27 Consecutive Missed Threes: When Data Exposes the Limits of the Three-Point Revolution

On May 28, 2026, at the Toyota Center in Houston, the Rockets led the Warriors 54-43 midway through the third quarter of Game 7 of the Western Conference Finals. A whole city believed that tonight they would break a twenty-three-year curse. Then the tragedy began, quietly, like a corrupted log file: 27 consecutive missed three-pointers. Not two, not five. Twenty-seven. When the final buzzer sounded, the Rockets lost 92-101, and the entire basketball world gained one more story to tell about the price of living by the three.

I sat in front of the screen that night, not to cry with the crowd, but to re-read Houston's three-point shooting log for the entire season. What I found was not an accident. It was the inevitable consequence of a philosophy this very team had chosen, and of a belief that if the data says threes are more efficient than twos, then you simply shoot them. The problem lies in this: the data is right, but it does not say everything.

Context: A revolution that began with a painted line

To understand why 27 consecutive misses matter so much, we need to go back nearly four decades. The NBA introduced the three-point line in the 2026-2026 season. For many early years it was merely decoration, an option for specialized perimeter shooters rather than a strategic weapon. Coaches of that era treated a shot from beyond the 7.24-meter arc as a rare gamble, used only when the game had already stalled.

The turning point came from a small group of data people, most notably Daryl Morey, who took control of the Houston Rockets in 2026. Morey viewed basketball as a resource-allocation problem. He realized something simple that many overlooked: in a basketball game, every possession is a scarce resource, and the real question is not "is this shot pretty or ugly" but "how many expected points does this shot yield per unit of resource."

Morey's math was clean. An average two-point shot in the NBA converts at about 50%, meaning 1.0 point per attempt. An average three-point shot converts at about 35-36%, meaning 1.05 to 1.08 points per attempt. The gap is only a few percentage points, but multiplied across thousands of possessions in a season, that gap becomes a massive cumulative advantage. And if you have three-point shooters converting at 40%, the number jumps to 1.2 points per attempt, far beyond any ordinary two-point shot.

From this came what analysts call "Moreyball": maximize the two most efficient shot types, layups and threes, while nearly erasing the mid-range shot. The mid-range shot, in the eyes of a data analyst, is the worst of all shots: it yields only two points, it usually converts below 45%, and it consumes a unit of resource that should have gone to a more valuable shot.

The Golden State Warriors took this philosophy to its artistic peak. In the 2026-2026 season, they won 73 of 82 games, breaking the all-time record set by the 2026 Chicago Bulls. Stephen Curry, then twenty-eight, made 402 three-pointers in a single season, a number with no precedent. The Warriors did not just shoot more; they shot so accurately that they shook the belief that the three was a gamble.

When I began tracking games in this period, I recorded every shot attempt by every team into a spreadsheet. I wanted to see clearly that the three-point revolution did not happen uniformly, but spread in layers. The leading group adopted it first, achieved efficiency, and then the whole league followed. By the 2026-2026 season, the Boston Celtics won the title as one of the highest-volume three-point shooting teams in league history, posting a 64-18 regular-season record before their championship run.

The core: When a beautiful number collides with a small sample

This is where the story becomes interesting, and also where I must say something many do not want to hear. Morey's math is correct at the season level, but wrong at the single-game level, and especially wrong at the level of a playoff series. That is the entire problem of the 27 consecutive misses.

Let me separate two concepts that the crowd often conflates. The first is expected value, the long-run average. The second is variance, the degree of fluctuation around that average. A three-point shot has a higher expected value, but also far higher variance than a close two. In other words, the three is both a favorable bet and a volatile bet.

With a large sample, variance does not matter. If you shoot 2,000 threes in a season, the law of large numbers pulls your rate close to the true value. But a single game has only about 90 to 100 shot attempts, and a single quarter only about 20 to 25. At that small sample size, variance dominates. A 38% shooter can go 0-for-10 in one night, and that does not prove he shoots badly. It only proves the sample is too small to conclude anything.

