基于ARFIMA模型的波动率预测及交易策略研究

发布时间：2018-01-05 08:07

本文关键词：基于ARFIMA模型的波动率预测及交易策略研究　出处：《苏州大学》2013年硕士论文　论文类型：学位论文

更多相关文章： 波动率 长记忆性 ARFIMA 交易策略

【摘要】：随着我国金融创新的不断深入，市场上对于股票期权产品的需求呼声越来越高。传统的Black-Scholes定价公式中，我们一般假设波动率是已知并且固定不变的，但是这并不符合市场的客观情况。因此，前人曾运用ARCH Model、GARCH Model、SV Model等对股票市场波动率进行刻画，，然而，通过研究我们发现，对于新兴市场国家，股票波动率是具有长记忆性的，而上述模型并不能很好刻画这一特点。基于上述原因，我们引入时间序列理论体系中一个新领域ARFIMA模型，以此刻画波动率序列的长记忆性，并对其进行预测和交易策略研究。本文主要研究了基于我国股票市场的波动率的长记忆性，并运用ARFIMA模型对长记忆性序列进行建模及预测，同时基于所预测波动率序列对期权交易策略进行研究。具体可以分为以下几个方面：首先，研究了股票市场波动率的度量方法。给出了“已实现”波动率可以作为波动率的无偏有效估计的结论，并通过二次移动平均消除了“噪声”，得出1天高频数据取样频率为5分钟时最佳。其次，对时间序列ARFIMA模型进行了探讨。比较了LW方法较R/S方法、修正R/S方法、GPH方法的优势，借助于Matlab软件计算出d值，确定模型的参数，并运用三种方法对未来波动率进行预测。最后，研究了基于波动率的期权交易策略。以江铜CWB1为例，提出了波动率异常的识别方法及综合运用，并分别详细介绍了买入波动率策略和卖出波动率策略。
[Abstract]:With the deepening of financial innovation in China, the demand for stock option products in the market is increasing. Traditional Black-Scholes pricing formula. We generally assume that volatility is known and fixed, but this does not conform to the objective situation of the market. Therefore, the previous use of the ARCH Model GARCH Model. SV Model and others depict the volatility of stock market. However, we find that the volatility of stock market has a long memory for emerging market countries. Because of the above reasons, we introduce a new domain ARFIMA model in the system of time series theory to describe the long memory of volatility series. And carries on the forecast and the trading strategy research to it. This paper mainly studies the long memory of volatility based on the stock market in China, and uses ARFIMA model to model and predict the long memory sequence. At the same time, based on the predicted volatility series to study the options trading strategy. The specific can be divided into the following aspects: Firstly, the paper studies the measurement method of volatility in stock market, and gives the conclusion that "realized" volatility can be used as an unbiased efficient estimate of volatility, and eliminates the "noise" by the quadratic moving average. The best sampling frequency of 1 day high frequency data is 5 minutes. Secondly, the time series ARFIMA model is discussed, and the advantage of LW method is compared with that of the R / S method and the modified R / S method. With the help of Matlab software, the d value is calculated, the parameters of the model are determined, and three methods are used to predict the volatility in the future. Finally, the option trading strategy based on volatility is studied. Taking Jiang Copper CWB1 as an example, the identification method of volatility anomaly and its comprehensive application are put forward. And the buy volatility strategy and the sell volatility strategy are introduced in detail.
【学位授予单位】：苏州大学
【学位级别】：硕士
【学位授予年份】：2013
【分类号】：F832.51;F224

【参考文献】