Chris Paul, the San Antonio Spurs and help defense as a cost function

May 2, 2015; Los Angeles, CA, USA; San Antonio Spurs forward Tim Duncan (21) fouls Los Angeles Clippers guard Chris Paul (3) in the fourth quarter of game seven of the first round of the NBA Playoffs at Staples Center. Paul hit both free throws to tie the game at 109. Clippers won 111-109. Mandatory Credit: Jayne Kamin-Oncea-USA TODAY Sports

I guess by now all of you have seen Chris Paul’s game winner a hundred times. While looking at one single play is no quantitative analysis, it nicely illustrates how the Spurs help defense was reluctant to give up a shot to anybody else in this series. Seth observed it as well recently and it seems like the Spurs game plan was to make Chris Paul more of a scorer than a distributor. Using nba.com playtype data, we can see that the Spurs defense seems to have taken their defensive scheme to the extreme

During the regular season, the Clippers scored in 100 possessions around 14.4 points (0.9 PPP * 16.0% playtype frequency) with their Pick and Roll Ball handler and 19.8 (1.05PPP * 18.9%) with spot up shooting. During the first round of the Playoffs this changed to 18.0 points (1.03PPP * 18.9%) with the ball handler and a paltry 12.4 points (0.86PPP * 14.4%) with spot up shots.

The 1.03 points per possession for the ball handler is staggering. During the regular season, the Clippers mark of 0.9 points per possession was already league leading (of course, a smaller sample size is helpful for staggering numbers. Even the Bulls ball handler scored 0.99 points per possession during the last six games). The Clippers pick and roll ball handler scored 1.1 PPP on 67 possessions (Paul) and 1.08 PPP on 37 possessions (Crawford). It only came down to 1.03 PPP team average because Austin Rivers scored 0.63 PPP on 16 possessions (I’m sure there is a hidden joke somewhere).

Why did the Spurs allow this to happen?

Of course I am not inside Gregg Popovich’s brain (but I am sure that is a very beautiful place), but the general idea of how much help-defense one should send is a long and ongoing debate. The general consensus right now seems to be that teams tend to overhelp on plays like isolations, pick and rolls and post-ups therefore giving up more efficient plays like spot-up shots, cuts and easy scoring from the PnR roll guy. This is obviously disadvantageous, because the first group of plays (‘non-pass plays’) usually yield around 0.8 points per possession, while those plays that are subsequent to a pass yield around 1 point per possession.

So, why do players help at all?

An interesting aspect of this whole thing is if we look at the correlation between defensive PlayType frequency and Points per possession. If we look at the instance of the Pick and Roll where the ball handler scores, we find that there is a clear correlation between play type frequency and points, which indicates that teams that help less have the drawback of conceding more points due to this play:

(Interestingly, the correlation for isolation plays is negative (-0.36). But isolation plays occur much less frequent and are often used as a last resort.) ‘Pass plays’ on the other hand show basically no correlation between occurrence and points per possession. Showing spot up shooting as an example:

The reason is that you usually take these shots very deliberately. You take the corner three after a pass only when you are open and you can hardly become more open than open. The same goes for cuts. Once a cut worked out you are getting to the basket and the amount of subsequent help only differs slightly.

Minimization of a cost function

Coming back to the Spurs. As you can see in the previous two plots, most of the good defensive teams are rather allowing your pick and roll ball handler to score, while taking away your additional shooting threads. The Spurs probably tried exactly this. This strategy can work, if there is only a slight increase in efficiency for the pick and roll ball handler. And usually you probably concede 3.5 points more to the ball handler, while taking away 4 points from the spot-up shooters (Note: In the beginning I thought about writing a small formula. But it would have been so crude that the rule of thumb is probably just as good) . Unfortunately for the Spurs, Chris Paul does not care about your cost minimization strategy.