Forecasting high-frequency stock returns: a comparison of alternative methods | Annals of Operations Research
Skip to main content

Advertisement

Springer Nature Link
Log in

Forecasting high-frequency stock returns: a comparison of alternative methods

  • S.I.: Risk Management Decisions and Value under Uncertainty
  • Published:
Annals of Operations Research Aims and scope Submit manuscript

Abstract

We compare the performance of various advanced forecasting techniques, namely artificial neural networks, k-nearest neighbors, logistic regression, Naïve Bayes, random forest classifier, support vector machine, and extreme gradient boosting classifier to predict stock price movements based on past prices. We apply these methods with the high frequency data of 27 blue-chip stocks traded in the Istanbul Stock Exchange. Our findings reveal that among the selected methodologies, random forest and support vector machine are able to capture both future price directions and percentage changes at a satisfactory level. Moreover, consistent ranking of the methodologies across different time frequencies and train/test set partitions prove the robustness of our empirical findings.

This is a preview of subscription content, log in via an institution to check access.

Access this article

Log in via an institution

Subscribe and save

Springer+ Basic
¥17,985 /Month
  • Get 10 units per month
  • Download Article/Chapter or eBook
  • 1 Unit = 1 Article or 1 Chapter
  • Cancel anytime
Subscribe now

Buy Now

Price includes VAT (Japan)

Instant access to the full article PDF.

Institutional subscriptions

Similar content being viewed by others

Notes

  1. Consequently, Fama (1970) classifies informative efficiency into three categories: (i) Weak efficiency, when the current price contains information from the past series of prices; (ii) Semi-strong efficiency, when the past price contains all the public information associated with that asset; and (iii) Strong efficiency, when the price reflects all public and private information relating to that asset.

  2. Being world’s 17th largest economy, Turkey has been a destination of extensive capital flows in the last few years and its stock market has attracted a lot of attention. With 614 TL bn. market capitalization and more than 1 TL tn. traded value at the end of 2016, equity market of Istanbul Stock Exchange is ranked 6th in traded value among all emerging markets in the world. Moreover, it is ranked 3rd in the whole world with a share turnover velocity of more than 200% in the same year. These statistics show that there is a high level of trading activity at a global scale in Istanbul Stock Exchange and makes it a perfect ground to test our trading strategies.

  3. See also Fornaciari and Grillenzoni (2017) and Ho et al. (2017).

  4. It was not possible to obtain this tick-by-tick data from data providers so the data comes directly from the database of the stock exchange. Since this is a unique data set, it was not possible the extend it beyond 2016.

  5. It is important to note that portfolios managed by financial institutions are very large and small abnormal returns (forecasting signs above 50% and profit ratio of 10%) could represent extra earnings worth millions of dollars. Similarly, trading thousands of times in a day with small abnormal returns would accumulate up to considerable amount of excess dollar gain in the market.

References

  • Avci, E., Bunn, D., Ketter, W., & van Heck, E. (2019). Agent-level determinants of price expectation formation in online double-sided auctions. Decision Support Systems, 124, 113068.

    Article  Google Scholar 

  • Brasileiro, R., Souza, V. L. F., & Oliviera, A. L. I. (2017). Automatic trading method based on piecewise aggregate approximation and multi-swarm of improved self-adaptive particle swarm optimization with validation. Decision Support Systems, 104, 79–91.

    Article  Google Scholar 

  • Breiman, L. (2001). Random forests. Machine Learning, 45(1), 5–32.

  • Cheng, C.-H., & Yang, J.-H. (2018). Fuzzy time-series model based on rough set rule induction for forecasting stock price. Neurocomputing, 302, 33–45.

    Article  Google Scholar 

  • Christensen, H. L., Murphy, J., & Godsill, S. J. (2012). Forecasting high-frequency futures returns using online Langevin dynamics. IEEE Journal of Selected Topics in Signal Processing, 6, 366–380.

    Article  Google Scholar 

  • D’Ecclesia, R. L., & Clementi, D. (2021). Volatility in the stock market: ANN versus parametric models. Annals of Operations Research, 299(1), 1101–1127.

