Walk-Forward Validation Quant Interview Guide
Walk-forward validation quant interview guide for rolling train/test periods, expanding windows, strategy evaluation, limitations, and examples.
Candidates evaluating time-series models and trading strategies.
Walk-forward validation simulates repeated deployment
Walk-forward validation trains on an earlier window and tests on the next period, then moves forward. It is useful when models would be refreshed over time.
Rolling versus expanding training
A rolling setup drops older data as it advances, while an expanding setup keeps all past data. Rolling can adapt faster; expanding can reduce estimation noise if older data remains relevant.
Concrete example
Train a signal on year one, test on the next quarter, then advance the window and repeat. The sequence gives a more detailed view than one lucky holdout period.
It does not remove every risk
Walk-forward validation can still be overfit if many variants are tried or if the test windows share the same favorable regime. Costs and selection discipline still matter.
Common mistakes
Candidates often say walk-forward and stop there. A better answer specifies window sizes, retraining cadence, metrics, and what failures would invalidate the model.
Practice the pattern
Use the LeetQuidity curriculum and calibration to turn this topic into a focused practice plan.