Wiener process (brownian motion), normal distribution, log-normal, and mean-reversion are 4 most frequently used stochastic processes in modelling.
i wonder, given 30-50 sample points, is there a way/process to identify which of the 4 distributions best describe a random variable X(t)?
Even Wiener process might not be easy to distinguish as the sample points is not too many, -- we are not assuming X(t) must be one of the 4 processes, the target is just to find the best distribution to describe X(t).
It would be most preferred if there's some "industry de facto standard process" of such pattern recognition.
