The Empty Report: When the NBA Analytics Room Refuses to Lie

Three in the morning in Chicago. I sit alone in the small studio with a forty-two-page document a friend inside an NBA analytics department sent over. Page one is a heat map. Page fifteen is an RAPM grid. Page twenty-three is a ten-thousand-run simulation of a seven-game series. The last page is blank, and on it one sentence is typed exactly fourteen times, with a date and the department head's signature: "Insufficient data to conclude."

I read that page longer than the other forty-one combined.

Forty years behind a microphone and in press rooms, I have been handed hundreds of files like this. Nearly all of them end with a confident verdict: trade this player, go small, extend the contract now. That blank page was the first time I saw a machine refuse to lie. And the first time I realised professional basketball is carrying an illness nobody wants to name.

The illness is not in the data. It is in people using data to answer questions data was never designed to answer.

To understand how an analytics department could file a blank report, you have to go back twenty years.

In 2026, Mike D'Antoni's Phoenix Suns ran and shot threes until the rest of the league called it a gimmick. In 2026, Daryl Morey was named general manager of the Houston Rockets and turned the analytics room into the brain of the entire organisation. In 2026, Sam Hinkie carried that philosophy to Philadelphia and gave it a name that sounded like a religious slogan: The Process. By 2026, Golden State had won a title with a roster every spacing model had dreamed of, and the data war ended with the guys holding the laptops on the winning side.

Twenty years later, every team has a data room. Every draft has a model. Every contract has a projected value curve. The problem is that when everyone owns the same weapon, the weapon stops deciding anything. The person holding it does — and that person is showing up at the arena less and less.

I still remember June 2026 in Kazan. Germany lost 0-2 to South Korea and were eliminated in the group stage, and the world was stunned. I walked into a local beer hall, bought a few rounds for Korean reporters, and wrote a long piece about how Germany completed twelve percent fewer vertical wide passes than they had four years earlier. That was data. But what let me write the line "Germany had probably already lost before the first ball was kicked — people simply had not been sharp enough to see it" was not the spreadsheet. It was the final training session I watched: three players standing ten metres apart, none of them looking at each other.

Data confirms what the eye already suspects. Data does not generate suspicion.

And this is where the story gets worrying.

NBA analytics departments are being asked to answer locker-room questions with classroom spreadsheets.

On 28 May 2026, the Houston Rockets walked into Game 7 of the Western Conference Finals. Chris Paul sat out with a hamstring injury. Houston led in the first half. Then the streak began: twenty-seven consecutive missed three-pointers, ending in a 92-101 loss to Golden State. A record no team wants to hold.

People called it bad luck. It was not. It was a decision.

Houston's philosophy was clear: a missed three is still a three, and over the long run the percentages regress to the mean. That logic holds mathematically across ten thousand simulations. But a game does not have ten thousand second halves. A game has forty-eight minutes, twelve of which feature a team collapsing psychologically, a few thousand fans holding their breath, and one player standing in the corner twice turning his head toward the bench looking for a wave of the hand.

That wave never came.

In my notebook I wrote this line: the model has no variable for "an emotional corridor". It does not model people. It models shots, and the shots in that game missed twenty-seven times in a row because the men behind them had stopped believing.

That was the first example, and the mildest. From there the illness gets worse.

In the summer of 2026, Kawhi Leonard and the San Antonio Spurs entered a divorce that lasted a full year. Leonard had a quad injury. The medical staff and the recovery model said he could return. The player and his camp said his body disagreed. He played nine games that entire season. A players-only meeting took place in March 2026. Words were said that hurt. In July, Leonard was shipped to Toronto, and in his first season there he won a championship.

Reading that saga again, I see an argument over who gets to interpret a body. For decades, a player's body belonged to the player and the team doctor. Then the computer walked in, and the body became a stream of input data: load, accelerations, cumulative impact, reinjury probability. When the model says "can play", the question in the room is no longer "how do you feel" but "why aren't you playing".

I still remember the first time I watched a player asked to sign a consent form clearing him to play based on a recovery model. He held the pen, looked at the paper, then looked at me as if to ask: since when did the one thing I am certain of — my own pain — become something that needs scientific verification?

Nobody has answered that. But it echoes through every locker room in this league.

Then there is the draft. This is where the illness peaks.

Nikola Jokić was taken with the forty-first pick in 2026. A Serbian with slow feet, a soft frame, no clear position in any model of that era. The models were not wrong — they simply had nowhere to put him. The man who saw Jokić was an international scout who sat night after night in Poland and Serbia watching a big kid pass like a guard.

Giannis Antetokounmpo went fifteenth in 2026. Same problem: a two-metre-eleven player who handled like a point guard, with no comparable in the database. And Stephen Curry — before he became the sport's new definition — signed a four-year, forty-four-million-dollar extension in October 2026, while every spreadsheet tagged him "high ankle risk". What followed was four championships, two MVPs, and the most underpriced contract in the history of the sport.

The biggest mistakes in the analytics world are not wrong predictions — they are the things with no column to predict.

A sixty-million-dollar player will not necessarily make more difference than a shy kid at the academy who knows how to watch.

