AI帮你理解科学

AI 生成解读视频

AI抽取解析论文重点内容自动生成视频


pub
生成解读视频

AI 溯源

AI解析本论文相关学术脉络


Master Reading Tree
生成 溯源树

AI 精读

AI抽取本论文的概要总结


微博一下
We utilized a 2-D Deep Convolutional Neural Network model to be used with financial stock market data and technical analysis indicators for developing an algorithmic trading system

Algorithmic financial trading with deep convolutional neural networks: Time series to image conversion approach.

Applied Soft Computing, (2018): 525-538

被引用51|浏览387
EI
下载 PDF 全文
引用
微博一下

摘要

•We converted 1-D financial technical analysis data to 2-D images for classification.•We used 2-D deep convolutional neural network for trend forecasting.•We propose a robust algorithmic trading model that works in any market condition.•To best of our knowledge, 2-D CNN with TA has not been used for financial trading before.•Model outperf...更多

代码

数据

0
简介
  • Stock market forecasting based on computational intelligence models have been part of stock trading systems for the last few decades.
  • As a result, trading systems based on autonomous intelligent decision making models are getting more attention in various different financial markets globally [1].
  • Deep learning based prediction/classification models started emerging as the best performance achievers in various applications, outperforming classical computational intelligence methods like SVM.
  • The application of deep neural networks on financial forecasting models have been very limited
重点内容
  • Stock market forecasting based on computational intelligence models have been part of stock trading systems for the last few decades
  • Deep learning based prediction/classification models started emerging as the best performance achievers in various applications, outperforming classical computational intelligence methods like support vector machines (SVM)
  • Image processing and vision based problems dominate the type of applications that these deep learning models outperform the other techniques [2]
  • We propose a novel approach that converts 1-D financial time series into a 2-D image-like data representation in order to be able to utilize the power of deep convolutional neural network for an algorithmic trading system
  • We utilized a 2-D Deep Convolutional Neural Network model to be used with financial stock market data and technical analysis indicators for developing an algorithmic trading system
方法
  • For the algorithmic trading model, the authors propose a novel method that uses CNN to determine the “Buy” and “Sell” points in stock prices using 15 different technical indicators with different time intervals and parameter selections for each daily stock price time series to create images.
  • As can be seen in Fig. 2, the proposed method is divided into five main steps: dataset extract/transform, labeling data, image creation, CNN analysis and financial evaluation phases.
  • Dataset extract/transform phase, the downloaded prices are normalized according to the adjusted close prices
结果
  • The results indicate CNN-TA trading performance is significantly better than all models over the long run (2007–2017) for both Dow30 stocks and ETFs.
结论
  • The authors utilized a 2-D Deep Convolutional Neural Network model to be used with financial stock market data and technical analysis indicators for developing an algorithmic trading system.
  • The authors used Dow Jones 30 stock prices and ETFs as the financial time series data.
  • The results indicate this novel approach performs very well against Buy & Hold and other models over long periods of out-of-sample test periods.
  • The authors will analyze the correlations between selected indicators in order to create more meaningful images so that the learning models can better associate the Buy–Sell–Hold signals and come up with more profitable trading models
表格
  • Table1: Selected ETFs and their descriptions
  • Table2: Confusion matrix of test data (Dow-30)
  • Table3: Evaluation of test data (Dow-30)
  • Table4: Confusion matrix of test data (ETFs)
  • Table5: Evaluation of test data (ETFs)
  • Table6: Comparison of annualized returns of the proposed system (CNN-TA) with BaH, RSI, SMA, LSTM, MLP Reg. models (ETFs – test period: 2007–2017)
  • Table7: Comparison of annualized returns of the proposed system (CNN-TA) with BaH, RSI, SMA, LSTM, MLP Reg. models (ETFs test period: 2007–2012)
  • Table8: Comparison of annualized returns of the proposed system (CNN-TA) with BaH, RSI, SMA, LSTM, MLP Reg. models (DOW30 – test period: 2007–2017)
  • Table9: Comparison of annualized returns of the proposed system (CNN-TA) with BaH, RSI, SMA,LSTM, MLP Reg. Models (DOW30 – test period: 2007–2012)
  • Table10: TTest results and average results of the proposed CNN-TA model for Dow30
  • Table11: TTest results and average results of the proposed CNN-TA model for ETFs
  • Table12: TTest results of annualized return of Dow30 stocks
  • Table13: TTest results of annualized return of ETFs
Download tables as Excel
相关工作
  • 2.1. Time series data analytics

