摘要
目前，我国各大基金公司、券商等金融企业纷纷研发高频交易系统与交易策略。但对金融市场高频交易价格趋势的研究均不够成熟，金融市场高频交易的特征的规律还尚待深入挖掘。本文结合期货市场的实际高频数据，将价格趋势分为上涨、下跌和稳定三类，从而将期货价格的预测研究转化为分类问题研究，将支持向量机模型。通过对黄金期货市场的高频数据进行实证研究，分析将机器学习算法用于价格预测研究的效果。研究发现支持向量机算法可以较好地预测结果。
Currently, many fund companies, security companies and other financial enter- prises have developed high frequency trading system and strategies. However, the research on the high frequency transaction price trend of the financial market is not mature e- nough, and the characteristics of the high frequency transaction in the financial market still need to be studied. Based on the actual high-frequency data of futures market, the research divide price trend into three categories： rise, fall and stability, and then trans- form the prediction of future price into the classification problem. Through the empirical study on the high frequency data of the gold futures market, study the effect of the ma- chine learning algorithm in the price prediction. It is found that the support vector ma- chine （SVM） algorithm has better prediction results and regularity.
出处
《价格理论与实践》
CSSCI
北大核心
2018年第1期114-117,共4页
Price:Theory & Practice
关键词
随机森林
支持向量机
高频交易
价格趋势
Random Forest
Support Vector Machine
High Frequency
Price Trend
