What “validated” actually means to us

Anyone can produce a chart that curves up and to the right. The hard part — the only part that matters — is knowing whether it means anything. In Lucy AI Markets we treat “validated” as a high bar with a specific definition, not a feeling.

Gate-verified data

A result is only as trustworthy as the data under it. Before anything is tested, the historical data has to pass a series of integrity checks. Mixed-interval or subtly corrupted data can manufacture an “edge” that never existed — so we refuse to test on data we haven't verified.

Real costs, honest fills

Spreads, financing, and slippage decide whether a strategy that looks profitable actually is. We measure per-instrument costs and model fills honestly — including the unpleasant cases, like a stop that gaps straight through its level.

Counting trials honestly

If you try enough strategies, one will look brilliant by chance. We account for how many things we tried, so a survivor has to clear a bar that rises with the search — not just beat a single lucky run.

Why we're strict about it

Because the alternative is optimising happily inside a measurement error and mistaking it for the world. A number that can't survive honest evaluation doesn't ship — and we don't pretend it did.