On the other side, there is a case that makes me bow to the very trade I doubt. In the 2026-23 season, the whole league called Denver a sleeping giant on top of the West. They drifted through the regular season looking slow, with a defence under suspicion, and people talked about them as a team that would exit early. But late-season pace data said something else: with all five starters on the floor, their half-court offence was the best in the league. They entered the playoffs and went 16-4. That sleeping giant had simply been walking slowly in order to arrive on time.

This brings me back to the blank page in that three-a.m. file.

I asked my friend one question: "If you did write a conclusion, what would it say?" He answered: "I'd say they should trade the player. But I have no evidence. And if I'm wrong, someone loses their job." That answer kept me silent for a long time.

Because this is the submerged part of the iceberg. In an organisation where every decision must be justified by numbers, an analyst is trapped between two deaths: say "I don't know" and be treated as useless, or invent a conclusion and get promoted. That pressure pushes the whole industry into a strange state where the most honest reports are the ones rated lowest.

In the boardroom, the dynamic runs the other way. Once data walks in, it stops being a tool — it becomes a hostage. People quote it to shut down debate, to end a meeting before anyone can tell a story. "The model says so." Those four words kill more ideas than any ego in this sport.

Think of it this way. Ask a model whether Player A should play on Tuesday night, and it answers from a percentage computed on games that resemble this Tuesday in almost no way: different opponent, different arena, different injury, different contract, different mood. The model is not wrong. It is simply answering a different question from the one people believe they asked.

The same illness shows up in the personnel market wearing a different jacket. In European football, people audit every cent of a transfer fee while free-agent deals glide through unexamined. In the NBA, the equivalent is the buyout market: a team waits for a star to be bought out by his old club, signs him for the minimum, and no balance sheet reflects the true value of that transaction. What is monitored is scrutinised to the last decimal. What is not monitored walks in through the back door and nobody writes it down.

And then the nightmare arrived.

In the 2026-24 season, the NBA was forced to introduce a player participation policy, with escalating fines for teams resting healthy stars in nationally televised games. Read that again: a league had to write a rule to stop teams from believing their own models. Analytics departments had calculated that resting a star one night in January raises the probability of winning in May. Correct on probability. But there is one thing no spreadsheet can price: a twelve-year-old taken to his first game by his father, watching his hero sit on the bench in a warm-up jacket.

That is why I call this an illness rather than a mistake. A mistake can be fixed with better data. An illness needs a cultural transfusion.

I do not want to paint a picture where everything is the computer's fault. I was in Kazan in 2026, and I know it was the vertical-passing numbers that saved my piece from becoming a lament. I know it was expected goals and dribble speed that gave me the grounds to shout on air in a Chicago studio in 2026 that Mohamed Salah would break the Premier League scoring record, when he had eleven goals in eighteen matches and the forums were laughing at me. He finished with thirty-two.

Which is where I have to turn around and interrogate myself.

Maybe that blank page is not a sign of decay. Maybe it is the most honest document in that building, and my praising it is just a coded way of mourning my own trade. Maybe I am the nostalgic, the man who still believes the human eye outweighs the model, while twenty years of history prove the opposite: the teams that applied data most rigorously — Houston, Golden State, Boston, Denver — have won the most.

I could be wrong. And if I am, I am wrong here: I am conflating two different things. The first is data used to understand. The second is data used to reassure.

Understanding is looking at the twelve-percent drop in Germany's vertical wide passes and saying the system has lost its connections. Reassurance is looking at the same numbers and saying the team still has sixty percent possession, so everything is fine. Same data, two different heads. I am against the second, not the first.

There is one more thing I owe you. The night I said Salah would break the record, I shouted first and went looking for statistics to defend myself afterwards. The story I have told for years has been sanded down into a small legend. The instinct came first, and I was lucky it was right. In basketball we only remember the people who guessed correctly. Hundreds of models missed Jokić, and we only mention the one scout who saw him. That is a data trap too, and a more dangerous one, because it wears the clothes of intelligence.

There is one more blind spot, and it belongs to me. When I said Germany would collapse, I was right about the structure and wrong about the timing. I did not predict they would lose to South Korea in the final group game; I only predicted that they would fall somehow. A prophecy correct in direction but wrong in clock is not a perfect prophecy. It is a lucky one. Being honest with yourself about that is far harder than shouting something provocative on air.

So what should be done with the blank page?

I do not think the answer is firing analysts, nor trusting them absolutely. The answer is giving them the right to say the three hardest words in a sports meeting: "I don't know." An industry only runs healthily if blank reports are treated as assets rather than admissions of failure.

My prediction, and it is verifiable: within the next three seasons, at least three NBA teams will appoint a former player to their analytics department with honorary veto power over rotation recommendations — someone who can stand up between the model and the floor. When that happens, remember the blank page in Chicago.

Because in basketball as in football, every fallen giant is a slap in the face for those who collect names instead of people. And what we have been chasing all these years was never a number. For three years we chased a ball that seemed to belong to no one, and it turned out we were chasing the silence in the middle of the crowd.