wc-captain-ladder — the rolling-captaincy edge

Implements MB2 — Captain Ladder from footballfantasy/context/frameworks/building-blocks.md, the captaincy clauses of league-config.md ("Captaincy — the rolling-captain edge"), and the captaincy-in-EV note in scoring-rules.md. It supplies the captaincy_uplift term that wc-fitness-eval adds inside raw_xEV (see frameworks/fitness-function.md).

This is the single biggest structural difference from FPL and the largest single lever in wc-matchday-tactician's claimed ~20–40 pts/round active-management edge. In FPL the captain is locked at deadline; here the armband doubles the round's points and can be moved during the round to any captain candidate who has not yet kicked off. So the question is never "who is the highest-xEV captain?" — it is "what is the expected value of the best armband I can realise by watching the early games and re-deciding before each later candidate kicks off?"

That makes captaincy a sequential decision under uncertainty, not a point estimate. A merely-good captain who plays in the late kickoff is worth more than a marginally-better one who plays first, because the late slot lets you bank an early hauler and only ride the late man if the early games disappoint. This skill prices that optionality.

Why captaincy_uplift is not 2× the best player

Define xEV_cap(p) as the (already fixture-scaled, minutes-weighted) expected captain bonus from player p — i.e. the extra points the armband adds, which equals one more copy of p's round score: xEV_cap(p) = xEV(p). The naive captaincy value is max_p xEV(p): pin the armband on the top player and never touch it.

The ladder beats that by an option premium. With candidates on different days you observe the early candidate's actual return, then keep-or-switch. You capture the maximum of correlated-but-staggered draws rather than a single fixed draw: