本文先回顾了投资组合理论和DCC-GARCH模型在国内外的相关研究文献。然后从 程序量化角度出发，通过第三方平台实现程序量化交易系统来验证国内股票价格的时 间序列分析中的统计模型AR-GARCH模型、DCC-GARCH模型、投资组合理论中的有效 前沿分析，是否能在交易系统的数据回测中体现出投资收益的增加。 研究标的是上证50成份指数中的50个标的股票的日线价格数据，通过对乖离率 建立GARCH模型，在原有均线系统上叠加股票收盘价对均线的平均偏差，实现改进传 统的双均线交易系统，通过长短周期的均线交叉为明确的买卖点，交易最多5个股票。 通过DCC-GARCH模型预测下一日的波动值和动态相关系数，再使用蒙特卡洛法，在有 效前沿边界上寻找最佳的配置参数。通过逐步分析和调整参数,建立交易系统，经过 各种组合的多次数据回测，对比采用不同统计模型后，发现的确能增加交易系统的收 益，并且在两年多的模拟交易时段里明显战胜了比较基准上证50指数。 本文的研究，主要是基于DCC-GARCH模型和有效前沿分析等理论基础上，通过程 序量化过程构建交易系统，对股票资产动态配置方向上进行量化实践和测试,来验证 统计理论的有效性, 并可以为FOF基金、大类资产配置、期权期货风险对冲组合以及 私募股票投资等资产的动态配置提供参考。 关键词：动态条件相关性模型，广义自回归条件异方差模型，有效前沿，资产配置， 量化交易 II Dynamic Asset Allocation Of Stock Market – A Study Based On DCC GARCH And Effective Frontier Abstract This paper first reviews the relevant literatures on portfolio theory and DCC-GARCH model at home and abroad. Then, from the perspective of program quantification, the paper verifies whether the AR-GARCH model, the DCC-GARCH model and the effective frontier analysis in the portfolio theory can reflect the increase of investment returns in the data regression by implementing the program quantification trading system on the third-party platform. The research target is the daily close price data of 50 stocks in the Shanghai Stock 50 index. By establishing the GARCH model based on the BIAS index, the average bias of the stock closing price to the average line is added on the original average line, and the traditional double-average trading system is improved，by trading stocks when the quick average line cross the slow average line. The trading system hold at most 5 stocks for easy comparing. The DCC-GARCH model is used to predict the next day's fluctuation and dynamic correlation coefficients, and then Monte Carlo method is used to find the optimal configuration parameters on the effective frontier boundary. Through the gradual establishment of the trading system, after many data tests and adjustments of parameters, compared with the use of different statistical models, it does increase the returns of the trading system, and in more than two years of trading period obviously beat the benchmark Shanghai 50 index. Based on the theory of DCC-GARCH model and effective frontier analysis, this paper constructs a trading system through program quantification process, which is a beneficial attempt to the dynamic allocation of stock assets.。。。。。。以下内容略