Edge

Does Sonar actually have an edge? Measured on real recorded outcomes — every flagged name, held to today, after estimated trading costs. The honest gate before any automation.

Marginal — costs eat most of it

Gross looks fine (+18.4%) but after spreads/fees it's only +16.3%. Tradeable edge is thin; be selective, not systematic.

129

Signals scored

+18.4%

Avg gross

+16.3%

Avg after costs · est.

20%

Win rate (net) · gross 31%

Does the conviction score predict outcomes?

High conviction (80+)9 sig33% win-11.9%

Mid (65–79)96 sig20% win+24.8%

Watch (<65)24 sig17% win-7.1%

If higher tiers earn more, the score has predictive value. If not, the engines need work.

Which engine carries the edge?

Stocks40 sig18% win+64.3%

Crypto73 sig21% win-5.4%

Options16 sig25% win-4.6%

Outcome distribution (net of costs)

≤ −20%

−20 to −5%

−5 to +5%

+5 to +20%

> +20%

Honest caveats. Returns are current price vs. the price when first flagged — live, not a fixed-horizon backtest (no defined exit). Costs are estimated (penny 3% / lottery 4% / other 1.5% round-trip) — your real fills may differ. Renamed or delisted tickers drop out (survivorship). Sample is still small. This measures whether the signals have edge — not a promise of future results. Research, not advice.