基于已实现波动率的ARFIMA模型在股指期货高频数据中的实证研究

An Empirical Study on ARFIMA Model Based on Realized Volatility in High Frequency Data of Stock Index Futures

本文利用连续的股指期货指数5分钟高频数据，基于“已实现”波动率进行实证研究，结果表明股指期货“已实现”波动率序列的分布是非正态分布且具有长记忆性，最后建立ARFIMA模型，并对波动率进行预测，预测平均误差7．20％。

This paper uses the continuous stock index futures index of 5 minutes high frequency data, based on the ＂realized＂ volatility of the empirical study, the results show that stock index futures ＂realized＂ volatility sequence distribution is non-normal distribution and long memory, and finally The ARFIMA model is established and the volatility is predicted to predict the average error of 7.20%.