    Article  Google Scholar 

  • Fama, E. F. (1970). Efficient capital markets: A review of theory and empirical work. Journal of Finance, 25, 383–417.

    Article  Google Scholar 

  • Fama, E. F. (1991). Efficient capital markets: II. Journal of Finance, 46, 1575–1617.

    Article  Google Scholar 

  • Fernandes, F. D. S., Stasinakis, C., & Zekaite, Z. (2019). Forecasting government bond spreads with heuristic models: Evidence from the Eurozone periphery. Annals of Operations Research, 282, 87–118.

    Article  Google Scholar 

  • Fernández-Rodríguez, F., González-Martel, C., & Sosvilla-Rivero, S. (2000). On the profitability of technical trading rules based on artificial neural networks: Evidence from the Madrid stock market. Economics Letters, 69, 89–94.

    Article  Google Scholar 

  • Feuerriegel, S., & Gordon, J. (2018). Long-term stock index forecasting based on text mining of regulatory disclosures. Decision Support Systems, 112, 88–97.

    Article  Google Scholar 

  • Fornaciari, M., & Grillenzoni, C. (2017). Evaluation of on-line trading systems: Markov-switching vs time-varying parameter models. Decision Support Systems, 93, 51–61.

    Article  Google Scholar 

  • Freeman, J. A., & Skapura, D. M. (1991). Neural networks, algorithms, applications, and programming techniques. Boston: Addison-Wesley Publishing Company.

    Google Scholar 

  • Gencay, R. (1999). Linear, non-linear and essential foreign exchange rate prediction with simple technical trading rules. Journal of International Economics, 47, 91–107.

    Article  Google Scholar 

  • Göçken, M., Özçalici, M., Boru, A., & Dosdogru, A. T. (2016). Integrating metaheuristics and artificial neural networks for improved stock price prediction. Expert Systems with Applications, 44, 320–331.

    Article  Google Scholar 

  • Ho, C.-S., Damien, P., Gu, B., & Konana, P. (2017). The time-varying nature of social media sentiments in modeling stock returns. Decision Support Systems, 101, 69–81.

    Article  Google Scholar 

  • Hudson, R., & Urquhart, A. (2021). Technical trading and cryptocurrencies. Annals of Operations Research, 297(1), 191–220.

    Article  Google Scholar 

  • Iglesias Caride, M., Bariviera, A. F., & Lanzarini, L. (2018). Stock returns forecast: An examination by means of artificial neural networks. In C. Berger-Vachon, A. M. Gil Lafuente, J. Kacprzyk, Y. Kondratenko, J. M. Merigó, & C. F. Morabito (Eds.), Complex systems: Solutions and challenges in economics, management and engineering: Dedicated to Professor Jaime Gil Aluja (pp. 399–410). Cham: Springer.

    Chapter  Google Scholar 

  • Isasi Viñuela, P., & Galván León, I. M. (2004). Redes de neuronas artificiales. Un enfoque práctico. Prentice Hall: Pearson.

    Google Scholar 

  • Jeong, G., & Kim, H. Y. (2019). Improving financial trading decisions using deep Q-learning: Predicting the number of shares, action strategies, and transfer learning. Expert Systems with Applications, 117, 125–138.

    Article  Google Scholar 

  • Karhunen, M. (2019). Algorithmic sign prediction and covariate selection across eleven international stock markets. Expert Systems with Applications, 115, 256–263.

    Article  Google Scholar 

  • Kim, H. Y., & Won, C. H. (2018). Forecasting the volatility of stock price index: A hybrid model integrating LSTM with multiple GARCH-type models. Expert Systems with Applications, 103, 25–37.

    Article  Google Scholar 

  • Kyriakou, I., Mousavi, P., Nielsen, J. P., & Scholz, M. (2021). Forecasting benchmarks of long-term stock returns via machine learning. Annals of Operations Research, 297(1), 221–240.

    Article  Google Scholar 

  • Lahmiri, S. (2016). Intraday stock price forecasting based on variational mode decomposition. Journal of Computational Science, 12, 23–27.