In Game 7 of 2026, the Rockets shot 7-of-44 from beyond the arc, roughly 16%. It was a bad night in the lower tail of the probability distribution. Given the team's season average of about 35% from three, the probability of missing 27 straight is not zero. It is low, but it exists, and more importantly: it rises systematically when you shoot more and when fatigue sets in.

And here is a detail few notice: when you build your entire offense around the three, you push yourself into a situation I call "single-point dependency." If the threes fall, you win big. If the threes go cold, you have no fallback, because you have cleared the mid-range shot out of your playbook. You have no player who can retreat to the middle of the floor and hit a stable two to break the drought. You can only keep shooting threes, and keep hoping the law of large numbers returns before the game ends.

I have spent many evenings re-reading teams' playoff three-point data across years. One pattern repeats: the teams with the highest three-point dependency tend to have the largest game-to-game variance in results. They can win one game by thirty and lose the next by fifteen, with nearly the same roster and the same tactics. The difference is not in tactics, but in variance.

This leads me to a judgment I believe is core: in basketball, expected value decides who wins in the long run, but variance decides who wins a seven-game series. The NBA playoffs are among the most brutal formats in sports in this respect, because they combine two toxic factors: few games, and enormous psychological pressure that degrades shooting quality. A team can be the better team all season and still be eliminated by one bad shooting week.

Looking back at history, this is not a new phenomenon. In 2026, the Houston Rockets won the title as a sixth seed, beating teams that were stronger on paper. In 2026, the Dallas Mavericks toppled LeBron James's Miami Heat. But what is different in the three-point era is that variance is amplified. When you shoot more, you do not only raise expected value; you also widen the distribution of outcomes. You are betting more money on a more volatile game.

There is another way to look at it that I find useful: imagine each team as an investment portfolio. A team that shoots many threes is like a high-growth stock portfolio, strongly volatile. A team that shoots many twos is like a government bond, low yield but stable. Over a long investment year, growth stocks win. But over a short month with limited capital, bonds can save you from bankruptcy. The playoffs are precisely that short month.

The contrarian angle: The mispriced mid-range shot

Now I want to push back a little against my own analytics community. I still believe in data. But I believe in data so much that I must admit something many in the field avoid: the three-point revolution has gone too far, and the mid-range shot market is mispriced.

Think in the language of the transfer market. When the whole league races to shoot threes, the value of a good three-point shooter soars, while the value of a player who can create a stable two from the middle of the floor falls. That is an imbalance. In an efficient market, when one asset class becomes expensive because everyone wants to buy it, the abandoned asset class is usually the bargain.

The mid-range shot is not a bad shot. It is a misunderstood shot because it is placed in the wrong context. In a system where every other shot is either a three or a layup, the mid-range shot appears in the worst situations: when the shot clock is nearly dead, when the defense has read the play, when the player must shoot while covered. Of course it converts poorly, because it has been pushed into the role of last resort.

But try placing the mid-range shot in the role of a proactive option. The most successful teams in recent history each have at least one player who can generate a stable two when the game needs breaking open. Nikola Jokić of the Denver Nuggets is the clearest example I track: a center with scoring ability at every range, and that very versatility makes him nearly impossible to defend with a single scheme. When his teammates' threes go cold, Jokić can still score from mid-range, and that keeps the offense alive through bad nights.

This is the point I want to stress, and I will say it plainly: absolute dependence on a single shot type, no matter how efficient, is a tactical vulnerability rather than a strength. It is like a team that only knows how to attack down one wing. When the opponent locks that wing, you have nothing left. Variety is not a concession to data. Variety is a form of risk insurance.

I remember a conversation with a young coach in Da Nang a few years ago. He told me the three was the future, and that those still clinging to the mid-range shot were the backward ones. I only asked one question back: "If your team goes 0-for-15 in the third quarter of the final, what will you tell your players to do?" He fell silent. That is the whole answer. Every philosophy is beautiful until it collides with a quarter where everything misses.

