Let us avoid rushing into hasty conclusions; performing Kolmogorov-Smirnov tests alone does not guarantee a reliable evaluation of statistical similarity if input conditions are not robust.
The validity of this realization is intrinsically linked to data quality and the relevance of thresholds chosen, a crucial point for any fund manager.
If empirical data come from a non-representative sample or reference data are poorly defined, an unexpected drawdown could occur.
For example, comparing Canadian venture capital portfolios to global indices without adjusting for local structural biases risks creating an illusion of convergence, ignoring specific geographical risks.
The mandate of fail-safe requires extreme caution in such interpretations that could mask underlying weaknesses.