It is imperative to consider failure conditions before getting excited. A Kolmogorov-Smirnov test, even well executed, cannot compensate for an inherent selection bias in the data. For example, including truncated time series without adjustment for missing data could mask significant divergences that are not detected.