利用RAVI等5种趋向技术指标将股票价格时间序列映射到高维特征空间,建构了支持向量机分类器,在不同趋势中选择相适应的不同的交易策略。为获得风险与收益的帕累托最优解,运用NSGA—Ⅱ算法对MA策略和KD策略进行了参数优化。经过测试,所建立的交易策略在上证指数2000至2009年中取得明显的收益,远远高于简单持有策略。
The time series of stock price are reflected into high dimensional character space, using RAVI and four other technical indicators of tendency,SVM classifier is constructed to assort trends. Furthermore, trade decisions can be chosen responding to different kinds of trends respectively. In order to obtain the Pareto optimal solution be- tween risk and earnings, NSGA - Ⅱ is utilized to optimize parameters of both the MA and the KD strategy. 2000 2009 SSE composite index data is used in empirical study, and the profit gained from the trading strategy constructed above is much higher than the counterpart which owns a simple buy and hold investment scenario.