The complete Game Shape V2 view, plus a mutable V3 worksheet for new measures. V2 remains frozen and Golden Model 6 is unchanged.
Research viewThe comparison updates immediately; team colors anchor each side without implying a forecast.
Hover, focus, or tap underlined terms for plain-English definitions.
Experimental V2 estimate plus leakage-safe available-yards history.
Home field is estimated inside this research model.
Not a Golden Model 6 projection or a betting recommendation. Team ratings use the current season only and remain experimental. The uncertainty is much larger than the displayed margin, especially with early-season samples.
Who held the stronger game state over time?
How consistently did offensive drives create value?
Available yards taken on offense and allowed on defense.
Lab 0.1 uses total captured drive yards divided by total starting yards to goal. Offensive touchdowns finish at 100%; negative drives remain negative; clock-drain possessions are excluded. Starting-field difficulty is displayed but not yet adjusted to an expected baseline.
Descriptive event effects—not a complete measure of turnover dependency or repeatable skill.
A reading of the observed sample, without naming a projected winner.
Most recent five games currently in the Game Shape 2 file.
Uses chronological play-by-play and a win-probability model trained only on earlier seasons. It summarizes sustained control and score-state movement. It is not reconstructed from final box-score totals.
Labels use classification version 2.1. “Back-and-forth” requires both teams to hold meaningful control and control to change sides at least twice. Movement between contested and opponent-controlled states alone does not qualify.
Comebacks and collapses require actual within-game state changes. Turnover leverage, non-offensive scoring, and fourth-down WPA identify high-impact events. They do not prove that a team is dependent on those events, nor that those events will repeat.
Available yards are calculated separately from every V2 measure. V3 reads accepted V2 drive data, writes separate artifacts, and can change iteratively without changing the frozen V2 page or its 2026 confirmation track.
None of the values on this page are inputs to Golden Model 6. This page is an experimental research layer and cannot alter production projections or stamped forecasts.
Game Shape V3 Lab · complete V2 view plus additive research worksheets · V2 and Golden Model 6 unchanged