生存分析在股市期市涨跌预测中的应用

An Application of Survival Analysis in Forecasting Changes of Stock Index Futures

运用生存分析（Survival Analysis）方法和台湾2009-2011年每日发布的公开数据，建立Cox回归模型（Cox Regression Model），预测台湾隔日加权股价指数期货涨跌。实证分析显示：涨幅模型预测结果的正确率为74．47％，绩效验证中平均收益为35．73点；跌幅模型预测结果的正确率为75．32％，绩效验证中平均收益为37．99点；将涨跌幅模型综合在一起考虑时，在绩效验证中可知平均收益可达38．68点。

In this article, we apply the Cox Regression Model to predict the changes of Taiwan Stock Price Index Futures by using the daily public information provided by mass media. The sample period covers from 2009 to 2011. Empirical results indicate that： The accuracy rate of predicting the rising change is 74.47% and the return in the performance of verification is 35.73; The accuracy rate of predicting the falling change is 75.32%and the return in the performance of verification is 37. 99; overall, the return in the performance of verification is up to 38. 68.