The personnel-conditional NFL cube

The Ultimate
Gameplan Predictor.

Every NFL play, every personnel grouping, every formation — sliced into a 20-dimensional cube. Then layered with a coach-tendency vector, college lineage prior, and roster-archetype composition function. The result: a predictive engine that tells you what a new coaching hire is actually going to do.

Get the White Paper See the Engine →
10Seasons of per-play personnel
500K+NFL plays in the cube
85Coach DNA vectors
4,180Player archetype-seasons

Three layers. One engine.

01

Descriptive Cube

Every team's gameplan fingerprint across 20 dimensions: formation mix, personnel mix, situational pass rates, EPA by situation. Cross-season normalized. Coach-aware drift over time.

SHIPPING NOW

02

Predictive Composition

f(new coach, current roster) → predicted Year-1 fingerprint. Blends coach DNA, college lineage, roster archetypes, team-history anchor. Empirically validated: L2 error halved across five iterations.

BETA Q3 2026

03

Monte Carlo Simulator

Sample per play from P(outcome | personnel, coverage, coach play-call). Run a game 10,000 times. Output: win probability, total distributions, player prop probabilities. The betting tier.

ENTERPRISE PIPELINE

Look at what the cube sees.

Each chart below is generated directly from the cube in seconds. Not mockups. Real plays, real personnel, real coaching staffs.

SF Gameplan Fingerprint 2023

Gameplan Fingerprint — SF 49ers 2023

Shanahan's identity in one chart: 38% under-center (above league), 31% 21-personnel (vs 8% league avg), +0.46 EPA on early-down passes. The fullback-heavy outside-zone DNA, isolated.

BAL Coaching-Era Drift 2016-2023

Coaching-Era Drift — BAL 2016–2023

The Roman→Monken OC change is visible in a single column. Pistol collapsed 33%→7%. Under-center appeared. Shotgun exploded to 78%. The OC drives gameplan more than the head coach.

SF Star Player Footprint

Irreplaceable Players — SF 49ers

Personnel-conditioned EPA delta. Deebo, Aiyuk, Purdy, Kittle, McCaffrey — five players each shifting team EPA by +0.13 to +0.22. The Shanahan offense isn't a system. It's a system that requires those five specific players.

BAL Defensive Fingerprint

Defensive Identity — BAL 2023

Mike Macdonald's match-quarters scheme exposed: 39% Cover-1 (vs 30% league), −0.39 EPA allowed on 3rd-and-long. The Defensive Player of the Year award in three bars.

League Overview Matrix

Identity Distinctiveness — 32 NFL Teams 2023

Every team's z-score deviation from baseline across 10 dimensions. Ordered by total identity distinctiveness. ATL/MIA/SF lead. JAX/NE/CIN cluster at league-average. Find the outliers in seconds.

Predictive Backtest — Monken at BAL 2023

Predictive Backtest — Monken at BAL 2023

Hold-out validation: BAL 2023 excluded from coach vector build. Engine predicts the rest. 11-personnel prediction within 0.4 points. 1st/2nd/3rd-down pass rates within 7 points. L2 error 28.0 — and dropping with each iteration.

Sixteen years in the making.

2010

The Cube

A multi-dimensional cube design in VBScript. 4,826 lines. Three full iterations before the architecture converged. High-probability gameplan inference on simulation-game data. The technique worked. The substrate was the limit.

2026

The Substrate Caught Up

NFL participation data: per-play personnel groupings, jersey numbers on the field, coverage shells, time-to-throw. CFBD: college coach tendencies, rookie QB college backgrounds. The cube design translates directly. What used to be a Dictionary of dictionaries in VBScript is now a Polars DataFrame on a 146 MB Parquet store.

Now

The Engine Ships

Descriptive renders for every team. A composition function that predicts coaching-hire outcomes from college lineage and roster archetypes. A Monte Carlo game simulator on the roadmap. The 2010 design didn't need to change. The data finally arrived.

Free White Paper

The personnel-conditional NFL cube — full technique walkthrough

20-page technical paper covering the cube design, coach vector aggregation, player archetype clustering, college-to-NFL convergence priors, and the Monken→BAL 2023 backtest case study with full per-iteration L2 reduction analysis.

We'll only email you about Gridiron Science releases. Unsubscribe any time.