Abstract
Financial processes are frequently explained by econometric models, however, data-driven approaches may outperform the analytical models with adequate amount and quality data and algorithms. In the case of today’s state-of-the-art deep learning methods the more data leads to better models. However, even if the model is trained on massively parallel hardware, the preprocessing of a large amount of data is usually still done in a traditional way (e.g. few hundreds of threads on Central Processing Unit, CPU).
In this paper, we propose a GPU accelerated pipeline, which assesses the burden of time taken with data preparation for machine learning in financial applications. With the reduced time, it enables its user to experiment with multiple parameter setups in much less time. The pipeline processes and models a specific type of financial data – limit order books – on massively parallel hardware. The pipeline handles data collection, order book preprocessing, data normalisation, and batching into training samples, which can be used for training deep neural networks and inference. Time comparisons of baseline and optimized approaches are part of this paper.
This is a preview of subscription content, log in via an institution to check access.
Access this chapter
Tax calculation will be finalised at checkout
Purchases are for personal use only
Notes
- 1.
Such an order book does not exist for the so called dark pool, because the orders are typically not published in dark pools [1].
References
Ganchev, K., Kearns, M., Nevmyvaka, Y., Vaughan, J.W.: Censored exploration and the dark pool problem (2012)
LeCun, Y., Bengio, Y., Hinton, G.: Deep learning. Nature 521(7553), 436–444 (2015)
Black, F., Scholes, M.: The pricing of options and corporate liabilities. J. Polit. Econ. 81(3), 637–654 (1973)
Ranaldo, A.: Order aggressiveness in limit order book markets. J. Financ. Mark. 7, 53–74 (2004)
Cont, R., Stoikov, S., Talreja, R.: A stochastic model for order book dynamics. Oper. Res. 58, 549–563 (2010)
Beck, J., et al.: Sensing social media signals for cryptocurrency news. In: Companion Proceedings of the 2019 World Wide Web Conference on - WWW 2019 (2019)
Mirtaheri, M., Abu-El-Haija, S., Morstatter, F., Steeg, GV., Galstyan, A.: Identifying and analyzing cryptocurrency manipulations in social media (2019)
Tsantekidis, A., Passalis, N., Tefas, A., Kanniainen, J., Gabbouj, M., Iosifidis, A.: Using deep learning for price prediction by exploiting stationary limit order book features (2018)
Ntakaris, A., Magris, M., Kanniainen, J., Gabbouj, M., Iosifidis, A.: Benchmark dataset for mid-price forecasting of limit order book data with machine learning methods. J. Forecast. 37(8), 852–866 (2018)
Bibinger, M., Neely, C., Winkelmann, L.: Estimation of the discontinuous leverage effect: evidence from the NASDAQ order book. In: Federal Reserve Bank of St. Louis, Working Papers, April 2017
Wei, H., Wang, Y., Mangu, L., Decker, K.: Model-based reinforcement learning for predictions and control for limit order books (2019)
Di Persio, L., Honchar, O.: Artificial neural networks architectures for stock price prediction: comparisons and applications. Int. J. Circ. Syst. Signal Process. 10, 403–413 (2016)
Sirignano, J., Cont, R.: Universal features of price formation in financial markets: perspectives from deep learning (2018)
Zhang, Z., Zohren, S., Roberts, S.: DeepLOB: deep convolutional neural networks for limit order books. IEEE Trans. Signal Process. 67(11), 3001–3012 (2019)
Acknowledgment
The research presented in this paper, carried out by BME was supported by the Ministry of Innovation and the National Research, Development and Innovation Office within the framework of the Artificial Intelligence National Laboratory Programme, by the NRDI Fund based on the charter of bolster issued by the NRDI Office under the auspices of the Ministry for Innovation and Technology, by the European Union, co-financed by the European Social Fund (EFOP-3.6.2-16-2017-00013, Thematic Fundamental Research Collaborations Grounding Innovation in Informatics and Infocommunications), by János Bolyai Research Scholarship of the Hungarian Academy of Sciences and by Doctoral Research Scholarship of Ministry of Human Resources (ÚNKP-20-5-BME-210) in the scope of New National Excellence Program. We gratefully acknowledge the support of NVIDIA Corporation with the donation of the Titan V GPU used for this research.
Author information
Authors and Affiliations
Corresponding authors
Editor information
Editors and Affiliations
Rights and permissions
Copyright information
© 2020 Springer Nature Switzerland AG
About this paper
Cite this paper
Burján, V., Gyires-Tóth, B. (2020). GPU Accelerated Data Preparation for Limit Order Book Modeling. In: Nicosia, G., et al. Machine Learning, Optimization, and Data Science. LOD 2020. Lecture Notes in Computer Science(), vol 12565. Springer, Cham. https://doi.org/10.1007/978-3-030-64583-0_35
Download citation
DOI: https://doi.org/10.1007/978-3-030-64583-0_35
Published:
Publisher Name: Springer, Cham
Print ISBN: 978-3-030-64582-3
Online ISBN: 978-3-030-64583-0
eBook Packages: Computer ScienceComputer Science (R0)