基于机器学习方法预测股市的系统性风险

发布时间：2018-03-03 00:22

本文选题：股票市场系统性风险　切入点：新的K线数据　出处：《天津工业大学》2017年硕士论文　论文类型：学位论文

【摘要】：股票市场的系统性风险作为金融系统的主要风险之一,长期受到人们的关注和研究,而目前我国的股票市场风险状况仍然不容乐观,影响股票市场的系统性风险依然存在。本文针对股票市场系统性风险做了三方面的工作。首先通过标准化原始数据、在计算出价格上升期间所需成交量后,对所需成交量以及价格进行拟合,得到系统性风险发生前交易量和价格的关系;其次,将标准化的数据以周、月为单位进行合并创造出新的具有统计意义的K线数据K1,并对K线数据K1进行拟合,得到早于原始数据价格见顶的K线数据K2;第三,本文又进一步根据K线数据K1和K2创造出12个特征,并使用SVR和Liner Regression方法对特征数据进行拟合,获得可以预测未来一年价格增长率的线性拟合函数。通过真实的股票数据验证表明,这三方面的工作都是有效的,且有助于帮助广大投资者了解股票市场的变化情况,本文的工作对于预测和防范股票市场的系统性风险也提供了一种可行的方法。
[Abstract]:As one of the main risks in the financial system, the systemic risk of stock market has been paid attention to and studied by people for a long time. However, the situation of stock market risk in our country is still not optimistic. The systemic risk that affects the stock market still exists. This paper has done three works on the systemic risk of the stock market. Firstly, through the standardized raw data, after calculating the transaction volume required during the period of price rise, The required volume and price are fitted to obtain the relationship between transaction volume and price before systemic risk occurs. Secondly, the standardized data are weekly. Month as the unit to create a new statistical significance of K line data K1, and K line data K1 fitting, before the original data price peaked K-line data K2; third, In this paper, we further create 12 features according to K line data K1 and K2, and use SVR and Liner Regression methods to fit the feature data. To obtain a linear fitting function that can predict the rate of price growth in the coming year. Real stock data verify that these three aspects of the work are effective and help investors understand the changes in the stock market. The work of this paper also provides a feasible method to predict and prevent the systematic risk of stock market.
【学位授予单位】：天津工业大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：F832.51;TP181

【参考文献】