Hands-On Machine Learning for Algorithmic Trading

Model definition

To model the Sharpe ratio as a probabilistic model, we need the priors about the distribution of returns and the parameters that govern this distribution. The student t distribution exhibits fat tails that are relative to the normal distribution for low degrees of freedom (df), and is a reasonable choice to capture this aspect of returns.

Hence, we need to model the three parameters of this distribution, namely the mean and standard deviation of returns, and the degrees of freedom. We'll assume normal and uniform distributions for the mean and the standard deviation, respectively, and an exponential distribution for the df with a sufficiently low expected value to ensure fat tails. Returns are based on these probabilistic ...