The Four Stats: LOAD, cTOV%, UW‑rTS

Good afternoon beautiful people. Recently published four stats:
Original X Thread here

Offensive Load (LOAD) — inspired by Ben Taylor

Creation Turnover Percentage (cTOV%) — inspired by Ben Taylor

Usage-Weighted Relative True Shooting (UW-rTS)

SHAI v1

The Stats

LOAD : FGA + 0.555·FTA + potAST + 0.555·FT_AST volume of possessions that end via a player's own shot or shot created for teamate

CREATION% LOAD can split into possessions ended by shooting vs by passing. Sister of USG

cTOV% TOV / LOAD — share of a player's offensive load lost to turnovers

Usage Weighted-rTS usage% × relative TS% — shooting efficiency scaled by volume.

TWOHAT UW rTS + creation%×(lg cTOV% − cTOV%) — Scoring efficiency and playmaking security, weighted by volume.

LOAD

volume of possessions that end via a player's own shot or shot created for teammate

this is the stat that started this whole binge.

When I hear USG%, I think: how often is the ball in this player's hands?

Not how often does their shot end the possession which is the actual definition:

Now USG is like this is for good reason.

Doing it that way allows everyone on a team to have a summation. It allows the stat to sum to 100.

With load, if you do an additive count of players on the team or even a starting five, you will get above 100%. As passing becomes tightly intertwined with shooting, double counting will happen.

Though I don't think a stat like this needs to summate to a hundred, it can stand on his own relatively for comparison.

It also uses potential assists rather than just assists so players aren't punished for their teammates' conversion, simply measuring how often you are the catalyst for a shot.

LOAD has value because it measures better via the box score, offensive hubs, rather than the most active shooters.

All Time LOAD

cTOV%

With LOAD as a foundational stat, we can then see different offensive actions weighted or scaled by how much offensive responsibility a player has.

In this case we have the amount of turnovers divided by the amount of generative possessions a player is responsible for.

The issue with Raw TOV% is that it divides turnovers by shots only, so it structurally punishes high-volume passers.

cTOV% uses both shooting and creation in the denominator, measuring turnovers against true offensive workload

A perfect case study for how cTOV is better than raw TOV is Chris Paul.

Routinely ranked as average in terms of ball security if you just look at turnover percentage. Again it uses a tiny denominator of shots for a player who is much more pass-heavy than shot-heavy.

Prior art

To be clear improving USG and TOV are ideas that have been floating around the space for a bit. LOAD and cTOV are extensions of the beautiful work done by @ElGee35.

Differences between the one that will be published on court-share.com/luma:

LOAD: instead of box creation as the creation input — (AST − 0.38·BoxCreation)·0.75 — it uses tracking data for Potential Assists (2014+), and a retuned free throw rate of 0.555 vs 0.44, as well as no TOV included in the formula (separated into cTOV only).

UW-rTS

This next stat is my favorite because it encapsulates what I set out to see. It was extremely simple. I didn't actually create it at all. I just merged two existing concepts.

Usage Rate and Relative TS to fundamentally see the most efficient shooters scaled by volume in one number

Curry is LOBOS, remember this is just an onball stat, doesnt take into account gravity like oTS may. Simply his own shooting efficieny and volume prowess

TWOHAT

TWOHAT will get its own thread later it is essentially Scoring efficiency and playmaking security, each weighted by volume, fused into one number.

so UW RTS + creation% × (league cTOV% − cTOV%) A high TWOHAT (+2 and up) means: a player carries a heavy offensive load, and both halves of it above league rate