Where the NBL shoots from

Why this was worth a look

There is a table in our database with 139,207 shots in it, every one carrying the coordinates it was taken from, going back to 2021. In five years nobody had drawn a single one of them, including me, so I drew this season’s.

The map

25,902 shots were taken in the 2026 NBL season across 178 games, and the feed records where every one of them came from. Mirrored into one half and binned, they make the shape anyone who watches basketball expects: a dense knot under the basket, a thick band beyond the arc, and not much in between.

At the rim, meaning within a metre and a quarter of it, the league took 6,400 shots and made 65.9%. In the paint but not at the rim, 6,989 shots at 46.9%. Beyond the arc, 10,316 shots — 39.8% of everything — at 32.6% from above the break and 35.6% from the corners.

The paint is the largest bucket on the floor and the least rewarding of the close ones. It holds more shots than the rim does and goes in 19 percentage points less often, which is the whole argument for getting the last two metres closer before letting go.

That leaves the mid-range: 2,197 shots, 8.5% of the total, at 40.2%. The middle of the floor is where shots go to be average.

The corners are the quiet exception. A three from the corner went in 35.6% of the time against 32.6% from above the break, and the league took 1,970 of them against 8,346 from up top. The best three in basketball is also one of the rarest.

Distance does what you would think, then stops

The same shots again, sorted by how far from the rim they were taken rather than by where on the floor they came from.

Accuracy falls fast and then flattens. Inside a metre, 67.9%. From 1 to 2 metres, 54.4%. By 3 to 4 metres it is 40.1%, and from there to the arc it barely moves: 41.8%, then 39.6%, then 37.3%.

A shot from four metres is worth two points and goes in 41.8% of the time. A shot from seven metres is worth three and goes in 33.7%. The arithmetic is why the middle of the floor is empty, and it is the same arithmetic in every league that has counted it.

Distance here is derived from the coordinates the feed stores, not a number the feed reports.

Two players, two maps

Two players from the same season, drawn on one scale, to show what the league map averages away: it is a picture of people doing completely different jobs.

Bryce Cotton of Adelaide 36ers took 757 shots, more than anyone: 342 of them threes, 80 at the rim, and 44.1% went in. Jo Lual-Acuil Jr of Perth Wildcats took 456, of which 18 were threes and 164 were at the rim, and made 55.3%.

The maps barely overlap. Both are good seasons.

Neither map carries many shots per patch, which is why the thin ones are left off both — four shots is not a shooting percentage. How much a small sample can carry, and where it stops meaning anything, is its own article here.

Nothing is rising

Three-point share has been flat for six seasons: 38.6%, 39.9%, 39.8%, 36.6%, 40.1% and 39.8% this year. The accuracy has been flat too, between 32.3% and 35.2%. Whatever changed basketball's shot chart happened before the seasons we hold.

What this cannot tell you

A shot chart knows where a shot came from and whether it went in. It does not know who was guarding it, how long was left on the clock, or whether the pass before it was the reason the shot existed at all. A three taken with two seconds left is not the same shot as a three taken freely, and on this map they are the same dot. If the feed ever carries the shot clock I would redraw the mid-range first, because that is where I would expect the desperate ones to be hiding.