Moving a player across eras is a translation. A translation has a confidence.
Not a barstool claim, a translation done factor by factor. The engine takes an archetype and walks it through the differences in rules, run environment, mound height, ball construction, bullpen usage, and athletic baselines, sums them to a raw net, scales by how far the eras sit apart, attenuates for level, clamps to the data tier, and outputs a projected KR with the full per-factor math. It shows every factor it moved and by how much, and flags the whole thing model-derived until it is validated against real cross-era data, because the honest output is the projected number plus how much to trust the crossing.
Walk him back through the matrix, one factor at a time.
Take a modern archetype and move it to an earlier era. The engine looks the archetype up in the era matrix, sums the factor magnitudes to a raw net, scales by era distance, applies level attenuation, and clamps to the data-tier maximum. Every factor moves the number, and you can see by how much.
The deader ball and the higher mound push his number down; the weaker bullpens and the shallower talent pool push it back up, and the two nearly cancel, so a raw net of -2.0 scales and attenuates to a projected 82.4. The output is the per-factor breakdown, not a single adjusted number handed down. The modern athlete's edge against a shallower field almost pays for the deader ball. Almost.
Illustrative engine read on the real backward era projection (archetype lookup in the era matrix, factor sum to raw net, era-distance scaling, level attenuation, data-tier clamp), with a confidence and a model-derived flag. Composite modern slugger, demonstration figures.
Into a faster, deeper game, at a wider band.
Move a historical archetype into the Statcast era, against modern athletic baselines, modern bullpen usage with its third-time-through penalty and leverage reliever, and the modern ball and strike zone. The forward projection carries wider bands than the backward one, because projecting into a faster, deeper, more specialized game asks more of the translation. The confidence says so.
Modern athletic baselines and specialized bullpens are the two big negatives, only partly offset by the livelier ball and the lower mound, for a raw net of -6.0 and a projected 77.0 at a wide band. The engine does not pretend the forward crossing is as sure as the backward one. The further and the faster the crossing, the wider the band, and the confidence says which is which.
Illustrative engine read on the real forward era projection into the Statcast era (modern athletic baselines, bullpen usage, ball and zone), carrying a wider band than a backward crossing, with a confidence and a model-derived flag. Composite historical hitter, demonstration figures.
The table you can argue with, not the verdict you have to take.
The per-factor table is the product. For a modern power hitter dropped into the dead-ball era, the biggest crossing in the sport, every era difference is listed with its sign and magnitude for this archetype, summed to the raw net before scaling and attenuation. The point is that a reader can see exactly which differences moved the number and challenge any single one.
Six factors, each with a sign, a magnitude, and a tier, summed to a raw net of -5.0 before the era-distance scale and the level attenuation ever touch it. A reader can see the whole crossing and challenge any single line, which is the entire point of showing the table instead of the answer. Auditability is the product: the factors you can challenge, not a number you have to accept.
Illustrative engine read on the real per-factor era matrix (mound height, ball construction, run environment, bullpen usage, athletic baselines, rules and strike zone), each with its sign, magnitude, and tier, summed to the raw net before scaling and attenuation. Composite modern power hitter, demonstration figures.
An era projection is a translation, and a translation has a confidence. The engine shows every factor it moved and by how much, flags the whole thing model-derived until it is validated against real cross-era data, and refuses to launder a fun hypothetical into a settled number. The mound height, the ball, the run environment, the bullpens, and the athletic baselines each get a sign and a magnitude you can challenge, the forward crossing carries a wider band than the backward one because it asks more, and the honest output is the projected KR plus how much to trust the crossing, not a barstool verdict dressed as arithmetic.
Cross the eras. Show the factors, flag the crossing.
Era Projection walks an archetype through the era matrix factor by factor, scales and attenuates and clamps the net, and hands back a projected KR with its per-factor math, a confidence, and a model-derived flag, so a crossing is a translation you can audit, not a claim you have to settle.