Re-run and quantify all backtest results affected by missing required data

Re-run the backtest and research-harness results produced while the live system’s required-data list exceeded the replay’s requested-data list, and determine the size and direction of the resulting errors.

Background

Defect 5 caused the replay harness to omit data that the live path downloaded, causing a decision input to read as zero or to be only partially computed. The defect persisted for approximately four months and affected every harness run during that period.

Although restoring the missing data reproduced the live value and corrected the affected replay, the paper reports that the prior results have not yet been systematically recomputed. Consequently, the magnitude and direction of the bias in the affected backtests and research rejections remain unresolved.

References

Their size and direction of error are unknown, because none has been re-run yet.

— Where Backtests and Live Trading Part Ways: Six Defects from One Intraday System  (2610.21846 - Vidovich et al., 15 Oct 2026) in Section 2.5, Defect 5; Section 5, Limitations