After a playoff career high of 26 points, Josh Hart had this to say after the New York Knicks Game 2 victory over the Cleveland Cavaliers:

“At a certain point, they’re a lamp post to a dr*nk person, you can lean on them but they won’t get you home.”

A few days later, following the Cavaliers’ Game 3 loss to the Knicks, Kenny Atkinson cited poor shooting luck as part of the reason that Cleveland was in a 0-3 hole: “I think analytically, I think we’ve won the — I said three out of three [games in the series], we’re two out of three in the expected wins,” Atkinson told reporters Sunday. “I don’t know if you guys follow that — the expected score. We’ve won two out of three.”

As you can tell, analytics and their place in basketball have been a major talking point during the Eastern Conference Finals. So much so that people are getting social media clout for poking fun at Atkinson’s remarks:

So, who is right? Hart or Atkinson? And is it time we start to pump the brakes on data-driven decision-making?

What Atkinsons Mean By “Analytically”

Before we dig into who is right/wrong, I think it is important to put some color on Atkinson’s remarks. Since the game of basketball is, in its essence, dictated by a round ball colliding with a bunch of living/non-living beings, there is a level of randomness at play, especially when you are dealing with small samples.

This is true even in a best-of-seven playoff series. Contrary to public belief, the best team does not always advance. In fact, in his 2016 book, “Thinking Basketball,” Ben Taylor concluded that it would take 22 games to obtain a sufficiently large sample to guarantee that the better team would always emerge victorious.

There is nothing wrong with the best team not always winning. Part of why humans are drawn to sports is the idea of an underdog triumphing over a superior adversary. Besides, who in their right mind would sit through a 22-game series?

However, we can’t disregard the prospect that the outcome of a series doesn’t necessarily tell the whole story. The Cavaliers surely have a more nuanced internal version of this statistic, but what Atkinson is likely referring to with his comments is that the Cavaliers’ estimated shot quality was better than the Knicks in two of the first three games of the series.

Location effective field goal percentage, via Cleaning the Glass, tells us what a team’s effective field goal percentage would be if they shot the league average from all those spots on the floor. The Cavaliers fared higher in Games 1 and 3 of the series. So, in theory, you would have expected them to win those games, particularly when there wasn’t a significant difference in turnovers/offensive rebounds in those matchups.

The Limitation Of Analytics

As an aside, along with my writing, I also serve as an assistant high school varsity basketball coach. And before the Knicks’ Game 4 victory, I was texting with my head coach about Atkinson’s remarks. As a 48-year-old man who has coached for over two decades and has seen the evolution of the game before his very eyes (and, for the most part, embraced those changes), his perspective on the matter is quite interesting. So, I asked him, what would he do in Atkinson’s shoes? What would he change if he knew his team was (again) theoretically acquiring and conceding the types of shots they wanted?

“Are these ‘wide open’ looks happening more in transition or the halfcourt? Are they coming off of offensive rebounds? Who is taking these wide open/contested looks for both teams? So many questions to answer before I give you an answer. Heck, Donovan Mitchell could be hurt and not as explosive as he normally is. So, that means less help is needed. He still creates what Atkinson will say is a wide open look, but maybe the shooter is actually rushed because the guy recovering is a little quicker. This is what I think Atkinson is missing. He did not explain in detail what is actually happening. Instead, he is just like we are getting good shots and missing, and they are not.”

Basically, what my head coach is calling for is the part that gives data the color we need to properly use it: context. To know what the data is actually trying to tell us, you need to take into account the circumstances surrounding it.

Let’s take a break from the talk of shot quality and look at on/off point differential, which tells us how much better/worse a team is with a certain player on the floor compared to when they are on the bench. Here are the top five players in that stat from the regular season (minimum 1,000 minutes played):

The first four names (Nikola Jokic, Victor Wembanyama, Giannis Antetokounmpo, and Kawhi Leonard) should be of no surprise to anyone. Depending on who you ask, those may be four of the five best players in the league when they are completely healthy. But what about the player sitting at number five? Is Jalen Smith really having a superstar-level of impact?

I’ll keep it simple: no. Jalen Smith played on the Chicago Bulls, a team that was sorely lacking in center play (especially after they traded away Nikola Vucevic). So, Smith’s standing in this measure isn’t so much about Smith’s unique excellence as it is about the team’s roster construction (no shade to Smith, though).

Back to Hart’s comments, not to speak for the dude, but I think this is what he is getting at. The Knicks’ offense has reached new heights in part because of how they have embraced a more modern shot diet (they were 12th in 3-point attempt rate after being 28th in 2024-25). However, NBA offense is far more complicated than simply being able to improve one’s offensive efficiency merely by taking more threes. The Knicks have spent all season improving their offensive flow in a way that is difficult to explain by merely analyzing shot quality.

To put a bow on all of this, neither Hart nor Atkinson is inherently right or wrong. In many ways, this discussion is no different than the use of analytics in sports. To properly use it, you need to look at all the pieces of the puzzle.