基于生成对抗网络的高频算法及其回测研究

doi:10.3969/j.issn.0253-2778.2020.06.012

中国科学技术大学学报 ›› 2020, Vol. 50 ›› Issue (6): 801-810.DOI: 10.3969/j.issn.0253-2778.2020.06.012

基于生成对抗网络的高频算法及其回测研究

孟徐然，毕秀春，张曙光

1.中国科学技术大学管理学院,安徽合肥 230026;2.贵州财经大学数学与统计学院，贵州贵阳 550025

收稿日期:2020-03-06 修回日期:2020-06-21 接受日期:2020-06-21 出版日期:2020-06-30 发布日期:2020-06-21
通讯作者: 毕秀春
作者简介:孟徐然，男，1997年生，硕士.研究方向：量化金融.E-mail: mmxxrr@mail.usct.edu.cn
基金资助:
国家自然科学基金(11471304),贵州财经大学科研项目（2020YJ021，2020YJ026）资助.

High frequency algorithm and its back-testing results based on GAN

MENG Xuran, BI Xiuchun, ZHANG Shuguang

1.School of Management, University of Science and Technology of China, Hefei 230026, China; 2.School of Mathematics and Statistics, Guizhou University of Finance and Economics, Guiyang 550025, China

Received:2020-03-06 Revised:2020-06-21 Accepted:2020-06-21 Online:2020-06-30 Published:2020-06-21

摘要/Abstract

摘要： 在金融工程的分类任务中，由于金融数据噪音大、信息比率低的特点,传统深度算法的有监督训练模式往往过于依赖数据本身的绝对标签从而进一步放大了噪音对最终结果的影响.生成对抗网络(generative adversarial network，GAN)能够利用深度网络挖取数据特征,增强数据,减少噪音影响,应用于金融序列分析时效果优异.这里将GAN模型用于高频交易,具体做法为: 将数据按波动性分为有标签数据与无标签数据两类,利用生成网络G与判别网络D互相对抗训练来深度学习这些数据的内在特性,训练好后的D网络根据有标签数据信息亦可对真实数据进行类别判别,得到涨跌分类模型,进而给出量化交易策略.基于期货主力合约数据进行了实证分析,结果表明,基于GAN训练下的LSTM模型显著优于有监督训练下的LSTM等深度模型和Logistics回归模型.

关键词: 深度学习, 生成对抗网络, 涨跌分类模型, 量化策略

Abstract: In the financial classification mission, due to the big noise and low information-ratio in financial data, traditional supervised-learning regime may extend the noise influence because of the over dependent on the data label. GAN(generative adversarial network) can learn the data characters and reduce the influence of noise. When it is used to analyze the financial data, it has great results. We apply GAN to the high frequency trading: set the data labeled or unlabeled based on its volatility, then use the adversarial training between generative network G and discriminative network D to learn the intrinsic characters of the data, finally use the well trained D to get the up and down classification model and the quantization strategy. The sample is based on the future data, and the final results show that the LSTM model training by GAN is better than the deep learning models such as LSTM with supervised training and the Logistic regression model.

Key words: deep learning, generative adversarial network, up and down classification model, quantization

中图分类号:

F830.9

孟徐然，毕秀春，张曙光. 基于生成对抗网络的高频算法及其回测研究[J]. 中国科学技术大学学报, 2020, 50(6): 801-810.

MENG Xuran, BI Xiuchun, ZHANG Shuguang. High frequency algorithm and its back-testing results based on GAN[J]. Journal of University of Science and Technology of China, 2020, 50(6): 801-810.

