Diagnostic view: which exit profile actually differs from which, and by how much. Read-only, computed live.
Context
Every section below reflects this exact strategy/timeframe combination.
Winner
The current net-performance leader for this context, if the sample clears the eligibility floor (n≥20, at least 3 losses) — matches the League Table's own ranking logic.
scoped to context above
Evaluation funnel
Evaluation-grain (distinct evaluation IDs), never mixed with the profile-row-grain table below — how many theses actually reached a point where exit profiles could be compared at all.
Per-variant comparison
Target-reached uses exit_reason='target' (or a larger sibling's crossed-json), never status='win' alone. Lifecycle win rate (gross) and net-profitable rate (economic) are always shown separately, since a row can be a lifecycle "win" with net_pips ≤ 0 after costs.
MFE vs. target
Profile-window MFE is execution-experience only (truncated at each profile's own exit, not comparable across profiles). Common-path MFE only uses evaluations where every applicable profile reached a terminal state.
Duplicate-target detection
Variant pairs whose target distance matched on every shared evaluation — a confirmed non-match on any partial mismatch, never shown here.