    In literature, there are different adapted methodologies for time series data analysis. These can be listed as follows: statistical and mathematical analysis, signal processing, extracting features, pattern recognition, and machine learning. Statistical and mathematical analysis in time series data can be achieved through determining the mathematical parameters such as maximum, minimum, average, moving average, variance, covariance, standard deviation, autocorrelation, crosscorrelation and convolution in the sliding window [6]. Curve fitting, regression analysis, autoregressive moving average (ARMA), autoregressive integrated moving average (ARIMA), Bayesian analysis, Kalman filter methods are the mathematical methods that are generally used to analyze and forecast time series data in literature [7]. In addition, signal processing methods such as Fourier and wavelet transforms are used to analyze the time series data. Discrete Fourier transform (DFT), discrete wavelet transform (DWT), piecewise aggregate approximation (PAA) are also used to analyze time series data to extract features and find the similarities within the data [8]. Unlike traditional approaches, machine learning models are also used in analyzing time series data and predictions. Machine learning algorithms that are mostly used in time series data analytics are listed as follows: clustering algorithms [9], hidden Markov models [10], support vector machines (SVM) [11,12,13], artificial neural networks (ANNs) [14,15,16,17] self organizing maps (SOM) [18,19,20].
基金
  • This study was funded by The Scientific and Technological Research Council of Turkey (TUBITAK) under grant number 215E248
引用论文
  • R.C. Cavalcante, R.C. Brasileiro, V.L. Souza, J.P. Nobrega, A.L. Oliveira, Computational intelligence and financial markets: a survey and future directions, Expert Syst. Appl. 55 (2016) 194–211.
    Google ScholarLocate open access versionFindings
  • A. Canziani, A. Paszke, E. Culurciello, An Analysis of Deep Neural Network Models for Practical Applications, 2016 arXiv:1605.07678.
    Findings
  • C. Krauss, X.A. Do, N. Huck, Deep neural networks, gradient-boosted trees, random forests: statistical arbitrage on the s&p 500, Eur. J. Oper. Res. 259 (2) (2017) 689–702.
    Google ScholarLocate open access versionFindings
  • J.-F. Chen, W.-L. Chen, C.-P. Huang, S.-H. Huang, A.-P. Chen, Financial time-series data analysis using deep convolutional neural networks, in: 2016 7th International Conference on Cloud Computing and Big Data (CCBD), IEEE, 2016, pp. 87–92.
    Google ScholarLocate open access versionFindings
  • T. Fischer, C. Krauß, Deep Learning with Long Short-term Memory Networks for Financial Market Predictions, Tech. Rep., FAU Discussion Papers in Economics, 2017.
    Google ScholarLocate open access versionFindings
  • F. Ganz, D. Puschmann, P. Barnaghi, F. Carrez, A practical evaluation of information processing and abstraction techniques for the internet of things, IEEE Internet Things J. 2 (4) (2015) 340–354.
    Google ScholarLocate open access versionFindings
  • G.E. Box, G.M. Jenkins, G.C. Reinsel, G.M. Ljung, Time Series Analysis: Forecasting and Control, John Wiley & Sons, 2015.
    Google ScholarFindings
  • J.D. Hamilton, Time Series Analysis, vol. 2, Princeton University Press, Princeton, 1994.
    Google ScholarLocate open access versionFindings
  • G. Das, K.-I. Lin, H. Mannila, G. Renganathan, P. Smyth, Rule discovery from time series, AAAI (1998).
    Google ScholarLocate open access versionFindings
  • M. Ramoni, P. Sebastiani, P. Cohen, Bayesian clustering by dynamics, Mach. Learn. 47 (1) (2002) 91–121.
    Google ScholarLocate open access versionFindings
  • L. Cao, Support vector machines experts for time series forecasting, Neurocomputing 51 (2003) 321–339.
    Google ScholarFindings
  • N.K. Ahmed, A.F. Atiya, N.E. Gayar, H. El-Shishiny, An empirical comparison of machine learning models for time series forecasting, Econometr. Rev. 29 (5–6) (2010) 594–621.
    Google ScholarLocate open access versionFindings
  • M. Mohandes, T. Halawani, S. Rehman, A.A. Hussain, Support vector machines for wind speed prediction, Renew. Energy 29 (6) (2004) 939–947.
    Google ScholarLocate open access versionFindings