There is a psychological dimension that pure data cannot measure, and here I must be careful not to cross the line of what I can prove. When a player misses ten straight, there is a cumulative psychological effect. Confidence drops, the shooting mechanism changes, and the quality of the next shot drops with it. I do not have tight enough data to assert this mechanism quantitatively, but I have enough observation to say it exists. And if it exists, it is a factor that aggravates variance, turning a bad night into a catastrophe.

This is when I recall a line I often use when analyzing big games: "Every coach talks about feeling. I have no feeling; I have standard deviation." But I must also admit: standard deviation does not help you score. It only helps you understand why you did not score. And sometimes, that understanding is all you can carry out of a loss.

Expanded analysis: Numbers worth remembering

To clarify the picture further, I want to lay out a few numbers I have collected in my tracking. In the 2026-2026 season, the Warriors posted the highest effective field goal percentage (eFG%) in history at roughly 56.3%. That number did not come from shooting one shot type very accurately, but from allocating shots intelligently across the floor. They shot many threes, but they also turned those threes into open shots through elite ball movement.

Conversely, teams that lead in three-point volume but lack corresponding passing quality tend to have lower efficiency. This is a correlation I always remind myself to treat carefully: correlation is not causation. A team shooting many threes does not automatically become efficient. It may be that ball movement and off-ball movement are the true cause, and shooting many threes is merely an accompanying effect.

This is a trap I see many fall into. They see the champion shoot many threes, and conclude that shooting many threes is the road to a championship. But if you look closer, you see the champion also has the best defense, the best passing, and the best roster depth. Shooting threes is only a surface expression of a far more comprehensive system.

I once wrote an analysis of striker Gastón Merlo of SHB Da Nang, showing that he had an expected-goals average of 0.8 per match but an actual scoring efficiency of only 0.4. A young coach of another club mocked me online. I did not argue. I published the full dataset of the next 12 matches, with shot counts and shot locations for each attempt. As a result, that club took only 9 of a possible 36 points, exactly as the model predicted. The lesson I drew was not "I was right," but: a model only has value when it dares to make a prediction that can be proven wrong. If my model only says what everyone already knows, it is useless.

Back to basketball. When I analyze a team, I always ask two questions in parallel. Question one: what shot type does their offense create? Question two: does that system have a fallback when the primary shot type is neutralized? Many teams have a good answer to question one and an empty answer to question two. That is their risk profile.

In the current regular season, I track the metric of shot attempts per unit of resource and also the volatility of that metric from game to game. A team with a high average efficiency but high volatility is a team that can beat anyone, and also lose to anyone. That is not a sustainable champion in any meaningful sense. It is a team that can win a title if luck smiles in exactly seven games.

What data cannot say

I must be honest about one thing. For many years, I believed data was the strongest weapon against prejudice. I still believe that. But I have also learned that data has limits no model can overcome.

First, data describes the past, not the future. Every prediction rests on the assumption that the context does not change. But the context always changes. Injuries, trades, coaching changes, and even off-court factors like a dense schedule or travel itineraries can break any model.

Second, data cannot measure the things that matter most in the moments that matter most. It cannot measure a player's calm before a decisive shot. It cannot measure a locker room's cohesion. It cannot measure the hunger of an underrated team. Those things lie outside every spreadsheet.

Third, and this is the most uncomfortable: data can be used to justify any decision, if you pick the right sample. Want to prove that shooting threes is right? Pick the Warriors' data. Want to prove that shooting threes is wrong? Pick the Rockets' data in Game 7 of 2026. Both are true. And both are truths cut loose from context.

That is why I always try to stand on the boundary. I do not invent a story from a number. I also do not deny the number. I try to tell the most honest story possible, using both data and hands-on experience, and with the humility to admit there are things I do not know.

A view from Vietnamese basketball

I grew up in Da Nang and have followed domestic basketball for many years. Vietnam's professional basketball league has a characteristic I always have to remind myself of when analyzing: the sample size is far smaller than the NBA's. A shorter season, fewer games, and fewer shot attempts per team in a season. That means variance dominates even more strongly.

