Let’s set the scene.
There are 53 seconds left in the 3rd quarter of a close game between the Milwaukee Bucks and Los Angeles Clippers. Lamar Odom secures a defensive rebound and the ball works up-court, eventually settling in the hands of All-Universe Point Guard (Point God?) Chris Paul. Paul pauses, takes a quick look at the clock, and then sets up a screen-and-roll with Odom. Paul comes over the screen, sees Samuel Dalembert stepping out to obstruct his path to the basket, but to no avail – Paul executes a nifty inside-out dribble, goes around Dalembert, and scoops a layup high off the glass and in.
On the ensuing Milwaukee possession, the Clippers play strong defense, and the Bucks are forced into an awkward shot with the shot clock winding down. The airball again finds Odom, and Odom gives the ball to Paul, who crosses half-court with 10 seconds on the clock. Again, he pauses, looks at the clock, and then sets up another screen-and-roll with Odom. This time, Dalembert hedges hard on the screen, but he bumps Paul as he goes by, resulting in two free throws as the Bucks were over the foul limit. Naturally, Paul makes both free throws.
What was a three-point game with 53 seconds left on the clock is now a seven-point game at the end of the third quarter. This sequence, in basketball vernacular, is called a two-for-one. The Clippers were able to get two possessions at the end of the quarter to the Bucks’ one. Chris Paul is a wizard in these sequences, expertly managing the clock to ensure his team is the one that ends up with the extra possession.
The two-for-one has long been a point of contention in the basketball community. The statistical side of the argument is pretty easy to understand – two bad shots are better than one good shot. Even the most precise offensive team, like Oklahoma City or San Antonio or Miami can’t realistically expect to have a rate of success higher than, say, 55-60%. On the other hand, even a poor offensive team taking a poor shot can reasonably expect to score maybe 35% of the time. Simple math will tell us that two plays with 35% success rates each, in the long run, will yield more points than one play with a 60% success rate.
At last weekend’s MIT Sloan Sports Analytics Conference, former Orlando and Miami coach Stan Van Gundy took issue with the idea that going two-for-one is an inherent positive strategy. Normally when people argue against strategies that, at least on the surface, appear to be statistically-supported as beyond reproach, they end up sounding like curmudgeonly old men screaming at the crazy kids to get off their lawn. Nonetheless, Van Gundy presented a logically sound, nuanced argument that said maybe two-for-one basketball isn’t a slam dunk (pun absolutely intended).
Van Gundy said that NBA basketball, at its bare bones, is built on trust. Trust that when you help and rotate, someone else is going to help and rotate to your spot. Trust that when a player curls around a screen, he’ll be ready for the pass that’s already in the air. Trust that if you make the right plays, your coach will reward you with more playing time. And, most importantly, trust that everybody is on the same page and buying into the same system.
To Van Gundy, oftentimes going for that two-for-one proposition involves taking an ill-advised shot early in the shot clock. To him, that flies directly in conflict with his philosophy that low-efficiency shots early in the shot clock are inherently bad shots. If he allows his players to take one of those shots in a two-for-one situation, the system that everyone is (hopefully) buying into is being compromised, if even only a little. Players might start to think, “If I can get away with taking a bad shot there, what else can I get away with?” Think of a basketball team like a Jenga tower, with every “exception” to the core message being a piece taken from the bottom of the tower and placed on top. Yes, the analytics are backing up those “exceptions,” and as a result, yes, the tower is getting higher. But it’s also becoming much less stable.
The question that I would have wanted to ask Van Gundy would be in regards to the makeup of the team, and which players are actually taking those two-for-one shots. For example, when Van Gundy was coaching the Magic, his superstar was Dwight Howard and everyone else fit in around him. However, Howard is a big man, so at the end of a quarter, if a two-for-one opportunity presented itself, it would be players like Jameer Nelson or Hedo Turkoglu creating those shots. My question would be if those situations, where it’s a non-star player who are taking what would normally be considered an ill-advised shot, is any different than a situation for the Clippers, or the Heat, or the Thunder, where those two-for-one shots are being taken by the superstar player, who in most cases is given more latitude to begin with.
In other words, does Jameer Nelson taking a bad shot in a two-for-one erode at a coach’s message and system more than Chris Paul taking that shot? Or similarly, how does the philosophy change when the team in question is a high-character, veteran team with clearly defined roles (like Miami) compared to a young team with questionable direction and discipline (like Sacramento)?
Or going further, what if the two-for-one shots aren’t even bad shots to begin with? The brilliance of Chris Paul isn’t just that he manages the clock to make two-for-one opportunities available, it’s that he actually creates good looks at the basket out of these opportunities (like the driving layup against Milwaukee Wednesday night). Would Van Gundy change his philosophy on two-for-ones if he knew that the players on the court were able to create good shots, even quickly in the shot clock?
The takeaway here lines up with one of the broader themes of last weekend’s conference. The analytics themselves are no longer the most important aspect of this field. Communicating these analytics in easy-to-understand terms, and turning the data into actionable recommendations in a real-world environment is now the real challenge. In the case of the two-for-one, the math clearly shows us that two bad shots are better than one good shot. But when we translate that data from a spreadsheet and into a scouting report, we need to realize that we’re not dealing with X-Y coordinates and abstract statistical profiles. We’re dealing with people, and people are imperfect by nature.
In a vacuum, teams should always go for the two-for-one. But basketball games aren’t played in vacuums. They’re played by the little men that live inside our television sets.