What the Stack Already Measures, and What It Skips
A modern scouting stack is deep on the body. Combine numbers, force plates, GPS load, tracking data, shot charts, advanced box scores. We can tell you how fast a prospect runs, how high they jump, and how their on-ball value holds up against tougher competition. That layer is mature, and front offices are right to lean on it.
What the stack does not measure is temperament under pressure. How a prospect carries themselves in the final four minutes of a one-possession game. Whether composure recovers fast after a turnover or leaks across the next three plays. Whether the confidence in the pre-draft interview is congruent or performed. These are the questions that separate two players with identical analytics profiles, and they are the questions every room argues about anyway, usually from anecdote and tape someone watched twice.
The behavioural layer sits above the combine. It does not replace a single number you already trust. It adds the one dimension the body metrics structurally cannot see.
It Reads the Film You Already Own
There is no new capture session, no wired-up prospect, no controlled lab. The input is the footage a front office already has: broadcast and all-22 game film, pre-draft and combine interview clips, workout video. We read facial action units from 468 MediaPipe FaceMesh landmarks, head pose, and where the data supports it, body posture and voice. The behavioural signal was always sitting in that film. It was just never scored.
The Facial Action Coding System (Ekman and Friesen, 1978) has spent more than four decades establishing that specific muscle movements map to specific behavioural states, and that the coding is reproducible across observers. What changed is speed. A trained human coder needs roughly a hundred minutes to code one minute of video. The same pass now runs automatically against a prospect's full clip library.
Because it is the existing film, you can run it on the whole board, including prospects you will never get a private workout with. The evaluation is the same whether the player is in your building or three time zones away.
Three Reads That Matter Before a Draft
The Composure Index measures decision readiness under pressure. It tracks how often stress-indicator action units fire across rolling windows and, more tellingly, how fast the face returns to baseline afterward. A prospect who shows a brow furrow for three frames after a bad possession and resets within half a second is a different competitor than one who carries that signature for two seconds while the next play develops. Recovery speed is the read, not the absence of stress.
The Authenticity Signal works the interview tape. It measures emotional congruence: whether the affect on the face matches the moment, or whether you are watching suppression and a practiced answer. It maps where the congruence holds and where it thins, so your interviewers know which threads to pull in the next conversation.
The Momentum Signature reads the clutch profile across a session: whether focus is reached, held, and self-reinforcing through a run, or whether it fragments when the game tightens. Together these three turn 'great motor, questions about the moment' from a scouting cliche into something you can point at on a timeline.
One Scale for the Whole Board
The quiet advantage is consistency. Human evaluation drifts. The prospect scouted on a Tuesday after a long flight gets a harsher read than the one watched fresh on Saturday. A scout who loves a player's college program brings that into the room without meaning to. The engine scores the late-round flyer on exactly the same scale as the projected lottery pick, with the same definitions, every clip, every time.
That consistency is also auditable. When two prospects land in the same composure band, the room can look at the same timeline and the same action-unit evidence rather than two scouts' competing memories. The argument moves from 'I just trust this kid' to 'here is where his recovery speed degraded against ranked opponents.' You can still disagree. You are now disagreeing about evidence on a shared scale.
This is where the behavioural layer earns its place in the stack. It does not ask the room to feel differently about a prospect. It gives the room a measure that holds still while a hundred names move past it.
An Input for the Room, Not a Verdict
Every read carries a confidence level, and the engine abstains when the clip cannot support a score. Thin footage, a face turned from camera, heavy motion blur, a forty-second interview cut: the system returns 'not enough signal' rather than a confident number built on nothing. That abstention is the feature. A behavioural read you can trust is one that tells you when it has nothing to say, so a low-confidence guess never gets laundered into a draft-day certainty.
Use it the way you would a strong scout's report: as one input into the room, weighed against medicals, analytics, character interviews, and the eyes of people who have watched the player live. We built it to make those conversations sharper and more even-handed.
A prospect is a person evaluated across many signals, and the behavioural layer is one of them, weighed alongside the rest. What it adds is the dimension the body-metric stack could never measure, scored on one scale for every name on the board. That is the whole job.