, you can't compare players across teams. You have to do it within the team.

So if it wasn't Amir at -8.25, but someone who was a net-zero player, then Andrea's number would be more like 0.85? Or is Amir's -8.25 number inflated by Andrea's +9.10? How does that explain Amir's -8.93 in Detroit? Does his defense sway Charlie V and Andrea's numbers even further than they really are, or do their poor numbers sway his?

[Too Late Crew]

Anyone else surprised with how bad Green was last year defensively?

His 106 on would make him the best defender on the Raptors. That's how bad the D was last year. The "worst defender" on the Tunder would be our best.

I expect the D to improve this year. It can't get worse than being 30th.

Its funny BC was quoted in the media regarding advanced stats showing how well Amir and Bargs play together. Yet the advanced stats show how absiolutley horrible Bargs effect is on D yet he's still out there playing 35 mpg. Just more data that supports the idea that BC simply did not care about defense when he constructed last year's team.

[dinosaur_dan]

Consider the case of two very bad defenders subbing for each other. The Net DRTG would be close to 0, since they would both have low DRTG. Are they both average defenders? You get the same result for two very good defenders - a Net DRTG close to 0. This stat tells us only one thing: the relative impact of a player vs. their sub.

In the case of AB and Amir, the stat tells us that Amir is a much better defender than AB, which of course we already knew. Does that mean AB is the worst defender in the league and that Amir is average? Or does it mean that AB is average and that Amir is the best in the league? Net DRTG won't tell us that.

All we know for sure is that that the difference in defensive impact between AB and Amir Johnson represents the largest gap between a player and their sub in the league.

[Courtside]

This stat tells us only one thing: the relative impact of a player vs. their sub.
...

All we know for sure is that that the difference in defensive impact between AB and Amir Johnson represents the largest gap between a player and their sub in the league.

I was trying to figure out how valuable these numbers really are - but explained the way you did above - it makes a whole lot more sense.

The defensive difference between Amir and either Charlie V or Andrea is large - we all know that. Using this stat across 30 NBA teams and 450 players is where it's almost useless.

[RapsFanInVA]

So because Amir is at one extreme and Bargs is at the other, does that mean they account for each other? Either Amir's numbers look amazing because Bargs is horrendous, or Bargs' numbers are awful because he's being replaced with a superior defender. I'm assuming it's a bit of both.

It'll be interesting to see the numbers this year because I expect Amir and Bargs to play a lot together and be replaced by rookies and crap.

We might be able to twist the stats to convince ourselves Amir and Bargs are a good defensive combo.

[supersub15]

I was trying to figure out how valuable these numbers really are - but explained the way you did above - it makes a whole lot more sense.

The defensive difference between Amir and either Charlie V or Andrea is large - we all know that. Using this stat across 30 NBA teams and 450 players is where it's almost useless.

10-Worst players in DRTG ON:

Code: Select all

Player Name         Min    DRTG ON
Ryan Hollins        1224   113.90
Jonny Flynn         2341   114.41
Charlie Villanueva  1849   114.61
Jose Calderon       1817   114.69
Hedo Turkoglu       2271   115.13
Will Bynum          1669   115.14
DeMar DeRozan       1667   115.24
Reggie Williams      779   115.33
Andrea Bargnani     2801   115.87
Marreese Speights   1017   116.45


That's a fine collection...

[I_Like_Dirt]

One factor I don't see you mentioning is that those numbers are greatly influenced by how a coach uses a player as well. People sometimes don't like to believe it, but in cases where players don't play many/regular minutes, it's often the case that the coach puts the player in to see if he's effective or not and quickly yanks him if he isn't. In the event that the player succeeds, he gets left in to pad his stats, and over the relatively small sample size of minutes, it's a pretty significant source of error. It largely explains Bargnani's drop from looking decent under Mitchell when he was being taught through teaching to his failure under Triano where he was supposedly being taught by trying (and failing) and just being left in the game for 35+ minutes no matter how poorly he was doing and how little he looked like he understood what was going on out there.

For the Raptors last season, the player who was by far the most benefitted statistically from that effect was Marco Bellineli, but I'd suggest Amir Johnson came in a distant but solid second. Not only was he yanked when he struggled at times, but often he would go in, rack up 2 or 3 quick fouls where the opposition wouldn't score, and watch the Raptors go into the penalty afterwards. It wasn't an outrageously common thing with him, but far moreso with him than any other Raptor or most other players in the league. What this number says is that when he's on his game, he's a very effective defender, and the Raptors need to figure out how to teach him to be on his game all the time.