中文摘要：

支持向量机（support vector machine，SVM）方法是建立在统计学习理论的VC（Vapnik-Chervonenkis）维理论和结构风险最小原理基础上的，根据有限的样本信息在模型的复杂性（对特定训练样本的学习精度）和学习能力（无错误地识别任意样本的能力）之间寻求最佳折衷，以求获得最好的推广能力．基本原理是，以二维数据为例，如果训练数据分布在二维平面上的点，它们按照其分类聚集在不同的区域．通过训练，找到这些分类之间的边界．利用SVM方法，针对上海期货交易所挂牌交易的2011年期货铜主力合约500ms每tick的高频数据进行分析．分析结果表明，SVM方法可以取得较好的预测效果．

英文摘要：

Support vector machine （SVM） method is based on the statistical learning theory of Vapnik-Chervonenkis （VC） dimension theory and the structure risk minimum principle, on the basis of information by the limited sample in the com- plexity of the model （the specific learning accuracy of training samples） and learn- ing ability （not mistakenly＇s ability to identify random sample） to seek the best compromise in between, in order to get the best generalization ability. In the case of two dimensional data, if the training data distribution in the two-dimensional plane are points, they are classified according to be gathered in different regions. Through training, we find the boundary between these regions of classifications. In this paper, using the SVM method, in view of the Shanghai futures exchange traded in 2011 main copper futures contracts 500 ms tick high-frequency data analysis, the results show that compared has better prediction ability.