By Stephen Shea (@SteveShea33) and Christopher Baker (@ChrisBakerAM)
October 30, 2015
Defenses can consistently influence opponent shot selection.
Red Auerbach once said, “Basketball is like war in that offensive weapons are developed first, and it always takes a while for the defense to catch up.” Perhaps then, it should come as no surprise that basketball analytics has followed the same pattern. Offensive analytics are far ahead of defensive analytics.
Offensively, we count players’ passes, but does anyone measure the extent to which an opposing defender prevents the pass? We measure a shooter’s 3-point FG%, but who’s counting how often a defender runs a shooter off the 3-point line? We look at how often a player drives to the hoop, but do we measure how often a defender keeps his man in front of him and away from the rim?
Historically, there have been two serious impediments to defensive basketball analytics. First, often the best defensive contributions are the things that don’t happen. Here, “don’t happen” means aren’t regularly recorded and recognized as standard basketball activities. When Rudy Gobert gets a block, that’s a credit to him. However, for him to get that block, it’s possible that a perimeter player was beat off the dribble or on a backdoor cut. That’s a bad thing. From Utah’s perspective, the best-case scenario is that their perimeter defenders never get beat on the perimeter and Gobert never has to make the block. The best defense occurs when there is no penetration—a non-event.
Continuing with this argument, the best defenders may be challenged less. A smart NFL quarterback is less likely to challenge a great cornerback. Thus, it’s unfair to judge a cornerback by his knockdowns or interceptions. A cornerback could have such a great game that he’s never challenged and records no knockdowns or interceptions. The best NBA defenders might be challenged less, leaving fewer opportunities for steals or blocks.
The second major obstacle for defensive basketball analytics is the over-infatuation of basketball statistics with the individual. Basketball is a team sport. Yet, individual advanced basketball metrics (such as PER or RPM) far outnumber anything comparable for the team or lineup.
No player operates in a bubble. A big man might give up fewer points at the rim because he plays with perimeter defenders that keep their men from getting to the hoop. A perimeter defender might be able to press up on the 3-point shooters, lowering his opponents’ 3-point FG% because he can trust the help defense behind him. A defender might have to fight over fewer screens because he plays with teammates that can guard multiple positions.
Defense is a team activity. Yet, the only team or lineup defensive metric that is remotely popular is DRtg. DRtg provides an overview of team defensive performance, but doesn’t get into the specifics of how or why the team executes the way that they do. DRtg can tell us that the Warriors had a good defense last year. However, if a team was looking for ways to improve its defense, this metric doesn’t do anything more than suggest that they could try to be like Golden State. Where are the defensive metrics that tell us how Golden State defends the pick and roll on the wing? Where are the metrics that tell us when they bring double teams?
There is a huge void in the basketball analytics literature and it’s time we started to fill it.
Opponent PISS
Previously, we introduced “painfully ill-advised shots” (PIS). These are contested pull-up mid-range jump shots that are taken with at least 5 seconds on the shot clock. These are not good shots, and it’s hard to justify a team or player’s choice to ever take a PIS.
We let PIS selection or PISS be the player or team’s percentage of shots taken with at least 5 seconds on the shot clock that are PIS. So, if a player had a PISS of 10%, 1 in 10 of that player’s shots with at least 5 seconds on the shot clock were PIS.
It turns out that some defenses induce more PISS than others.
The table below presents the opponent PISS and opponent PIS FG% for all NBA teams in each of the last two seasons. Opponent PISS ranges from 13.1% for the 2015 Trail Blazers to 5.6% for the 2014 Heat.
| Season | TeamName | Opp. PISS | Opp. PIS FG% |
|---|---|---|---|
| 2015 | Portland Trailblazers | 13.1% | 40.7% |
| 2014 | San Antonio Spurs | 12.3% | 39.0% |
| 2014 | Chicago Bulls | 11.9% | 37.1% |
| 2014 | Indiana Pacers | 11.8% | 38.2% |
| 2015 | Chicago Bulls | 11.3% | 40.6% |
| 2015 | Charlotte Hornets | 10.8% | 38.5% |
| 2015 | Indiana Pacers | 10.6% | 37.8% |
| 2014 | Portland Trailblazers | 10.5% | 38.5% |
| 2015 | Memphis Grizzlies | 10.4% | 39.5% |
| 2015 | Golden State Warriors | 10.3% | 37.7% |
| 2014 | Orlando Magic | 10.2% | 39.4% |
