// Three findings. Fixed meanings.

THE FINDINGS SYSTEM

Every test ends in exactly one of three findings. Learn to read them is RAMP 02, and it is most of what separates evidence from a sales pitch.

// THE THREE STATES

EDGE CONFIRMED

The strategy beat its benchmark on BOTH compound growth (CAGR) and risk-adjusted return (Sharpe), in the training window AND in the held-out walk-forward window it never saw. Both axes, both windows, or it is not CONFIRMED.

TRADE-OFF

It beat the benchmark on one axis only: more return with worse risk, or better risk with less return. Not a failure, not a green light. You decide if the trade is worth it; the engine will not decide for you.

NO EDGE

It did not survive. The engine says so plainly, and so will your AI: the instructions it receives forbid softening this finding. Rejecting strategies is the job. The FALSIFIER badge exists for a reason.

// THE WALK-FORWARD SPLIT

Every study is split chronologically at the OOS SPLIT YEAR (default 2015). Criteria are formed and evaluated on the earlier window, then replayed untouched on the later window the strategy has never seen. Random splits are refused by design: shuffling time leaks the future into the past, and a leaked backtest is a fiction with good numbers.

// THE METHODOLOGY, STATED IN FULL

Survivorship-complete. 41,357 tickers including every delisting. Most tools quietly drop the dead companies and every strategy looks brilliant.
Point-in-time fundamentals. Joined by filing date. The engine only knows what the market knew on the day.
No look-ahead. Positions lag signals; formation-only filters; the forward return must span exactly one calendar period.
Gross returns. Costs, taxes, and slippage are not modelled, and every result says so. Real results would be lower.
Complete record. Losing periods are shown, always. A backtest is not a promise of returns.

That is the whole system. When your AI presents QuantGPT numbers, it repeats this methodology, because the manifest it receives on connect tells it to.