Nylon Calculus: The impact of passing on offense and winning

Photo by Meg Oliphant/Getty Images
Photo by Meg Oliphant/Getty Images /
facebooktwitterreddit

Do additional passes lead to better outcomes on offense? And, what does passing tracking data indicate about the current methods for capturing advanced metrics on NBA play?

The reverence for teams that share the ball is prominent at all levels of basketball, and the NBA is no exception. As a fan, watching an altruistic team dance as the defense scrambles to react is aesthetically pleasing, and brings to mind the axiom “the way the game was meant to be played.”

It’s quickly evident that not all passes are created equal when contextualizing a pass as a function of creating a better shot for a teammate. In certain situations, a pass represents hope, a building block, of a potentially successful offensive possession. Teammates share the ball to try to spark some advantage, either through their collective prowess or the exploitation of a mistake made by the opponent.

On Saturday, the Wizards opened the game with such a possession. The play started with little direction, the ball traveling east-west rather than north-south, and the Blazers did well to contain multiple Washington attacks. The additional passes from Bradley Beal, Mo Wagner, and Russell Westbrook were all hopes of an opening materializing, which did come to fruition when Westbrook attacked the paint and found Wagner for an open-corner 3 as the shot clock was expiring.

Other possessions begin more auspiciously, comprised of passes with more rhythm and purpose:

So at least anecdotally, we can conclude that more passes do not equal improved offense all else equal. But what do the data say?

The sample I used for this analysis is comprised of seven full seasons of team tracking and “traditional” data for each of the 30 teams from the 2013-14 season through the 2019-20 shortened season. Using the publically available tracking data provided by Second Spectrum on NBA Stats, I gathered aggregate passing totals per season per team and the corresponding offensive efficiency metrics. From the plot of the 210 data points, we see a rather surprising (or unsurprising depending on your initial hypothesis) trend.

From this sample, offensive efficiency has a negative relationship with aggregate passing data, which provides us with an immediate insight. We can reasonably conclude that aggregate passes don’t accurately capture the teamwork that we would commonly associate with possessions that have a higher pass volume. That is, we don’t effortlessly recall the high-pass-volume possessions that don’t conclude with points scored.

The data doesn’t discriminate, so an unsuccessful possession in which a team passes multiple times creating no true advantage carries the same weight as a possession in which each pass works to create on top of the previous one leading to an open shot.

We can also hypothesize on a couple of second-order effects that are slightly more discrete and rather esoteric. The first is that strong offensive teams, such as the new-look Brooklyn Nets with James Harden, don’t require multiple passes to find a high-percentage shot attempt (and again, these opportunities are equally weighted). Simply allowing Kyrie Irving, Kevin Durant, or Harden to create in isolation suffices.

A more existential argument is that teams that take longer to find acceptable shots ultimately see fewer possessions in a 48-minute game, and theoretically, they’d have to be even more efficient on such possessions to boost their overall offensive efficiency. While the likelihood that a team takes characteristically more time to find a shot and passes the ball is inherent to the offensive scheme itself, empirical data (i.e. watching games) would suggest that most team’s offensive styles are more varied.

The exact relationship developed by my model estimates that each additional pass made is detrimental to offensive efficiency by 0.04 or 4 points per 100 possessions. If accurate, that is enough of an effect to impact the outcome of close games. However, given the imperfect measurement in our data, it seems more probable that the relationship between passing and offensive efficiency is nebulous at best.

For the sake of discussion, let’s momentarily assume that we have debunked the myth that more passing leads to better offense, and next we want to explore the relationship between passing and turnovers. In this manner, we can view a pass as a risk-return proposition, since every pass has the simultaneous possibility of creating a better scoring opportunity or ending a possession. Using the same sample, we can plot the correlation between aggregate passes and each team’s turnover percentage to account for possession differences.

And here we find a result we’re expecting, with total passes positively correlated with a team’s turnover percentage.

There are two general categories of turnovers, dead-ball (e.g. shot-clock violation, offensive three-second violation, offensive foul, pass or dribble out of bounds, traveling, and double-dribbling) and live-ball (lost live dribble or a bad pass to an opponent) with the latter being more frequent.

Turnovers are obviously harmful to a team’s offense in losing possessions, but they are rarely considered for their detriment to a team’s defense.

And this is where we see the strongest correlation; a higher turnover percentage undoubtedly leads to more opponent points off turnovers.

For thoroughness, we can measure the total impact of each pass considering the holistic impact on both offense and defense. Using the same model to measure the relationship between passes made and turnover percentage and turnover percentage and opponent points off turnovers, we can make an estimation of the overall impact of each pass on opponent scoring. In addition to the original estimate of 4 points per 100 possessions lost on offense, the relationship between passing and opponent points off turnovers builds an additional 0.8 points per 100 possessions in detrimental impact. Were we to completely trust the available data, we could conclude that passing is net-negative to team performance at around 5 points per 100 possesions, a statistically significant result.

Is there a better offensive philosophy than “more passing”?

If passes are shown to have a neutral to possibly slightly negative impact on offensive efficiency but lead to more turnovers and more points for the opponent, should coaches advise their teams to pass the ball less?

Personally, I find this to be a slippery slope argument from which the recommendation can lead to more harmful effects than inertia. To truly be able to make a declaration on passing volumes within a given NBA offense, we would need to compare the model’s output with that of an alternative universe league that actively tries to minimize passes above all other offensive objectives except scoring.

We can hypothesize what this would look like, probably Trae Young or Luka Dončić on steroids — a sure-handed, high-volume scorer who doesn’t need to share the ball to create offense. And even with a seminal talent, does a team’s shot quality and ability to score the ball go up enough to offset the avoided lifeless possessions, turnovers, and easy opportunities for the opponent?

In my opinion, the more interesting possibility is for an offensive experiment with stretches of one-on-one play in between “standard” playbook possessions to find out which strategies work best with a given lineup. Rather than optimizing for one course over another, a team could design its roster to take advantage of both simultaneously. It would be easy to measure the results of these sandbox tests internally, and potentially capture micro advantages that may not otherwise have come to light.

One such example: would optimizing an offense for players with higher career assist-to-turnover ratios naturally increase offensive efficiency, if not alone through the elimination of harmful turnovers? And wouldn’t those in charge of roster decisions at least make an attempt to actively avoid some of the players with the highest career turnover percentage, since there are many current players atop the list?

There is certainly more nuance to these findings and an argument for both sides of the discussion. But if a team is struggling to create offense, why not investigate?

light. THE WHITEBOARD. Subscribe to our NBA daily email newsletter