In a league where each team plays only a few dozen games, a lucky or unlucky shooting streak can decide an entire season. A team can win a title on one hot week, and a theoretically better team can be eliminated by one bad night. This is something I see domestic fans sometimes fail to recognize when comparing their team to foreign ones.

I also notice a worrying trend in how young Vietnamese clubs copy the three-point model. They see the NBA shooting many threes, and they shoot many threes. But they ignore the prerequisite: to shoot threes efficiently, you need a system that creates open threes, and you need shooters capable of converting them. Without those two things, shooting many threes is just a way of passing the ball to the opponent.

When a young coach told me he would build a team around the three, I laughed. I touch the future with a keyboard. But I also told him that future only opens if he builds an entire system behind it, not just a single tactical decision.

Arguing with myself: What would make me wrong?

A principle I always follow when I write: if I make a prediction, I must state clearly the conditions under which it fails. So let me argue against myself.

If in the next few seasons the championship teams are all the highest-volume three-point shooting teams, with the lowest volatility, and simultaneously possess three-point shooters stable at 40% or above on high volume, then my argument that the three has gone too far weakens. If the rules of play change to further encourage the three, or if young players are trained to shoot threes better systematically, then the variance of this shot type will fall, and my argument loses weight.

I must also admit that the data on playoff variance is thin. There are not many playoff games per year, and isolating the three-point variable from others like defense or conditioning is very hard. If you want to believe the three is the absolute future, I cannot disprove you decisively. I can only point to patterns I observe, and to risks I believe are undervalued.

This is the boundary I always try to respect: distinguish sharply between what I can measure and what I infer from experience. When I say that three-point dependency correlates with result volatility, that is a verifiable observation. When I say that dependency is a tactical vulnerability, that is an inference from experience, and I must say so.

Lessons for data people

Over many years in this profession, I have realized that the biggest mistakes do not come from wrong data, but from interpreting right data too confidently. A correct number placed in a wrong context can lead to a disastrous decision. And the 27 consecutive missed threes of the Houston Rockets are proof of that.

If you are an analyst, a coach, or a fan trying to understand the game, here are the questions I advise you to always ask. One: in what context was this data collected? Two: is the sample size large enough to conclude? Three: is any variable being ignored? Four: what conditions would make this conclusion wrong? Five: if I am wrong, what will I lose?

Question five is the most important, and also the one fewest people ask. In a single game, the price of being wrong is a loss. In a season, the price of being wrong is a wasted year. In a team-building cycle, the price of being wrong is a lost decade. And in a decade, you cannot recover what was missed.

That is why I always remind myself that data is a monastery. The less noise, the more clearly you hear something trying to speak. But I must also remember that inside that monastery, there are secrets I will never hear. And humility before those secrets is what distinguishes a good analyst from an arrogant one.

Looking forward: The next cycle

If I must offer a judgment on the signals of the next cycle, here is what I will track. First, I will track the return of the mid-range shot as a strategic tool, not a last resort. Teams that realize early the value of variety will have an edge in the playoffs.

Second, I will track teams capable of reducing variance in their offensive system. These are teams that not only shoot threes well, but can also score in multiple ways when the primary way is neutralized. This flexibility will be the most important strategic asset over the next few seasons.

Third, I will track the effect of load management on variance. If a team can keep its three-point shooters fresh throughout the playoffs, it can reduce volatility. This is why I still track how teams rotate their lineups and allocate minutes. Load management is romanticized, but beneath it is a very practical data question: how to maximize shooting quality in the most important games.

I will not end with a summary. I will end with a question I want you to carry when watching your favorite team's next game. When you see your team miss shot after shot, will you think it is bad luck, or will you understand it is part of a distribution the team voluntarily accepted when it chose this path? And more importantly: does your team have any plan for the night that distribution betrays them?

People look at made shots to remember a game. I look at missed shots to understand how the game did not happen. And sometimes, in those missed shots, there is more truth than in the made ones. Numbers do not lie, but they do not tell stories either. It is my job, and yours, to learn how to tell the story the numbers cannot tell themselves.