Greatness is not era-proof. It is era-priced.
The engine holds ability fixed and reads how the game's shifts re-price a player's archetype, forward and back.
The rating does not move through time. Its value does.
The underlying trait identity is locked and era-agnostic, never touched. A separate Era-Projected KR reads how the documented shifts re-price this archetype, reported as a band, because an era projection is a counterfactual with no direct data. A composite ball-playing defender, projected forward from an older era into the modern game.
His 84 never moved; it is his ability, and ability is not era-dependent. The modern game simply values that ability far more, so his Era-Projected KR lands in a 90-to-94 band, reported honestly as a model-derived counterfactual, not a fact. The projection re-prices the archetype and stops. It does not say the modern game is better, only that it pays more for this player. The player is the constant. The era is the price.
Illustrative engine read on the real era-projection structure (locked era-agnostic OVERALL, a separate model-derived Era-Projected KR band, the per-factor re-pricing). Composite player, demonstration figures.
What one era elevates, the other buries.
The seven shifts re-price every archetype both directions: a tailwind projecting forward is a headwind of the same size projecting backward. The roles the modern game invented gain most going forward and nearly vanish going back; the roles it squeezed do the reverse. Two composite players, opposite archetypes.
The magnitude is symmetric; only the sign flips with direction, because if the modern game elevated a skill, removing that context removes the elevation. This is the death-of-the-number-10 case made precise: the free central creator and the target man read as premiums in one era and luxuries in another, priced both ways without ever ranking the eras themselves. An era does not make a player better or worse. It changes what his particular gifts are worth.
Illustrative engine read on the real bidirectional archetype matrix (forward tailwind equals backward headwind of the same magnitude; modern-elevated versus older-devalued roles). Composite players, real archetype magnitudes.
No big era swing on a shaky archetype.
An era projection rests entirely on the archetype, and an archetype is only as trustworthy as the data tier that produced it. So the maximum era swing is capped by confidence: a thinly-read player cannot be swung as far as a richly-measured one, whatever the raw factors say. When the raw net exceeds the cap, it is clamped and flagged.
The factors might say plus eleven, but if the archetype behind them is a reasoned guess, that confidence is not there to spend. The swing clamps to the tier, the clamp is flagged, and the whole thing is reported as a model-derived band, because an era projection is a thought experiment on one scale, not a measurement. The size of the claim can never outrun the quality of the read.
Illustrative engine read on the real confidence-gate structure (max era swing clamped by the archetype's data tier; GATE_CLAMPED flag; model-derived band). Composite player, demonstration figures.
The player is the constant and the era is the price, and the engine reads that price both ways without ever calling one era better.
Hold the player fixed. Price his era.
Era Projection re-prices a locked archetype across football's documented shifts, both directions, gated by what the read can support.