  • C.M. Arizmendi, J.R. Sanchez, N.E. Ramos, G.I. Ramos, Time series predictions with neural nets: application to airborne pollen forecasting, Int. J. Biometeorol. 37 (3) (1993) 139–144.
    Google ScholarLocate open access versionFindings
  • D. Srinivasan, A. Liew, C. Chang, A neural network short-term load forecaster, Electr. Power Syst. Res. 28 (3) (1994) 227–234.
    Google ScholarLocate open access versionFindings
  • I. Kaastra, M. Boyd, Designing a neural network for forecasting financial and economic time series, Neurocomputing 10 (3) (1996) 215–236.
    Google ScholarLocate open access versionFindings
  • K. Kalaitzakis, G. Stavrakakis, E. Anagnostakis, Short-term load forecasting based on artificial neural networks parallel implementation, Electr. Power Syst. Res. 63 (3) (2002) 185–196.
    Google ScholarLocate open access versionFindings
  • T. Kohonen, The self-organizing map, Neurocomputing 21 (1-3) (1998) 1–6.
    Google ScholarFindings
  • F. Mörchen, A. Ultsch, O. Hoos, Extracting interpretable muscle activation patterns with time series knowledge mining, Int. J. Knowl-Based Intell. Eng. Syst. 9 (3) (2005) 197–208.
    Google ScholarLocate open access versionFindings
  • S.-C. Kuo, S.-T. Li, Y.-C. Cheng, M.-H. Ho, Knowledge discovery with SOM networks in financial investment strategy, in: Fourth International Conference on Hybrid Intelligent Systems (HIS’04), IEEE, 2004, pp. 98–103.
    Google ScholarLocate open access versionFindings
  • A. Bezerianos, S. Papadimitriou, D. Alexopoulos, Radial basis function neural networks for the characterization of heart rate variability dynamics, Artif. Intell. Med. 15 (3) (1999) 215–234.
    Google ScholarLocate open access versionFindings
  • Q. Li, D. Liu, J. Fang, A. Jeary, C. Wong, Damping in buildings: its neural network model and AR model, Eng. Struct. 22 (9) (2000) 1216–1223.
    Google ScholarLocate open access versionFindings
  • D. Guan, W. Yuan, S.J. Cho, A. Gavrilov, Y.-K. Lee, S. Lee, Devising a context selection-based reasoning engine for context-aware ubiquitous computing middleware, in: Ubiquitous Intelligence and Computing, Springer Berlin Heidelberg, Berlin, Heidelberg, 2007, pp. 849–857.
    Google ScholarFindings
  • J. Choi, D. Shin, D. Shin, Research and implementation of the context-aware middleware for controlling home appliances, IEEE Trans. Consumer Electron. 51 (1) (2005) 301–306.
    Google ScholarLocate open access versionFindings
  • N. An, W. Zhao, J. Wang, D. Shang, E. Zhao, Using multi-output feedforward neural network with empirical mode decomposition based signal filtering for electricity demand forecasting, Energy 49 (2013) 279–288.
    Google ScholarLocate open access versionFindings
  • B. Krollner, B. Vanstone, G. Finnie, Financial Time Series Forecasting with Machine Learning Techniques: A Survey, 2010.
    Google ScholarLocate open access versionFindings
  • J.-Z. Wang, J.-J. Wang, Z.-G. Zhang, S.-P. Guo, Forecasting stock indices with back propagation neural network, Expert Syst. Appl. 38 (11) (2011) 14346–14355.
    Google ScholarLocate open access versionFindings
  • Z. Liao, J. Wang, Forecasting model of global stock index by stochastic time effective neural network, Expert Syst. Appl. 37 (1) (2010) 834–841.
    Google ScholarLocate open access versionFindings
  • A.-S. Chen, M.T. Leung, H. Daouk, Application of neural networks to an emerging financial market: forecasting and trading the Taiwan stock index, Comput. Oper. Res. 30 (6) (2003) 901–923.
    Google ScholarLocate open access versionFindings
  • E. Guresen, G. Kayakutlu, T.U. Daim, Using artificial neural network models in stock market index prediction, Expert Syst. Appl. 38 (8) (2011) 10389–10397.
    Google ScholarLocate open access versionFindings
  • O.B. Sezer, A.M. Ozbayoglu, E. Dogdu, An artificial neural network-based stock trading system using technical analysis and big data framework, in: Proceedings of the SouthEast Conference, ACM, 2017, pp. 223–226.
    Google ScholarLocate open access versionFindings
  • S. Dhar, T. Mukherjee, Performance evaluation of neural network approach in financial prediction: evidence from Indian Market, Proceedings of the International Conference on Communication and Computational Intelligence (2010).
