Not affiliated with or endorsed by the NBA. Team names are used descriptively.

Guide

Z-scores in fantasy basketball: what they are and how to read them

Published 7 min read

Short answer

A z-score says how far a player's average sits from the pool average, in units of that stat's typical spread, so points, rebounds and steals can be added together. Turnovers are flipped because fewer is better, and percentages are weighted by attempts so volume shooters count for more.

What a z-score is

A z-score says how far a player’s average sits from the average player, measured in the typical spread of that stat. Zero is average. +1 is one spread better. -1 is one spread worse. The point of the measure is that it puts points, rebounds, steals and percentages on the same scale, so you can add them.

Without it, you cannot compare a player who scores 24 points with one who blocks 2.5 shots: the numbers mean different things. With it, you ask the same question of every stat: how unusual is this number, in the direction I want?

The formula is one line:

z = (player’s average minus pool average) divided by the pool’s standard deviation

The standard deviation is the usual distance between a player and the pool average. A big standard deviation (points) means a +3 point gap is small. A small one (steals) means a +0.3 steal gap is large.

A worked example with a made-up pool

Every number here is invented. Take eight players, A to H, and four categories: points, rebounds, turnovers and field goal percentage. Here are their per-game numbers.

Player Points Rebounds Turnovers FGM / FGA
A 24 5 3.4 8.3 / 18
B 18 9 1.8 6.0 / 12
C 15 4 1.2 4.5 / 9
D 12 11 1.5 4.2 / 7
E 20 6 2.9 6.6 / 15
F 9 3 0.8 2.6 / 5
G 16 7 2.2 5.0 / 11
H 10 8 1.0 3.3 / 6

For points, the pool average is 15.5 and the standard deviation is 4.8. Player A scores 24, so A’s points z-score is (24 - 15.5) / 4.8 = +1.77. Player F scores 9, so (9 - 15.5) / 4.8 = -1.36.

For rebounds, the average is 6.62 and the standard deviation is 2.5. Player D, with 11, has (11 - 6.62) / 2.5 = +1.75. Player A, with 5, has -0.65.

Two categories need the special treatment below, but here are all four z-scores per player, added into a total.

Player Points Rebounds Turnovers FG% Total
A +1.77 -0.65 -1.79 -1.05 -1.72
B +0.52 +0.95 +0.06 +0.31 +1.84
C -0.10 -1.05 +0.75 +0.24 -0.16
D -0.73 +1.75 +0.40 +1.70 +3.12
E +0.94 -0.25 -1.21 -1.56 -2.08
F -1.36 -1.45 +1.21 +0.35 -1.25
G +0.10 +0.15 -0.40 -0.80 -0.95
H -1.15 +0.55 +0.98 +0.81 +1.19

Notice what the table says. Player A has the best scoring line by far and ranks near the bottom, because his turnovers and shooting cost more than his points earn. Player D scores the least but ranks first, because he is strong where the others are not. That is the idea of a z-score ranking: it measures total contribution across categories, not the headline stat.

Why turnovers are negative

Turnovers are the one counting category where less is better. A player with 3.4 turnovers is worse than average for you, so his z-score is flipped: (average minus his number) divided by the spread. A is at 3.4 against a 1.85 pool average, so his turnover value is -1.79. F, with 0.8, gets +1.21.

If you forgot the flip, turnovers would reward the players who hurt your team most, which is the most common mistake in homemade rankings.

Why percentages are weighted by attempts

You cannot z-score a percentage directly. A team’s FG% is total makes divided by total attempts, so a player who shoots 18 times a night moves it more than one who shoots 5 times, at the same percentage.

The fix is to measure made shots added over an average shooter taking the same attempts. In our pool the average shooter makes 48.8% (the pool’s total makes over total attempts). Player A takes 18 shots and makes 8.3. An average shooter would make 18 x 0.488 = 8.78, so A is 0.48 makes below average. Player D takes 7 and makes 4.2, against 3.42 for an average shooter, so D is 0.78 above. D’s shot percentage (60%) is better than A’s (46%), but the real gap is bigger than the percentages suggest, because A’s volume drags the team’s percentage down more than D’s volume lifts it up.

Then those made-shot differences are z-scored like any other stat. That is where the FG% column in the table above comes from: D is +1.70, A is -1.05. The same reasoning applies to free throw percentage. The site’s percentages explainer shows a worked pair of shooters at the same percentage and different volume.

Why a z-score is not a projection

A z-score is a description of what happened in the window you chose, compared to the other players in the pool. It says nothing about what happens next week. Rankings built from per-game averages are only as good as the window: a player back from injury may rank well on a full season and badly on the last month. Fantasy Hoops HQ does not forecast; it lets you choose the stat window and rebuilds the pool in that window. See stat windows.

The pool also matters. A z-score of +1 means one spread above the pool average, so it moves if the pool does. A 12-team league’s rostered players are a different pool from every player in the league.

What changes when you punt

Because values are added across categories, you can leave a category out and re-sort. If you punt turnovers in the table above, Player A’s total rises from -1.72 to +0.07, and he moves from near the bottom to near the middle. Player D loses 0.40 and drops to +2.72, still first. A punt build lifts the players who are weak only where you have stopped competing. The punting guide covers when that is worth doing.

Fantasy Hoops HQ weights a target category double, an in-play category once, and a punted one at zero. The rankings page applies your choices, and the category values section shows the calculation for one player.

Limits worth knowing

A z-score total treats a point of value in steals the same as a point in rebounds, though a head-to-head week may not. Standard deviations also work best on a roughly bell-shaped spread, and some NBA stats are not: blocks, for example, bunch near zero with a long tail of elite shot blockers. Basketball Monster’s rankings page says its DURANT values adjust for the non-normal distribution of NBA stats. Treat any z-score ranking as a good first sort, then adjust for your own league’s categories.

Common questions

What is a z-score in fantasy basketball?

It is (player's average minus pool average) divided by the pool's standard deviation. Zero is average and +1 is one spread better than average.

Why are turnovers negative in z-score rankings?

Fewer turnovers is better, so the z-score is flipped. A player with more turnovers than average gets a negative value.

Why are FG% and FT% weighted by attempts?

A team's percentage is total makes over total attempts, so a high-volume shooter moves it more than a low-volume one. The value is the made shots added over an average shooter on the same attempts.

Are z-score rankings projections?

No. They describe what happened in the window you chose, compared with the other players in the pool, and say nothing about future minutes.

Sources

Checked on .

Try in tradeCompare

Per-game numbers in the named window. Games are this season's schedule under your league's games limit.