This project refers to the BARRA’s Multiple-Factor Model (MFM). According to the research ideas of constructing the MFM, in total 48 factors from the respective 5 aspects including technical indices, fundamental economy, market access return, industry allocation as well as firm characteristic factors are used to divide the individual stock abnormal return. These chosen factors are responsible for measuring various relevant situations of the risks and benefits of the individual stocks in China’s A-share market. This kind of modelling puts itself in favorable position in dividing the individual return into factor returns and specific returns cleanly. After rolling the factor return estimation window during every 120 months into a dynamic model, every month this paper calculates the total risk of a portfolio given a certain weight of positions. Total risk here is measured by the annual standard deviation estimated from the variance-covariance matrix of the predicted individual abnormal return by MFM. With this tool the portfolio manager can firstly observe the trend and the current total risk he is bearing. The last section researches into the way of attributing the risks to every asset in the portfolio. By taking the partial derivative of the total risk with respect to the weights of stocks, the marginal contribution of each asset to the risk can be acquired. This serves as an application for the investor to make corresponding adjustment in his positions in the whole portfolio according to their marginal contribution on total risk.
zhibzeng/MultipleFactorRiskModel
This project explores the way to construct the multiple factor risk model to calculate the risk contribution of each factor and the total portfolio risk using daily stock data from China.
PythonMIT