    Google ScholarLocate open access versionFindings
  • B. Vanstone, G. Finnie, T. Hahn, Creating trading systems with fundamental variables and neural networks: the Aby case study, Math. Comput. Simul. 86 (2012) 78–91.
    Google ScholarLocate open access versionFindings
  • R. Aguilar-Rivera, M. Valenzuela-Rendón, J. Rodríguez-Ortiz, Genetic algorithms and Darwinian approaches in financial applications: a survey, Expert Syst. Appl. 42 (21) (2015) 7684–7697.
    Google ScholarLocate open access versionFindings
  • A.M. Ozbayoglu, U. Erkut, Stock market technical indicator optimization by genetic algorithms, in: Intelligent Engineering Systems through Artificial Neural Networks, vol. 20, ASME Press, 2010.
    Google ScholarLocate open access versionFindings
  • Y.-K. Kwon, B.-R. Moon, A hybrid neurogenetic approach for stock forecasting, IEEE Trans. Neural Netw. 18 (3) (2007) 851–864.
    Google ScholarLocate open access versionFindings
  • O.B. Sezer, M. Ozbayoglu, E. Dogdu, A deep neural-network based stock trading system based on evolutionary optimized technical analysis parameters, Proc. Comput. Sci. 114 (2017) 473–480.
    Google ScholarLocate open access versionFindings
  • C. Evans, K. Pappas, F. Xhafa, Utilizing artificial neural networks and genetic algorithms to build an algo-trading model for intra-day foreign exchange speculation, Math. Comput. Modell. 58 (5) (2013) 1249–1266.
    Google ScholarLocate open access versionFindings
  • C.-F. Huang, A hybrid stock selection model using genetic algorithms and support vector regression, Appl. Soft Comput. 12 (2) (2012) 807–818.
    Google ScholarLocate open access versionFindings
  • M. Pulido, P. Melin, O. Castillo, Particle swarm optimization of ensemble neural networks with fuzzy aggregation for time series prediction of the Mexican Stock Exchange, Inform. Sci. 280 (2014) 188–204.
    Google ScholarLocate open access versionFindings
  • J. Wang, R. Hou, C. Wang, L. Shen, Improved v-support vector regression model based on variable selection and brain storm optimization for stock price forecasting, Appl. Soft Comput. 49 (2016) 164–178.
    Google ScholarLocate open access versionFindings
  • S. Mabu, M. Obayashi, T. Kuremoto, Ensemble learning of rule-based evolutionary algorithm using multi-layer perceptron for supporting decisions in stock trading problems, Appl. Soft Comput. 36 (2015) 357–367.
    Google ScholarFindings
  • M. Ballings, D. Van den Poel, N. Hespeels, R. Gryp, Evaluating multiple classifiers for stock price direction prediction, Expert Syst. Appl. 42 (20) (2015) 7046–7056.
    Google ScholarLocate open access versionFindings
  • Y. LeCun, Y. Bengio, G. Hinton, Deep learning, Nature 521 (7553) (2015) 436–444.
    Google ScholarLocate open access versionFindings
  • A. Krizhevsky, I. Sutskever, G.E. Hinton, Imagenet classification with deep convolutional neural networks, Adv. Neural Inform. Process. Syst. (2012) 1097–1105.
    Google ScholarLocate open access versionFindings
  • A. Karpathy, G. Toderici, S. Shetty, T. Leung, R. Sukthankar, L. Fei-Fei, Large-scale video classification with convolutional neural networks, Proceedings of the IEEE conference on Computer Vision and Pattern Recognition (2014) 1725–1732.
    Google ScholarLocate open access versionFindings
  • S. Lawrence, C.L. Giles, A.C. Tsoi, A.D. Back, Face recognition: a convolutional neural-network approach, IEEE Trans. Neural Netw. 8 (1) (1997) 98–113.
    Google ScholarLocate open access versionFindings
  • D.C. Ciresan, U. Meier, L.M. Gambardella, J. Schmidhuber, Convolutional neural network committees for handwritten character classification, in: 2011 International Conference on Document Analysis and Recognition (ICDAR), IEEE, 2011, pp. 1135–1139.
    Google ScholarLocate open access versionFindings
  • Y. Kim, Convolutional Neural Networks for Sentence Classification, 2014 arXiv:1408.5882.
    Findings
  • N. Kalchbrenner, E. Grefenstette, P. Blunsom, A Convolutional Neural Network for Modelling Sentences, 2014 arXiv:1404.2188.
    Findings
  • X. Ding, Y. Zhang, T. Liu, J. Duan, Deep learning for event-driven stock prediction, IJCAI (2015) 2327–2333.
    Google ScholarLocate open access versionFindings
