关键词: β系数, 量化投资策略, 均值-方差模型, 动态投资组合策略

Abstract: A new 2-level dynamic portfolio management strategy for stock investment is proposed in this paper. First of all, industrial sectors are selected according to their relative strength in terms of adaptive beta coefficients. Then stocks of the selected sectors are grouped into a stock portfolio which is managed through a dynamic multi-period mean-variance model. This dynamic model uses the classic mean-variance theory as the kernel, and the model parameters-the portfolio weights-are solved out using the data from the immediate preceding period with a certain length L. The portfolio with the solved weights is then held on for the immediate follow-up period with a certain length H. These two exogenous parameters (L, H) are optimized through historical data. This dynamic portfolio management strategy has been tested using historical data from the Chinese stock market, showing better performance than passive index-tracking investment strategies in three performance appraisal metrics (including annualized returns, risk-adjusted returns and prediction market return). In particular, the dynamic portfolio strategy can earn better excess risk-adjusted returns. The 2-level dynamic portfolio strategy provides a computationally feasible and operationally reliable stock portfolio management approach.
All in all, a quantitative and dynamic asset management methodology for investors is proposed in this paper. On one hand, it verifies the Chinese stock market is ineffective; on the other hand, it can help investors decide how to allocate their wealth.

Key words: adaptive beta coefficient, quantitative investment strategy, mean-variance model, dynamic portfolio strategy