Robust portfolio selection with smart return prediction
Description
We propose a robust portfolio framework that optimizes parameters in return prediction by maximizing robust mean-variance utility. The proposed framework effectively combines return prediction and portfolio decisions while mitigating the uncertainty in return prediction. The data includes monthly firm-level returns and firm-level characteristics from January 1976 to December 2021.
Files
Steps to reproduce
The folder "RPO_code" contains the code for both simulation and empirical results. 1."01RPO_simulation" is the folder for simulation code. Run "01Portfolio_run_MV.py", "02Portfolio_run_MV_L1.py", and "03Portfolio_run_MV_L2.py" to calculate the simulation results of the related strategies. Figure 1, runs script "plot_picture_simulation.py". 2. "02RPO_experiments" contains the code for the benchmark strategies and three RPO strategies. Runs the numbered 01-06 py files to get portfolio weights in each folder. Table 2, Table 3, Table 4, and other results to run "get_other_result.py".