After allowing an average of 119 points in the first two games of their first-round series against the Miami Heat, both losses, the Charlotte Hornets turned things around to win Game 3 96-80 at home on April 23.
There were plenty of explanations for how much better Charlotte's defense performed, including a change to the team's starting lineup, better energy and the effect of home-court advantage. But Hornets coach Steve Clifford, who had publicly downplayed the importance of adjustments after Game 2, offered a simpler reason.
"Coach [Pat] Riley is the person that gets credit in this league when anybody says, 'It's a make or miss league,'" Clifford told reporters in the postgame news conference. "The nights when you make, you look good. The nights when you miss, it looks bad."
Clifford's comments might sound like a reductive coaching cliché to outside observers. But they're not.
Advances in statistical analysis have allowed us to better separate the quality of a shot from the outcome of a shot. And these advanced stats support Riley's idea that the result of a game has one dominant factor: whether shots go in or not.
"Make or Miss League"
The phrase "make or miss league" surely isn't new to anyone who has regularly listened to ABC and ESPN telecasts featuring former NBA coach Jeff Van Gundy. While the phrase apparently originated with Riley, it's Van Gundy, who worked under Riley as an assistant with the New York Knicks and later had Clifford as his assistant coach, who has popularized the notion, first as a coach and now as an analyst.
Among others, Los Angeles Clippers coach Doc Rivers is fond of the saying as well.
Van Gundy said he has "been more convinced" that the NBA is a make or miss league the longer he's around basketball.
"It means to me that when you have great shooting, your offensive system is going to look better," he said, "and when you have great passing and great shooting you can play at historically good levels offensively."
There's another layer, too, that speaks to the importance of making the shot above and beyond its quality.
"You can have a great offensive possession and miss the shot, and it gets marked down as an inefficient possession," Van Gundy said, "and you can have a horrible offensive possession bailed out by great shot making, and it goes down as great offense."
But how do we tell the good shots from the bad shots on a systematic level? That's where the NBA's new player tracking data comes in.
Separating Shot Quality from Shot Making
Using the tracking data, we can separate the outcome of a shot into two elements: quantified Shot Probability (qSP), which estimates the expected value of each shot based on the identity and location of the shooter and location of defenders at the time it is taken, and the difference between this and the actual effective field goal percentage, or quantified Shot Making (qSM).
To figure out the relative value of shot quality and shot making, we can look at how well they explain the outcome of games. First, though, we should start by looking at the relative importance of shooting in determining who wins in the NBA using effective field goal percentage, one of the four factors of basketball identified by pioneering analyst Dean Oliver.
During the 2015-16 regular season, the difference in teams' shooting explained almost half of the difference between them on the scoreboard. Making up the rest were their turnover rate and offensive rebounding, both a little more than 20 percent, and their ability to get to the free throw line, just 7 percent.
When we replace eFG with qSP and qSM, it becomes clear that, within a single game, quantified Shot Making is the most important factor that determines winning and losing.
The difference in teams' shot making explains more than twice as much of their difference on the scoreboard compared to their qSP.
Another way to consider the issue is to look at the win-loss record of teams based on their shot quality and shot making. When teams had the edge in both categories, they won more than 90 percent of their games during the 2015-16 regular season. But when they were split, teams with the better qSM beat teams with the better qSP 56.8 percent of the time (355-270).
Over 82 games, there is time for the randomness in shot outcomes to even out, so shot quality becomes paramount. But when it comes to making shots in a single game, it's better to be lucky than good.
Shot making doesn't carry over game-to-game
How do we know that teams with higher than expected shooting percentages are simply having a good night rather than showing skill that isn't captured in qSP? Looking at how they perform game to game throughout the playoffs helps make that point.
Even though the two offenses and defenses are the same, shot making tends to vary widely from one game to the next. And if you're looking to predict how much better than expected a team will shoot in the next game, you're better off looking at their qSP in the previous game than their previous qSM. That suggests that the difference is about more than just those adjustments Clifford diminished as overrated.
Perhaps because of the factors qSP can't entirely capture, including size mismatches and whether a shot is actually contested or the nearest defender is simply a bystander, the process of taking good shots tends to lead to better shot making than hot shooting in the previous game.
As Kyle Wagner explored last week at FiveThirtyEight, the Charlotte-Miami series was a good example of how shot making tends to even out over the course of a series. But it's not the only case from this year's postseason, and maybe not even the best one.
Consider the Oklahoma City Thunder's playoff run. After the Thunder lost Game 2 to the Dallas Mavericks in the first round, scoring just 84 points in the process, there was widespread concern about Oklahoma City's uncreative offense and reliance on stars Kevin Durant and Russell Westbrook.
According to qSP, the Thunder's shots in Game 2 could have been expected to yield 51.5 percent effective shooting, similar to the team's 52.4 percent average in the regular season. Instead, Oklahoma City posted an eFG of just 37.9 percent, in large part because Durant shot 7-of-33 from the field. In Game 3, the Thunder bounced back with an eFG of 67.3 percent, the highest of any team in the playoffs. Yet Oklahoma City's qSP was actually slightly worse in Game 3, at 50.2 percent. The gap between the best shooting performance in the postseason and one of the worst apparently was nothing more than shot making.
Something similar happened between Games 1 and 2 of the Thunder's series with the San Antonio Spurs, this time on the other end of the court. San Antonio's shot making in Game 1 was second only to Oklahoma City's in Game 3 against Dallas among all playoff teams, and the Thunder were criticized for their lack of effort defensively. In Game 2, by qSP the Spurs actually got better shots but were unable to hit them as frequently as expected this time, and Oklahoma City won despite getting worse shots than Game 1.
The importance of quantified Shot Making, and the way it doesn't carry over from game to game, might be surprising to some observers, but not to experienced coaches such as Van Gundy.
"Unfortunately, sometimes we don't want the simple explanation," he said. "I think we want the game to be more complicated than sometimes our eyes tell us -- and maybe even the numbers tell us -- it is. You try to get the best possible players, you try to put them in a sound system at both ends. And then when the ball goes up, when it's in the air is when you judge the shot. You don't wait to see if the ball goes in."