All scores →
Loading…
Loading…
Loading…

Analytics myths: five things the numbers do not say

By 4Basketball Staff · · 11 min read

Basketball analytics is often summarized in slogans that the research does not support. This article takes five common ones, from the midrange ban to the hot-hand myth, and explains what the numbers actually say.

Analytics has a public image problem in basketball, and part of it is earned. The field has produced real discoveries, and it has also produced slogans: no midrange shots, only threes and layups, the hot hand is a myth, one stat tells you who is best. Slogans travel faster than the studies behind them, and by the time they reach a broadcast or a comment section they are often the opposite of what the research found. The numbers rarely say as much as people claim, and what they do say tends to be more qualified, more interesting and more useful.

This article takes five widely repeated claims and sets each against the evidence. The purpose is not to defend or attack analytics. It is to help a fan tell the difference between what the data show and what has been added on the way to the screen. If you want the underlying concepts, the plain-English companion is Basketball stats glossary: from PER to RPM in plain English. The thread running through all five myths is the same: a number describes something specific, and trouble begins when it is stretched beyond that.

Myth one: the numbers say never take a midrange shot

The claim: analytics proved that the midrange jumper is a bad shot, so smart teams ban it.

What the numbers say: on average, a long two-pointer produces fewer points per attempt than a three or a shot at the rim. That is a real and robust finding, and it explains why the league's shot mix changed so much. It is covered in Why teams stopped taking midrange shots: the efficiency argument.

What they do not say: that the shot is bad for every player in every situation. An average is a blend of excellent shooters and poor ones, of open looks and desperate ones. A player who makes half of his long twos is producing a point per attempt, which is competitive with the league's average three, and a defense that takes away the arc and the rim leaves the midrange as the shot on offer. The sensible reading is a question, not a ban: is this the best shot available to this player right now? Teams that treated the finding as a prohibition often discovered that they had made themselves predictable. The full story, including how the shot returned in the hands of specialists, is in Why the midrange jump shot fell out of fashion, and why it is returning.

The lesson generalizes. A finding about the league average is a statement about a population. It tells you where to look for improvement, and it says nothing certain about a single case.

Myth two: one number can settle who is better

The claim: pick the right all-in-one statistic and it will tell you who the best player is.

What the numbers say: composite statistics are useful summaries. Player Efficiency Rating, created by analyst John Hollinger, collapses a box score into a single per-minute figure. Win shares, box plus-minus and value over replacement player attempt to turn production into wins or points of value. These are discussed in Player efficiency rating: what it measures and what it misses and Win shares, BPM and VORP: the all-in-one stats. They agree with each other and with the eye test on most players, and they are a good first screen.

What they do not say: that the ranking they produce is the truth. Each formula embodies choices about how much to credit a rebound, how to treat a missed shot, what to do about defense and how to split credit among teammates. Defense in particular is poorly captured by box-score statistics, because a stopped possession leaves no entry in the box. A player who guards the opponent's best scorer well can appear in the box score as a modest contributor. The difficulty is laid out in Defensive metrics: why defense is hard to measure.

The other tools have their own problems. Plus-minus is affected by who shares the floor with a player, which is why Plus-minus: on/off and why the raw number lies treats the raw figure with suspicion. Usage rate, discussed in Usage rate: how much of the offense a player uses, shows how much of a team's offense runs through a player but not how well he uses it. A fair summary is that the different numbers answer different questions and a careful analyst reads several together. When two all-in-one statistics disagree about a player, the disagreement is information: it usually points to the type of value that one of them misses.

Myth three: the hot hand is a myth

The claim: players do not get hot; streaks are an illusion, and analytics proved it.

What the numbers said, and then said again: in 1985, psychologists Thomas Gilovich, Robert Vallone and Amos Tversky published a famous paper that analyzed shooting records from NBA games and a controlled experiment. They found no evidence that a made shot made the next one more likely, and concluded that the belief in streak shooting was a cognitive illusion, a case of people seeing patterns in random sequences. For decades the result was cited as a model of how intuition misleads.

What they do not say: that the question was closed. In 2018, economists Joshua Miller and Adam Sanjurjo reported a subtle statistical bias in the way the original analysis, and many like it, had looked at finite sequences of shots, a bias that tends to hide any real streakiness. Correcting for it, they found evidence of a modest hot-hand effect in the data they re-examined. The dispute has continued in the literature, and the honest summary is that the effect, if present, is much smaller than fans believe and harder to detect than anyone assumed.

Two ideas from this story are worth keeping apart. One is that people, including players and fans, read far too much into short streaks, because random sequences contain more runs than intuition expects. That point is as sound as ever, and it is the heart of Small samples in basketball: when a hot streak is noise. The other is that a player's shooting probability might actually vary from one shot to the next, a claim the data can address but not easily. "Most hot streaks are noise" is well supported. "No one is ever hot" is not something the numbers establish. A good habit is to be wary of a claim that is both surprising and absolute.

Myth four: the numbers have made the eye test obsolete

The claim: scouts and coaches who watch games are being replaced by spreadsheets, because the data see what the eyes miss.

What the numbers say: data are very good at certain tasks. They count what happens across thousands of possessions, which no observer can do, and they catch patterns that are invisible in a single viewing, such as which kinds of shots a team gives up. Tracking systems record the positions of every player many times a second, which has made it possible to measure things that once required guesswork, as discussed in Expected points and shot quality in the tracking era.

