Positionless basketball and the collapse of positional ranking

What positional labels were doing

Traditional positions grouped competitors by height, location on the court and the specific responsibilities that followed from both. The grouping was useful because it made comparison meaningful, since two competitors with the same label faced broadly similar demands. It was always an approximation, and the approximation held well enough while tactical convention kept responsibilities reasonably fixed.

As offences began using competitors interchangeably, the label stopped predicting what somebody would actually be asked to do. A comparison framework built on labels then compares people whose jobs have diverged considerably despite sharing a category. The framework does not announce its own obsolescence, so it keeps producing confident orderings long after the categories underneath have stopped meaning anything.

Why the change happened

Increased emphasis on spacing rewarded competitors who could operate away from the basket regardless of their size. Defensive schemes that switch assignments require every competitor to defend several types, which erodes the specialisation positions encoded. Improvements in skill development mean attributes once concentrated in particular sizes are now distributed much more widely.

The result is that responsibilities are assigned by capability rather than by category, which is what positionless describes. The label survives in broadcast and in registration because it is convenient, not because it describes the game being played. Convenience is a perfectly good reason for a label to persist, and a very poor reason for an analytical framework to keep resting on it.

What replaces positional comparison

Grouping by role rather than by label restores meaningful comparison, since role describes what somebody is actually asked to do. Roles can be identified from behaviour, including where attempts are taken, how possessions are initiated and which assignments are defended. This grouping changes over time and even within a season, which makes it more accurate and considerably harder to maintain.

It also produces categories that do not match public vocabulary, which limits how readily such rankings are adopted. The trade between accuracy and legibility appears here as clearly as anywhere in ranking design. A category that readers recognise instantly is worth something real, and it is worth less than a category that corresponds to what actually happens.

Stale baselines and mismeasurement

Positional baselines derived from historical data describe the demands of a role as it used to be rather than as it is. When responsibilities shift, everybody in the affected category is measured against a standard that no longer applies to them. The error is systematic rather than random, which means it does not average out across a large sample of competitors.

Updating baselines continuously introduces instability, since a baseline that moves makes comparison across seasons harder. There is no setting that avoids both problems, and stating which one has been accepted is the available remedy.

Ranking without positions

One option is to abandon categories entirely and rank on contribution measures that do not require a positional reference at all. Impact-based measures do this naturally, since they estimate effect on outcomes without needing to know what somebody was supposed to do. The cost is the loss of interpretability, because a single number gives no account of how the contribution was made.

Pairing an impact measure with a behavioural description recovers some of that without reintroducing rigid categories. That pairing is probably where positional ranking is heading, and it will be less tidy than what it replaces. Tidiness was always the main virtue of the positional system, and it was purchased by ignoring differences that have since become too large to overlook.