Hands-On Machine Learning for Algorithmic Trading

Optimal investment – multiple assets

We will use an example with various equities. E. Chan (2008) illustrates how to arrive at a multi-asset application of the Kelly Rule, and that the result is equivalent to the (potentially levered) maximum Sharpe ratio portfolio from the mean-variance optimization.

The computation involves the dot product of the precision matrix, which is the inverse of the covariance matrix, and the return matrix:

mean_returns = monthly_returns.mean()cov_matrix = monthly_returns.cov()precision_matrix = pd.DataFrame(inv(cov_matrix), index=stocks, columns=stocks)kelly_wt = precision_matrix.dot(mean_returns).values

The Kelly Portfolio is also shown in the efficient frontier diagram (after normalization so that the absolute ...