Thanks, Carlos. Your comments are very appreciated and helpful. I’ve always preferred a data-driven approach, that’s why I start from data and then create something from that. Obviously, there’s more than data itself in stock markets. Concerning Welch’s test you are right about the dependency from normality and it’s an approximation i make in my article. I know that returns can be non-normally distrubuted, although there are models that make this assumption (es. Black-Scholes model for options). I’ve always thought that Student’s test is not so much dependent from normality as we may think and maybe some day I’ll write an article about this topic.