| 2014 | Golden State Warriors | 10.1% | 32.9% |
| 2015 | Washington Wizards | 9.9% | 37.5% |
| 2015 | San Antonio Spurs | 9.7% | 37.2% |
| 2015 | Boston Celtics | 9.4% | 37.1% |
| 2015 | Utah Jazz | 9.4% | 39.4% |
| 2014 | Charlotte Bobcats | 9.4% | 44.8% |
| 2014 | Boston Celtics | 9.3% | 39.5% |
| 2014 | Memphis Grizzlies | 9.1% | 37.7% |
| 2015 | Phoenix Suns | 9.1% | 32.5% |
| 2014 | Atlanta Hawks | 9.0% | 36.5% |
| 2014 | Denver Nuggets | 8.9% | 41.3% |
| 2015 | Sacramento Kings | 8.7% | 38.3% |
| 2014 | Los Angeles Clippers | 8.7% | 36.2% |
| 2015 | Los Angeles Clippers | 8.6% | 37.4% |
| 2015 | New York Knicks | 8.6% | 36.6% |
| 2015 | Orlando Magic | 8.6% | 39.7% |
| 2015 | Denver Nuggets | 8.6% | 41.6% |
| 2015 | Detroit Pistons | 8.6% | 38.0% |
| 2014 | Sacramento Kings | 8.5% | 34.4% |
| 2014 | Los Angeles Lakers | 8.5% | 34.9% |
| 2015 | Cleveland Cavaliers | 8.4% | 39.1% |
| 2015 | Atlanta Hawks | 8.3% | 40.3% |
| 2014 | Minnesota Timberwolves | 8.3% | 39.8% |
| 2015 | Dallas Mavericks | 8.2% | 35.3% |
| 2015 | New Orleans Pelicans | 8.2% | 36.0% |
| 2014 | Toronto Raptors | 8.1% | 36.1% |
| 2015 | Los Angeles Lakers | 8.1% | 38.0% |
| 2014 | Washington Wizards | 8.1% | 36.7% |
| 2015 | Miami Heat | 8.0% | 37.6% |
| 2015 | Toronto Raptors | 7.9% | 41.5% |
| 2014 | Phoenix Suns | 7.9% | 36.1% |
| 2015 | Minnesota Timberwolves | 7.7% | 39.7% |
| 2014 | Utah Jazz | 7.7% | 36.7% |
| 2014 | Cleveland Cavaliers | 7.7% | 33.2% |
| 2015 | Milwaukee Bucks | 7.6% | 34.6% |
| 2014 | Houston Rockets | 7.5% | 37.2% |
| 2014 | Dallas Mavericks | 7.4% | 37.5% |
| 2014 | Brooklyn Nets | 7.4% | 36.2% |
| 2015 | Brooklyn Nets | 7.3% | 35.2% |
| 2014 | Detroit Pistons | 7.2% | 39.1% |
| 2015 | Houston Rockets | 7.2% | 39.8% |
| 2014 | Philadelphia 76ers | 7.1% | 37.4% |
| 2014 | New Orleans Pelicans | 7.1% | 35.1% |
| 2014 | Milwaukee Bucks | 6.7% | 39.8% |
| 2015 | Oklahoma City Thunder | 6.6% | 42.0% |
| 2015 | Philadelphia 76ers | 6.6% | 39.6% |
| 2014 | Oklahoma City Thunder | 6.2% | 37.7% |
| 2014 | New York Knicks | 6.2% | 35.2% |
| 2014 | Miami Heat | 5.6% | 36.8% |
The chart below shows that in 2015, the Portland Trail Blazers were nearly 2 percentage points better in this category than any other team. This success is nothing new for Portland. They were 4th in the NBA in 2014. Playing Portland seems to have the same effect as chugging a large coffee and sinking your hand in a bucket of warm water.
Portland’s consistent PISS-inducing defense provides some evidence that defenses can actually influence their opponents’ shot selection. Note that if opponent PISS varied greatly from year to year, it would suggest that the statistic is unpredictable and largely out of a team’s control.
The year-to-year stability exhibited by Portland was not an anomaly. The chart below plots a team’s 2014 opponent PISS to their 2015 opponent PISS. The teams that were better at forcing PISS in 2014 tended to be better at the task in 2015.
How does a team induce PISS? Is it personnel dependent?
Inducing PISS is less about the perimeter defender’s ability and more about the team’s system and the interior defenders. In 2015, the top two in opponent PISS were Steve Blake (27.8%) and Damian Lillard (27.0%). Both played for Portland. Two of the next three in opponent PISS played for Chicago (Aaron Brooks and Derrick Rose). Two other Chicago guards, Tony Snell and Kirk Hinrich, were in the top 20. (Tom Thibodeau was the NBA’s catheter.) San Antonio’s two point guards Tony Parker and Cory Joseph were 7th and 8th with nearly identical opponent PISS. (In 2014, Parker and Patty Mills were both in the top 5.)
Meanwhile, certain guards that changed teams saw significant changes in their opponent PISS. For example, Mo Williams went from a PISS of 19.2% in Portland in 2014 to 14.8% last season with Charlotte and Minnesota. Steve Blake saw a drop from 27.8% in Portland in 2014 to 16.9% last season with the Warriors and Lakers.
PISS Matters
We’ve seen that teams can design and implement schemes that influence PISS, but does PISS truly matter? Should teams try to induce PISS?
The chart below plots opponent PISS to DRtg for each team in each of the last 2 seasons. The two statistics are correlated. Higher opponent PISS tends to predict a lower DRtg. Among the top 15 team seasons in Opponent PISS, 13 were in the top 18 in DRtg.
Conclusions
Our analysis leads to several conclusions.
- NBA defenses can have a significant impact on opponent shot selection.
- The impact on opponent PISS seems to be less a product of individual perimeter defensive skills and more a result of team scheme and interior defense.
- Inducing opponent PISS correlates with defensive efficiency.
The analytics suggest that teams can and should start inducing PISS.