基于深度学习的高频交易金融数据的波动率预测  

Volatility prediction of high-frequency trading financial data based on deep learning

作　　者：朱峰 郭文静[1] 阎希平 ZHU Feng;GUO Wenjing;YAN Xiping(College of Computer Science and Technology,Donghua University,Shanghai 201620,China;Shanghai Zhaoqian Investment Co.,Ltd,Shanghai 201107,China)

机构地区：[1]东华大学计算机科学与技术学院,上海201620 [2]上海兆前投资有限公司,上海201107

出　　处：《智能计算机与应用》2024年第9期82-87,共6页Intelligent Computer and Applications

摘　　要：随着信息化技术的发展,许多在线交易平台都可以提供高频的实时交易数据,为基于大数据的高频交易数据的波动率研究提供了基础。使用机器学习和深度学习算法分析大量的交易数据,建立波动率预测模型,可以帮助投资者更好地把握市场风险和机会,但金融高频交易数据存在大量噪声和非平稳性,导致模型的预测效果不佳。针对以上问题,本文构建了基于降噪自动编码器和不稳定注意力机制的深度学习模型,并利用该模型对高频交易数据波动率预测。实验结果表明该模型相较于常用的机器学习和深度学习方法拥有更准确的预测效果。With the development of information technology,many online trading platforms can provide high-frequency real-time trading data,which provides a basis for the study of volatility based on high-frequency trading data of big data.Many scholars use machine learning and deep learning algorithms to analyze a large amount of trading data to establish volatility forecasting models,helping investors to better grasp market risks and opportunities.However,financial high-frequency trading data has a lot of noise and non-stationarity,which leads to poor predictive performance of models.To address these issues,This paper constructs a deep learning model based on noise reduction autoencoder and unstable attention mechanism,and uses this model to predict the volatility of high-frequency trading data.The experimental results show that this model has more accurate predictive performance than commonly used machine learning and deep learning methods.

关 键 词：降噪自动编码 不稳定注意 波动率预测 

分 类 号：TP301[自动化与计算机技术—计算机系统结构]