A Note on Modifying the Mean-Absolute Deviation Portfolio Optimization Model to Account for Nonstationarity Biases