I would keep the raw rows beside the repaired series, with the source and adjustment version recorded. One regression case can use the same symbol and date with conflicting raw and adjusted closes. It should reject the mixed input before the backtest starts, then accept exactly one consistently adjusted row. That makes a later import unable to bring the phantom gap back unnoticed.
Arnold Holm
Challenge Pre-Check: know before you risk. Your rules, tested against prop limits. Historical simulation, not a forecast.
I would keep the raw rows beside the repaired series, with the source and adjustment version recorded. One regression case can use the same symbol and date with conflicting raw and adjusted closes. It should reject the mixed input before the backtest starts, then accept exactly one consistently adjusted row. That makes a later import unable to bring the phantom gap back unnoticed.