  • M. Längkvist, L. Karlsson, A. Loutfi, A review of unsupervised feature learning and deep learning for time-series modeling, Pattern Recogn. Lett. 42 (2014) 11–24.
    Google ScholarLocate open access versionFindings
  • A. Yoshihara, K. Fujikawa, K. Seki, K. Uehara, Predicting Stock Market Trends by Recurrent Deep Neural Networks, 2014, pp. 759–769.
    Google ScholarLocate open access versionFindings
  • F. Shen, J. Chao, J. Zhao, Forecasting exchange rate using deep belief networks and conjugate gradient method, Neurocomputing 167 (2015) 243–253.
    Google ScholarFindings
  • P. Tino, C. Schittenkopf, G. Dorffner, Financial volatility trading using recurrent neural networks, IEEE Trans. Neural Netw. 12 (4) (2001) 865–874.
    Google ScholarLocate open access versionFindings
  • Y. Deng, F. Bao, Y. Kong, Z. Ren, Q. Dai, Deep direct reinforcement learning for financial signal representation and trading, IEEE Trans. Neural Netw. Learn. Syst. 28 (3) (2017) 653–664, http://dx.doi.org/10.1109/TNNLS.2016.2522401.
    Locate open access versionFindings
  • Y. LeCun, B.E. Boser, J.S. Denker, D. Henderson, R.E. Howard, W.E. Hubbard, L.D. Jackel, Handwritten digit recognition with a back-propagation network, Advances in Neural Information Processing Systems (1990) 396–404.
    Google ScholarLocate open access versionFindings
  • Y. LeCun, L. Jackel, L. Bottou, C. Cortes, J.S. Denker, H. Drucker, I. Guyon, U. Muller, E. Sackinger, P. Simard, et al., Learning algorithms for classification: a comparison on handwritten digit recognition, Neural Netw. 261 (1995) 276.
    Google ScholarLocate open access versionFindings
  • I. Goodfellow, Y. Bengio, A. Courville, Deep Learning, MIT Press, 2016 http://www.deeplearningbook.org.
    Findings
  • M.M. Mostafa, Forecasting stock exchange movements using neural networks: empirical evidence from Kuwait, Expert Syst. Appl. 37 (9) (2010) 6302–6309.
    Google ScholarLocate open access versionFindings
  • Y. Zheng, Q. Liu, E. Chen, Y. Ge, J.L. Zhao, Time series classification using multi-channels deep convolutional neural networks, in: International Conference on Web-Age Information Management, Springer, 2014, pp. 298–310.
    Google ScholarLocate open access versionFindings
  • Z. Wang, W. Yan, T. Oates, Time series classification from scratch with deep neural networks: a strong baseline, in: 2017 International Joint Conference on Neural Networks (IJCNN), IEEE, 2017, pp. 1578–1585.
    Google ScholarLocate open access versionFindings
  • A. Le Guennec, S. Malinowski, R. Tavenard, Data augmentation for time series classification using convolutional neural networks, ECML/PKDD Workshop on Advanced Analytics and Learning on Temporal Data (2016).
    Google ScholarLocate open access versionFindings
  • M.S. Seyfioglu, A.M. Ozbayoglu, S.Z. Gurbuz, Deep convolutional autoencoder for radar-based classification of similar aided and unaided human activities, IEEE Trans. Aerosp. Electron. Syst. PP (99) (2018), 1-1.
    Google ScholarLocate open access versionFindings
  • E.K. Kabanga, C.H. Kim, Malware images classification using convolutional neural network, J. Comput. Commun. 6 (01) (2017) 153.
    Google ScholarLocate open access versionFindings
  • S. Yue, Imbalanced Malware Images Classification: A CNN Based Approach, 2017 arXiv:1708.08042. Omer Berat Sezer received the B.S. degree in Electrical and Electronics Engineering from Middle East Technical University NCC, Ankara, Turkey, in 2009, with an emphasis on telecommunications and computers; M.S. degree in Electrical and Electronics Engineering from Middle East Technical University, Ankara, Turkey, in 2013,with an emphasis on computer networks. He is currently a Ph.D. candidate at Department of Computer Engineering of TOBB University of Economics and Technology, in Ankara, Turkey and he is also working as a senior researcher and software engineer at The Scientific and Technological Research Council of Turkey – Space Technologies Research Institute, in Ankara, Turkey. His research interests are machine learning, Internet of things, big data and time series data analytics.
    Findings
作者
Omer Berat Sezer
Omer Berat Sezer
您的评分 :
0

 

标签
评论
数据免责声明
页面数据均来自互联网公开来源、合作出版商和通过AI技术自动分析结果,我们不对页面数据的有效性、准确性、正确性、可靠性、完整性和及时性做出任何承诺和保证。若有疑问,可以通过电子邮件方式联系我们:report@aminer.cn
小科