How much each algorithm scanned, and what it rejected.
Weekly and monthly views rank by consistency, not by a single day's spike.
Names ranked by several algorithms at once. Disagreement is shown, not averaged away.
The live ledger from day one — every purchase and sale with its booked profit or loss, and the money trail against its predicted range.
Solid line: actual NAV. Shaded fan: the range the engine predicted at inception (safe p25–p75, outer p5–p95). Grey dashes: the strategy's backtest before the live record began — context, not evidence.
Realized P&L from completed sales, net of costs. Grows as the record does.
Claude reviews the portfolio only when NAV departs its predicted range (guardrailed: cooldowns, monthly cap). Its findings are proposals for you — the strategy never changes itself.
The distribution of outcomes if you add money today, from the validated strategy's own measured weekly returns.