Spurs improve defensive efficiency by focusing less on defensive rebounding

Yeah I think they're just getting their perimeter players to play tighter and contest more shots. That means they get less defensive rebounds since the perimeter guys are further from the basket.

 

When advanced stats people talk about defensive rebounding performance, they're not referring to raw defensive rebound totals. They're talking about defensive rebound% which is the percentage of missed shots that are rebounded by the defending team. The numbers I posted in this thread was looking at rebounding as a percentage of missed shots.

 

Defensive rebounding percentage was correlated with winning tonight.

 

durvasa, yes I understand. I'm just saying you might be using stats to answer a flawed question. The very question is what I don't agree with. There are more variables. Playing "tight perimeter defense" can also mean less outside shot. You can funnel players into the defenders or make it so tight on the outside the opp is forced to pass it inside to lessor scorers.

Plus the converse. None of it could even mean a better defense. You can overplay the perimeter and allow dribble penetration for easier shots

Some people in this thread are making a wrong assumption by saying tight perimeter defense always equates to better defense. That all depends on many variables

The journalists is making a sweeping conclusion based on a flawed hypothesis, imo

 

The only point of view I see in the article is that contesting shots is more important than crashing the defensive boards. That's not to say defensive rebounding isn't important, but if you're going to emphasize one over the other than emphasize contesting shots. I think this is sensible, and the stats show its a smart strategy.

Also, it should be pointed out that over-playing on the perimeter and allowing dribbling penetration isn't the same thing as contesting shots. Effectively contesting shots includes containing dribble penetration, because if you can't do that then chances are you'll actually give up more open shots later in the possession.

 

durvasa. I'm saying "tight perimeter defense" does not equate to contesting shots efficiently. For instance, Pick n roll defense is more complicated than that. Sometimes PnR defense gives up the shot

Also lets not forget about fastbreak defense. Release for easy buckets and don't care as much about opp offensive rebounds because their 2nd chances don't score as much as your fast break first chances



Of course, I have now stats to defend my more complicated opinion

 

durvasa. Where do your stats prove teams are contesting shots on the perimeter. Aren't your stats just assuming that

 

I don't think its mere assumption to think that the effective FG% allowed to an opponent correlates highly with how well the team contests shots.

 

If you're giving up a good shot without contesting it, then generally speaking something went wrong in your defense.

Now, strategically giving up one shot to prevent a better shot may make sense. Giving up a shot to a really bad shooter may also make sense. But giving up a decent shot in favor of hanging back to get the rebound in case it misses? Probably not a good decision.

 

The journalist is not saying that at all. He's implying worse Drb% equates to a better defense

Am I missing something? Because that is a very naive statement

And while I respect your work more than I have told you. I think you are trying to prove a wrong assumption

 

I never fully argued that. I just don't see where your stats prove the journalist correct

imo you took his question as a challenge to prove. And I'm saying his very hypothesis is questionable

 

I don't agree that's what he's saying; maybe you can quote the part where you feel he's implying this. And that's definitely not what the stats I posted show. I'll repost the earlier plots:



Lower right corner shows opponent efficiency versus opponent's ORB%. Its positively correlated. If higher Drb% lead to worse defense, and lower Drb% lead to better defense, I would expect it to be negatively correlated.

 

Forgive me I'm in Germany watching this thread on a phone.im watching this thread because you are posting in it. I love your stats I can't really follow your charts on my iPhone in my hotel

In very simple terms. What are you showing. Please don't tell me you are trying to show proof for a new defensive strategy

 

Not sure I proved anything. But I wanted to see the relationship, if any, between various factors and offensive/defensive efficiency. I showed, based on data from 1974 through 2012, that eFG% allowed explains overall defensive efficiency to a significantly greater degree than DRB% explains it (R² of 72% versus 24%).

And I knew that work done in the past by Dean Oliver (who's like the basketball equivalent of Bill James) already determined that making shots and defending shots is the most important thing in basketball (link), so I'm not trying to show anything really novel here:

...
Having these weights on the Four Factors suggests how to build a team. Get players who can put the ball in the basket and, very importantly, players who can keep opponents from putting the ball in the basket. Those are the first things to consider in constructing a solid team. At the NBA level, shooting and stopping shooting are two very different skills and knowing which players to go after first becomes important. This relates to the scarcity of talent. At different times in NBA history, one type of player has been more difficult to obtain than the other. In the 1980's, offense won - so getting shooters was more important. In the 1990's, defense won and obtaining players who could shut down opposing shooters became more important. But note that improving shooting or shutting down shooting - whichever a team chooses - is more important than controlling turnovers or getting rebounds or getting to the line (or their defensive counterparts). Keeping this in mind while starting a team from scratch or while building a team from its current components will make for a shorter road to success.
...
​

Not exactly a radical viewpoint, but its nice to see it backed up by evidence.

 

Where in your stats do you show quality of shot? It's not there because your stats can't show it. That's what I'm saying. You are using stats to prove a hypothesis that could be entirely wrong. I contend you are jumping to a conclusion that is flawed at its very basis. And the only reason we are talking about this is because one journalist suggested it and you took that as a challenge to make charts

Well, I took it as a challenge too. And I love this game of bball. And I love defense. And no journalist is going to tell me he found a new defense. Not even JVG would tell me that

 

Teams that allow high eFG% are giving up too many open, quality shots. What other explanation is there for it?

Still not sure which hypothesis you think I set out to prove. And any hypothesis, by definition, can be wrong. The point of a scientific investigation is to try to test claims that are uncertain.

 

And what does that have to do with drb%. There is no correlation. You are trying to prove quality of shot from box scores. Then you are trying to match that to Drb%. Then you are further trying to say all of those assumptions mean the defense is better

Nope

 

durvasa. You can't use boxscore stats to make claims about nba defenses. Not when I'm around

 

Sorry, you lost me.

Let's re-quote R.C. Buford:

[rquoter]
“This summer, we looked at our defensive efficiency, which for years had been very high. And last year, we went in the 10-15 range. And I think we were valuing some things that weren’t nearly as important as the data showed us. We learned from the Celtics.

“While they were really high in defensive efficiency, they weren’t very high in defensive rebounding. And that was a big part of where our emphasis was, and it made us question is that really where we should be paying attention. And those were discussions that were then brought to Pop from our coaches and from our analytics team. And some great discussions came from that, that ended up having us reevaluate what was important to us.”
[/rquoter]

He says that the Spurs shifted their emphasis from defensive rebounding to other aspects of defense.

What do you think that means?

 

Using the same data (a sample of 1013 team-seasons from 1974-2012), I regressed the z-score for DRTG onto the z-scores for opponent eFG%, opponent ORB%, opponent TOV%, and opponent FT/FGA (or FTR). The resulting model had an R² of 96%:

DRTG_std = 0.74*oppEFG%_std - 0.53*oppTOV%_std + 0.33*oppORB%_std + 0.25*oppFTR_std

I interpret these coefficients as telling me the relative importance of each factor within this model, holding the other factors constant. The sign on the oppTOV%_std coefficient is negative because unlike eFG%, ORB%, or FTR, TOV% makes your efficiency worse.

In actuality, these factors aren't independent of eachother. While the above model says that TOV% is a bigger factor in resulting team efficiency than ORB%, teams that force more TOV% from their opponents also tend to do a worse job in the other 3 defensive factors which was shown the correlation table on the previous page. So a defensive strategy that focuses on forcing turnovers would not be sound if it comes at the expense of suffering in the other areas (contesting shots, defensive rebounding, keeping teams off the free throw line).