About this question
Medium · statistical_analysis · Quant Researcher interview question · forecast_evaluation, nested_models, clark_west, hypothesis_testing, statistical_analysis
Comparing a predictive model against a simpler, nested benchmark is a common task in quantitative finance. The standard Diebold-Mariano test is inappropriate for this, but the Clark-West (CW) test adjusts for estimation errors in the larger model. This provides a statistically sound method for determining if a new signal adds genuine out-of-sample predictive power. Task Implement the function clark_west_test(actual: list, forecast_null: list, forecast_alt: list) -> float to compute the Clark-We