The record, measured the same way every time
Fantasy scoring is mostly noise. A projection is a claim about the middle of a wide distribution, and one season can make a good model look poor or a poor one look sharp. So this page reports the whole record — every position, every season we have backtested — against a baseline that cannot be tuned after the fact.
Prior season
The player’s own per-game production from the season before. The naive forecast, carried forward.
RANK CORRELATION VS THE SEASON’S REAL FINISH · HIGHER IS BETTER · 13 SEASONS
Read the margins as small and consistent rather than large. Beating a naive carry-forward at every position is the claim; the size of the gap moves season to season, and any page presenting one season as the record is selling something.
Same measure, most recent three seasons. Included so a good year cannot stand in for the record.
- Metric
- SPEARMAN RANK CORRELATION
- Horizon
- SEASON-LONG, DRAFT BOARD
- Validation
- LEAVE-ONE-SEASON-OUT
- Population
- PLAYERS INSIDE THE PUBLISHED DEPTH
- Backfill
- NONE
We publish comparisons we can compute ourselves from public data. Anything measured against third-party sources stays internal, so nothing on this page depends on a number you cannot check.
There is no accuracy guarantee here and there will not be one. Variance decides most weeks. We are confident about the method, and the method is on the methodology page.