Opening the research
Opening the research
How Crosswake separates an interesting result from a reliable conclusion.
Each version has a written hypothesis, fixed rules, and pass/fail criteria before test data is fetched. Changing a rule creates a new version rather than rewriting the old result.
Development uses an earlier period. Evaluation uses a later period the strategy has not seen. Chronological separation matters because market information available in the future must not influence an earlier decision.
The main daily strategy developed on 2020–2024 and tested on January 2025–September 2026. These periods are historical research, not live returns.
Results include modeled fees and slippage. The daily trend model uses 30 basis points per round trip. The futures research also considers funding.
Actual execution can differ. A modeled cost assumption is not a promise of an attainable fill.
A positive mean is only one part of the result. Crosswake considers sample size, profit factor, and a 95% confidence interval. The validation standard requires positive expectancy with a confidence interval above zero.
v005 has a positive test average, but its confidence interval crosses zero. It has not established a proven edge. v003, v004 and v007 failed their tests and run only for observation.
Paper tracking records new signals as market data arrives. It helps compare forward behavior with the frozen research, without placing real orders.
Past performance does not guarantee future results. Paper evidence must be judged over enough independent observations, not a handful of winning trades.