参考文献

［1］王锴. 主成分Logistic回归模型在国债期货跨品种套利中的应用[N]. 期货日报, 2020-06-15.
[2] ALDRIDGE I. High-Frequency Trading: A Practical Guide to Algorithmic Strategies and Trading Systems[M]. Hoboken, NJ: Wiley, 2010.
[3] BROGAARD J, HENDERSHOTT T, RIORDANR R. High frequency trading and price discovery[J]. The Review of Financial Studies, 2014, 27(8): 2267-2306.
[4] ANGEL J, MCCABE D. Fairness in financial markets: The case of high frequency trading[J]. Journal of Business Ethis, 2013, 112: 585-595.
[5] HINTON G E, OSINDERO S, TEH Y W. A fast learning algorithm for deep belief nets[J]. Neural Computation, 2006, 18: 1527-1554.
[6] GOODFELLOW I, BENGIO Y, COURVILLE A. Deep Learning[M]. Cambridge, MA: The MIT Press, 2017.
[7] WANG M, DENG W. Deep face recognition[C]// [2020-02-10]. https://arxiv.org/abs/1804.06655.
[8] KARPATHY A, TODERICI G, SHETTY S, et al. Large-scale video classification with convolutional neural networks[C]// Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition. IEEE, 2014: 1725-1732.
[9] DING Y, LIU Yang, LUAN H, et al. Visualizing and understanding neural machine translation[C]// Proceedings of the 55th Annual Meeting of the Association for Computational Linguistics. Vancouver, Canada: ACL, 2017: 1150-1159.
[10] QIAN Y, BI M, TAN T,et al. Very deep convolutional neural networks for noise robust speech recognition[C]// IEEE/ACM Transactions on Audio, Speech, and Language Processing. IEEE, 2016, 42(12): 2263-2276.
[11] GOODFELLOW I, POUGET-ABADIE J, MIRZA M, et al. Generative adversarial nets[C]// Advances in Neural Information Processing Systems 27. ACM, 2014.
[12] SALIMANS T, GOODFELLOW I, ZAREMBA W, et al. Improved techniques for training GANs[C]// Advances in Neural Information Processing Systems 29. ACM, 2016: 2234-2242.
[13] YANG J, KANNAN A, BATRA D, et al. LR-GAN: Layered recursive generative adversarial networks for image generation[DB/OL]. [2020-02-10]. https://arxiv.org/abs/1703.01560.
[14] HOANG Q, NGUYEN T D, LE T, et al. MGAN: Training generative adversarial nets with multiple generators[C]// 6th International Conference on Learning Representations. La Jolla, CA: International Conference on Representation Learning, 2018.
[15] PERSIO L D, HONCHAR O. Artificial neural networks architectures for stock price prediction: Comparisons and applications[J]. International Journal of Circuits, Systems and Signal Processing, 2016, 10: 403-413.
[16] 龙奥明, 毕秀春, 张曙光. 基于LSTM 神经网络的黑色金属期货套利策略模型[J]．中国科学技术大学学报, 2018, 48(2): 125-132.
[17] 杨青, 王晨蔚. 基于深度学习LSTM神经网络的全球股票指数预测研究[J]. 统计研究, 2019, 36(3): 67-79.
[18] 孙达昌, 毕秀春. 基于深度学习算法的高频交易策略及其盈利能力[J]. 中国科学技术大学学报, 2018, 48(11): 58-67.
[19] LAKSHMINARAYANAN S K, MCCRAE J. A comparative study of SVM and LSTM deep learning algorithms for stock prediction[DB/OL]. [2020-02-10]. http://CEURWS.org/Vol-2563/aics 41.pdf.
[20] DEV S, WESLEY C, FARHANA H Z. A comparative study of LSTM and DNN for stock market forecasting[C]// 2018 IEEE International Conference on Big Data. IEEE, 2018.
[21] ARVAND F, SHERIDAN H. Deep learning for the prediction of stock market trends[C]// 2019 IEEE International Conference on Big Data. IEEE, 2019.
[22] 陆广泉, 谢扬才, 刘星,等.一种基于KNN的半监督分类改进算法[J].广西师范大学学报（自然科学版）, 2012, 30 (1): 48-52.
[23] 刘蓉. 半监督学习的Co-training算法研究[J]. 电脑编程技巧与维护, 2010(14): 6-7.

[24] KINGMA D P, BA J. Adam: A method for stochastic optimization[DB/OL]. [2020-02-10]. https://arxiv.org/abs/1412.6980.

()
()

[1]	陶陶，柏建树，刘恒，侯书东，郑啸. 基于WGAN反馈的深度学习差分隐私保护方法[J]. 中国科学技术大学学报, 2020, 50(8): 1064-1071.
[2]	白江波，杨阳，张文生. 基于改进CycleGAN的合成孔径雷达图像仿真[J]. 中国科学技术大学学报, 2020, 50(8): 1181-1186.
[3]	王悦，李京. 基于可视化的卷积神经网络优化方法研究[J]. 中国科学技术大学学报, 2020, 50(7): 959-967.
[4]	崔文泉，余厚莹，侯晓天. 不均衡数据情形的基于聚焦损失的CGAN的集成分类方法[J]. 中国科学技术大学学报, 2020, 50(7): 968-976.
[5]	杜淑颖，杜鹏，丁世飞. 基于CNN的假冒域名识别方法研究[J]. 中国科学技术大学学报, 2020, 50(7): 1019-1025.
[6]	谷睿哲, 汪平, 郑梦策, 胡红钢, 俞能海. 基于对抗攻击的侧信道防护方案[J]. 中国科学技术大学学报, 2020, 50(10): 1343-1358.
[7]	杜炤鑫, 谢海宁, 宋杰, 周德生, 邹晓峰, 陈冉, 曾平. 基于图像处理和深度学习的配网跳闸故障识别方法[J]. 中国科学技术大学学报, 2020, 50(1): 39-48.
[8]	杨子文，陈蕾，浦建宇. 基于两层迁移卷积神经网络的抽象图像情感识别[J]. 中国科学技术大学学报, 2019, 49(1): 40-48.
[9]	孙达昌，毕秀春. 基于深度学习算法的高频交易策略及其盈利能力[J]. 中国科学技术大学学报, 2018, 48(11): 923-932.
[10]	陈东杰，张文生，杨阳. 基于深度学习的高铁接触网定位器检测与识别[J]. 中国科学技术大学学报, 2017, 47(4): 320-327.

基于生成对抗网络的高频算法及其回测研究

High frequency algorithm and its back-testing results based on GAN

PDF (PC)

可视化

摘要/Abstract

引用本文

使用本文

参考文献

相关文章 10

编辑推荐

Metrics

本文评价