README

Regime Switching An UZH Asset Management: Advanced Portfolio Management FS17 Project

Author

Kun Yu kun.yu@uzh.ch 16-704-389 Luoyi Zou luoyi.zou@uzh.ch 16-743-536

File List

Data Files

Historical Return Data

• data.xls
• The original data file with data retrieved from BenchmarkReturns_v2.xlsx (provided by Dr. Rohner) and Bloomberg
• dataCSV.csv
• Same set of data, data.xls cleaned for the purpose of being imported to MATLAB
• new.mat
• Same set of data stored in MATLAB-friendly .mat format

Completed Estimation Results

• est.mat
• It takes tens of minutes to run the mainEstimation.m script to estimate all parameters, so we stored the resulting variables that can be simply loaded by backTesting.m to process the remaining calculation & run backtest.
• It is also possible to run the mainEstimation.m to go through from the beginning of the process.

MATLAB Scripts & Functions

• mainEstimation.m
• The entry point of the project. Estimate the P, Q and other regime-related parameters using the MS_Regress-MATLAB package.
• getBetas.m
• Called by mainEstimation.m to calculate the beta coefficients of assets for every observation windows.
• backTesting.m
• The script that calculates asset-and-regime-specific expected returns and volatilities. It also computes the optimal weights of assets across months, and calculate the overall portfolio return of the strategy.
• plotting.m
• Called by backTesting.m by the end of the script. Used to visualise the data estimated and calculated.

Acknowledgements

• Used in mainEstimation.m
• Marcelo Perlin - MS_Regress - The MATLAB Package for Markov Regime Switching Models - SSRN - GitHub Page