中文摘要：

为克服经典Markowitz均值一方差模型的不足，考虑了投资者对风险认知的不同理解以及最小交易量、交易费用、最大投资上限等实际因素，提出了符合我国国情的用VaR度量风险的资产配置优化模型，并设计了一种新兴的自适应遗传算法来求解该模型。通过Matlab语言编写的求解程序进行实例测试，可得到不同风险情况下的多组输出结果。实际算例表明，所提出的模型和算法均是合理、有效的。

英文摘要：

In order to overcome shortcomings of the classical Markowitz mean variance model, this paper proposes an asset allocation optimization model which applies VaR to measurement of the risk and consists with the condition of the Chinese stock markets. In this model, the paper takes into consideration the investors different comprehension of the risk, the least quantities of buys, transaction cost and the investment limited. The solution of the model is obtained by designing an adaptive genetic algorithm with Matlab language. Finally, this solution is tested by real data and many favorable outcomes are resulted. Empirical results show that both the proposed model and the algorithm are reasonable and efficient.