Draft kit

Players at each position are sorted by VORP — value over the replacement-level player at the same position, the number the rankings board already ranks by. A new tier opens wherever the drop from one player to the next is at least 2× that position's median drop. Median rather than mean, because the mean is dragged upward by exactly the cliffs the rule exists to find.

The bar is a multiple rather than a fixed number of points so it works unchanged across scoring formats whose values differ by two orders of magnitude. A position is cut into at most 8 tiers; when that binds, the smallest qualifying drops are the ones left inside a tier, so what is lost is always the weakest claim. Fewer than 4 players above replacement is reported as untiered rather than shown as one tier — “these three are equivalent” would be a finding nobody computed.

What this rule gets wrong. The boundaries depend on which players are in the pool: widening it lowers the median drop and promotes more drops to boundaries. The pool is therefore pinned to players at or above replacement level — the 120 players a 10-team league would draft — and cannot be changed by a query parameter. A genuinely smooth position comes back as one tier, which is the correct answer and looks like a broken feature; the per-position line under each board says how many drops cleared the bar so that case is legible. The full method behind the underlying values is published in the methodology.