Comparing Offensive Outcomes and More Hack-A-Context
By Ian Levy
Mandatory Credit: Thomas B. Shea-USA TODAY Sports
Last night, in their 124-103 win, the Houston Rockets sent DeAndre Jordan to the free throw line another 16 times. That gives him 40 free throw attempts in the last two games and 144 for the playoffs—more than any other single player and more than every team eliminated in the first round, except the San Antonio Spurs. Not all of those have been on intentional fouls of the Hack-A-____ variety, but the strategy is a big part of his inflated total.
Everyone has been writing about Hack-A-Jordan, including here at Nylon Calculus, with a focus on whether or not it works and to what degree. At the risk of exhausting our readers, I want to go back to the topic but with the more modest goal of contextualizing some of the numbers.
Setting aside the issue of whether the strategy actually creates an advantage and how big or small that advantage is, I think it’s worth illuminating exactly what sort of outcomes we’re talking about.
Jordan has made 42.4 percent of his free throws in the playoffs. We know that’s a horrible mark because we know what good-free throw percentages look like. We see free-throw percentages bandied about all the time and we, for the most part, understand the scale. Percentages in the 70s and 80s are objectively recognized as the marks of good or reliable shooters. The problem is, thinking about the Hack-a-Jordan strategy in the context of Jordan’s free-throw percentage as compared to the rest of the league is misleading. When an opponent intentionally fouls Jordan they are not forcing the Clippers into a binary switch between having him shoot free throws and someone more reliable shooting free throws. The opponent is selecting that offensive outcome—Jordan shoots two free throws—from the entire array of possible offensive outcomes.
To really create a proper statistical context around the strategy, it’s worth comparing the outcome of Jordan at the line for two free throws to a wider array of offensive outcomes than just someone else’s free-throw percentage.
The most basic way to split offensive outcomes is into shot attempts, free throw attempts and turnovers—those are the only ways a possession can end. Obviously, turnovers have an expected value of zero points per possession. However, we can compare the expected value of Jordan attempting two free throws—0.84 in these playoffs (42.4 percent times two free throw attempts)—to shots and free throw attempts. To add some granularity I’ve split shots out by location and added the average, minimum and maximum player expected value from each location[1. Minimum and maximum are from the pool of players with at least 100 attempts this season in the restricted area, mid-range, above the break threes, or from the free throw line. 50 attempts was the cut for corner threes and in the paint (non-restricted area.].
At 0.84, Jordan at the line for two free-throw attempts is a slightly more efficient outcome than the league average from either the In the Paint (Non-RA) or Mid-Range zones. It’s also roughly comparable to a three-point attempt or a shot in the restricted area from a bad scorer in either location.
Here we can also see how objectively valuable free-throw attempts usually are. The peak expected value of a pair of free throws is higher than a scoring opportunity from any other location. The average expected value on a pair of free throws is higher than the peak expected value of shots from three different locations—In the Paint (Non-RA), Mid-Range and Above the Break 3s.
We can also compare expected values of offensive outcomes by play type, using the Synergy Sports statistics now available at NBA.com. The graph below is set up the same way and I’ve limited things to the five most common play types. As always, the caveat with these Synergy classifications is that they represent the outcome only—points per possession on isolations are only those actually scored by the player. They do not include points scored off assists where an isolation collapsed the defense and the ball was kicked out to a shooter.
The expected value of these play types is heavily influenced by the shot locations they inherently produce. The vast majority of transition possessions end with a three-pointer or a shot at the rim, thus they typically produce efficient results. Spot-ups are almost always catch-and-shoot jumpers[2. Although Synergy does appear to classify a player catching and driving against a close-out as a Spot-Up possession] with the majority being three-pointers, hence their high numbers. We can see that Jordan taking two free throws is roughly comparable to an average possession used on an isolation, post-up or finished by the ball-handler in a pick-and-roll.
Also, I’m keeping this analysis fairly simple and not incorporating the likelihood of offensive rebounds on any of these play types or the shot locations above. Those probabilities could dramatically change the expected values of these different outcomes.
Obviously the three most efficient possession types—transition, spot-ups and possessions finished by the screener in a pick-and-roll—are influenced by the defense in terms of how often they can be run. A team could essentially get an isolation or post-up possession any time it wants. Those more efficient options usually require some sort of defensive breakdown which limits their frequency. This means that every team is forced to use a certain portion of their offense on isolations, post-ups and pick-and-roll screeners.
This is all to say that when a defense intentionally fouls Jordan and makes the choice of offensive outcome, they aren’t necessarily creating a catastrophically untenable offensive situation. Every team in the league is forced to use a huge number of possessions every game in scenarios with a similar expected value. If a team was able to send Jordan to the line every possession for an entire game, the strategy would almost certainly work. But there aren’t enough fouls on a roster and so defenses are simply working a few more of these possessions into the mix of spot-ups, isolations, transition, etc. possessions that the offense is able to achieve otherwise.
This is all to say that DeAndre Jordan at the free-throw line is not an ideal offensive outcome for the Clippers. But it’s not nearly as bad as that 42.4 free-throw percentage may imply.