> ## Documentation Index
> Fetch the complete documentation index at: https://docs.quanorix.com/llms.txt
> Use this file to discover all available pages before exploring further.

# Validation

> Walk-forward, robustness suite, and regime analysis on a Databank strategy.

# Validation

Use **Validation** to check whether a Databank strategy holds up outside the generation IS window. Three tabs: **Walk-forward**, **Robustness**, and **Regime**.

<Frame caption="Validation · Walk-forward. Click to zoom.">
  <img src="https://mintcdn.com/quanorix/2kJanlhXKp9-dfu0/images/es/validation-wf-dark.png?fit=max&auto=format&n=2kJanlhXKp9-dfu0&q=85&s=39d75c027d6f26a4e26de5d8ed5c6f5a" alt="Validation Walk-forward tab" width="1184" height="844" data-path="images/es/validation-wf-dark.png" />
</Frame>

<Tip>
  From [strategy detail](/en/databank/strategy-detail) you can open Validate with the strategy already selected.
</Tip>

## Shared across tabs

1. Pick a **Databank strategy**.
2. Review costs (spread, etc.): **prefilled from generation**, editable.
3. Data source: usually **Database** (your downloaded history).
4. Optional: run **in the background** and follow progress in [Jobs](/en/jobs/overview).
5. Export the report (HTML/Excel) when the UI offers it **after** results exist.

With no saved strategies there is nothing to validate → go back to Generator / Databank.

## Walk-forward (WFE)

Splits history into IS/OOS folds (**anchored** or rolling depending on parameters).

| Parameter | Role                            |
| --------- | ------------------------------- |
| Folds     | Number of windows               |
| OOS frac  | Out-of-sample fraction per fold |

Typical outputs:

* **WFE (efficiency)** — aggregated walk-forward metric
* **Aggregated OOS PF**, OOS Net PnL
* Per-fold table: IS/OOS PF, win%, trades, OOS net

<Warning>
  A high IS-only PF is not enough. Check OOS PF/Net and fold stability before exporting to MT.
</Warning>

## Robustness

Optional suite (enable at least one block):

| Block                  | What it checks                                                  |
| ---------------------- | --------------------------------------------------------------- |
| **Monte Carlo**        | Shuffle trades / bootstrap / price noise; PF and DD percentiles |
| **Sensitivity**        | Parameter perturbation; “robust” fraction                       |
| **Temporal stability** | Profitable months / periods over time                           |

Read P(profit), median DD, and whether small param changes kill viability.

## Regime

Classifies market segments (trend, range, high vol, warm-up) and measures edge by regime.

UI verdicts:

* **Stable edge**
* **Degraded edge**
* **Unstable edge**

Review the regime timeline, per-regime table, and PF halves/tail. If edge only lives in a rare regime, treat the strategy cautiously.

## After something “looks profitable”

1. Walk-forward with a decent OOS.
2. Robustness (at least MC or sensitivity).
3. Regime: is the edge stable?
4. Review [sessions/TZ](/en/settings/sessions-timezones) before [export](/en/export/mt4-mt5).
5. If it still holds → portfolio or evolve; otherwise discard or regenerate.

## Next

* [Databank](/en/databank/overview)
* [Portfolios](/en/portfolios/overview)
* [Jobs](/en/jobs/overview)