What they do not say: that watching games has nothing to add. Numbers record what happened; they rarely explain why. A player's shooting percentage cannot tell you whether his form is sound, whether he is playing hurt or whether he is out of position because a teammate missed an assignment. Context, such as the scheme, the roles and the quality of the opposition, is carried more reliably by people who study film. In the draft, for instance, the best-run approaches combine statistical models and scouting, as in How draft prospects are evaluated with data. Neither side has a monopoly on the truth, and the teams that use both generally make fewer mistakes than those relying on only one.

There is a subtler point. Numbers and film check each other. When a player's statistics are better than his film suggests, it is worth asking whether the stats are inflated by a role or a hot month. When the film looks better than the statistics, it is worth asking whether the player's contributions are the kind that statistics miss. The tension between them is often where the interesting analysis begins. A fan who sets the two against each other is giving up half the available information.

Myth five: analytics wants every team to play the same way

The claim: the numbers have made the game uniform, with every team shooting threes and layups and every player the same.

What the numbers say: there is a general pattern. The shot values described in The three-point revolution in the numbers pushed the league toward more threes and shots at the rim, and the shift in the shot mix is real. Teams across the league do look more alike in their shot charts than they did a generation ago.

What they do not say: that the best strategy is the same for every roster. The numbers describe the value of shots, and the right strategy depends on who is on the floor. A team with great shooters should use them. A team with a dominant big man or an elite pull-up scorer might do well to build around that, and plenty have. The mix of styles in any given season is wider than the slogan suggests, from teams built on pace and three-point volume to teams that grind in the half court, as seen in Pace, possessions per game and why it changes everything. And the numbers say nothing about whether a style is enjoyable to watch. That is a judgment about taste, not an output of a model, and people who like the older style are entitled to say so.

Defenses and positions are part of the picture. The same analysis that favors threes also favors defenders who can guard several positions, big men who can protect the rim and also step out, and players with varied skills. Variety returns through the back door, a point developed in Positionless basketball: do the five positions still matter?. If anything, the numbers reward being unpredictable in the right way.

Why the slogans spread

It helps to understand why these distortions are so durable, because the cause is not stupidity. A good slogan is short, and a careful finding is long. "The midrange is the worst shot" fits in a headline; "the average long two produces fewer points than the average three or rim attempt, though the gap depends on the shooter, the situation and the defense" does not. When a finding is repeated, the qualifications are the first thing to be dropped.

There is also a story people like to tell about numbers in sports, one that owes much to the baseball book Moneyball: the clever outsider with a spreadsheet beats the old guard that trusts its gut. The story is entertaining and sometimes true, and it casts the data as a weapon in an argument between camps. In basketball, where front offices employ both film analysts and statisticians, the camps are largely a fiction. Most teams use both kinds of information, and the real disagreements are about particular cases.

Finally, numbers carry an air of authority that makes it tempting to stop thinking. A table of figures feels like a verdict. But every figure was produced by someone who decided what to count, how to count it and what to leave out, and those decisions can be examined. The mark of a trustworthy analysis is that it tells you its limits. The mark of a slogan is that it does not.

A short field guide to overreach

The five myths share a structure, and recognizing it helps with the sixth and seventh you will meet. A table summarizes the common moves.

The move What it sounds like What to ask
Average becomes rule "Analytics says never do X" Who does X, and how well, compared with the average?
One number settles it "He leads the league in stat Y, so he is the best" What does Y leave out? Do other measures agree?
Absence of proof becomes proof of absence "Studies found no effect" How large was the sample? Were the methods later challenged?
Numbers versus eyes "The data say the scouts are wrong" What would film show that the data cannot?
Description becomes prescription "The numbers say everyone should do this" Does it hold for this team, with these players?

Each move takes a true, narrow statement and stretches it into a broad one. Often the person who stretches it is not misleading anyone deliberately; the narrow version is just less quotable.

Uncertainty is part of the answer

One more habit separates careful readers of analytics from careless ones: they ask how sure the number is. A shooting percentage from a hundred attempts has a wide margin of error, and a plus-minus figure from a small number of minutes is mostly noise. A forecast that gives a team a 70 percent chance leaves 30 percent for the other result. Good analysts report ranges and probabilities, and good readers keep them in mind. The pattern appears again and again, for example in Regression to the mean: why extreme seasons rarely repeat, which shows why an unusually good season tends to be followed by a less good one even when nothing has changed. Clutch statistics, which are famous for their small samples, are the standard cautionary tale: see Clutch statistics: what clutch means and why samples are small.

What to take away

Analytics is a way of counting carefully, and its best results are modest and specific: this shot is worth more on average than that one, this stretch of games is too short to tell, this stat captures offense better than defense. The slogans built on those results are bolder and usually wrong. When you hear that the numbers say something, ask what was measured, over what sample, for whom and with what uncertainty. Those questions will not make you cynical about the numbers. They will make you a more useful consumer of them. How this site's own numbers are sourced is set out on the data sources page, and the best way to build judgment is to look at a number and ask what it leaves out.

Search more on this

Keep reading

Quadrant records and the NCAA selection committee's NET

Quadrant records sort every college game by the NET rank of the opponent and the location. This article explains the four tiers, why a road win counts for more, how the cutoffs create cliffs, and how the committee uses them.

4Basketball Staff · Apr 27, 2026 · 11 min read

How to build a simple team rating at home

A useful team rating needs only scores and a spreadsheet. This article walks through a four-team example by hand, then shows how to scale it, adjust for home court, turn ratings into win probabilities and avoid the common traps.

4Basketball Staff · Apr 19, 2026 · 10 min read

Usage rate: how much of the offense a player uses

Usage rate estimates the share of a team's possessions a player ends with a shot, a trip to the line or a turnover. Learn the formula, the typical ranges and why usage and efficiency must be read together.

4Basketball Staff · May 6, 2026 · 11 min read