What Validated

• Held-out evaluation discipline: published claims are based on periods not used for model fitting.
• Robustness checks: results are reviewed across repeated runs to reduce single-run bias.
• Risk-first reporting: drawdown and consistency are presented alongside return metrics.
• Conservative publication policy: only claims with supporting artifacts are promoted.

What Remains Provisional

• Some strategy variants and market regimes remain mixed; we do not present universal profitability claims.
• Incremental enhancements are treated as exploratory until statistical stability is established.
• Backtest-derived evidence is treated as research evidence, not a guarantee of live outcomes.

Methodology Policy

• Chronological train/validation/test separation where applicable.
• Cross-checking published summaries against underlying result artifacts before release.
• Preference for mean/robustness expectations over peak-run outcomes.