The current configuration produces continuous, non-overlapping OOS windows.
Tools / Validation Design
Walk-Forward Analysis Designer
Build a dated rolling or anchored in-sample and out-of-sample schedule before running strategy validation.
Local processingCalendar-month windowsMethodology checks
Schedule review
Report note
Rolling walk-forward design using 36-month IS and 6-month OOS windows.
Window map
Chronological validation schedule
In-sampleEmbargoOut-of-sample
Run schedule
Exact window boundaries
Only complete OOS windows are included. Partial final windows are excluded to avoid unequal evaluation periods.
| Run | In-sample | Embargo | Out-of-sample | IS months | OOS months |
|---|
Methodology and limitations
Window design should be declared before examining OOS results. The designer uses calendar months and does not account for trade count, regime balance, parameter-selection rules or execution differences. Those must be reviewed after the actual runs.
Interpretation rule
Overlapping OOS windows reuse market observations. Their aggregate results are not independent and should not be presented as additional evidence without explicit disclosure.