    Article  Google Scholar 

  • Lanzarini, L., Iglesias Caride, J. M., & Bariviera, A. F. (2011). Are technical trading rules useful to beat the market? Evidence from the Brazilian stock market. In World congress of international fuzzy systems association 2011 and Asia fuzzy systems society international conference 2011 (pp. 21–25).

  • Malagrino, L. S., Roman, N. T., & Monteiro, A. M. (2018). Forecasting stock market index daily direction: A Bayesian network approach. Expert Systems with Applications, 105, 11–22.

    Article  Google Scholar 

  • McGroarty, F., Booth, A., Gerding, E., & Chinthalapati, V. L. R. (2019). High frequency trading strategies, market fragility and price spikes: An agent based model perspective. Annals of Operations Research, 282, 217–244.

    Article  Google Scholar 

  • Nadkarni, J., & Neves, R. F. (2018). Combining neuro-evolution and principal component analysis to trade in the financial markets. Expert Systems with Applications, 103, 184–195.

    Article  Google Scholar 

  • Nam, K., & Seong, N. Y. (2019). Financial news-based stock movement prediction using causality analysis of influence in the Korean stock market. Decision Support Systems, 117, 100–112.

    Article  Google Scholar 

  • Oztekin, A., Kizilaslan, R., Freund, S., & Iseri, A. (2016). A data analytic approach to forecasting daily stock returns in an emerging market. European Journal of Operational Research, 253, 697–710.

    Article  Google Scholar 

  • Qiu, M., Song, Y., & Akagi, F. (2016). Application of artificial neural network for the prediction of stock market returns: The case of the Japanese stock market. Chaos, Solitons & Fractals, 85, 1–7.

  • Rapach, D. E., & Zhou, G. (2013). Forecasting stock returns. In G. Elliott & A. Timmermann (Eds.), Handbook of economic forecasting (Vol. 6, pp. 328–383). Amsterdam: Elsevier.

    Google Scholar 

  • Riedmiller, M., & Braun, H. (1993). A direct adaptive method for faster backpropagation learning: The RPROP algorithm. In A. Anu (Ed.), IEEE international conference on neural networks (pp. 586–591). IEEE.

  • Ross, S. A. (2005). Neoclassical finance. Princeton: Princeton University Press.

    Book  Google Scholar 

  • Shang, H. L., Yang, Y., & Kearney, F. (2019). Intraday forecasts of a volatility index: Functional time series methods with dynamic updating. Annals of Operations Research, 282, 331–354.

    Article  Google Scholar 

  • Zuo, Y., & Kita, E. (2012). Stock price forecast using Bayesian network. Expert Systems with Applications, 39, 6729–6737.

    Article  Google Scholar 

Download references

Acknowledgements

Ahmet Sensoy gratefully acknowledges support from the Turkish Academy of Sciences under its Outstanding Young Scientist Program (TUBA-GEBIP).

Author information

Authors and Affiliations

  1. Department of Banking and Finance, Burdur Mehmet Akif Ersoy University, Burdur, Turkey

    Erdinc Akyildirim

  2. Department of Banking and Finance, University of Zurich, Zurich, Switzerland

    Erdinc Akyildirim

  3. Department of Business, Universitat Rovira i Virgili, Reus, Spain

    Aurelio F. Bariviera

  4. IPAG Business School, Paris, France

    Duc Khuong Nguyen

  5. International School, Vietnam National University, Hanoi, Vietnam

    Duc Khuong Nguyen

  6. Faculty of Business Administration, Bilkent University, Ankara, Turkey

    Ahmet Sensoy

Authors
  1. Erdinc Akyildirim
    View author publications

    You can also search for this author in PubMed Google Scholar

  2. Aurelio F. Bariviera
    View author publications

    You can also search for this author in PubMed Google Scholar

  3. Duc Khuong Nguyen
    View author publications

    You can also search for this author in PubMed Google Scholar

  4. Ahmet Sensoy
    View author publications

    You can also search for this author in PubMed Google Scholar

Corresponding author

Correspondence to Erdinc Akyildirim.

Additional information

Publisher's Note

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Appendix A. Results for the ‘all trades’ scenario

Appendix A. Results for the ‘all trades’ scenario

This appendix presents some descriptive statistics and the results for our main forecasting techniques applied to ‘all trades’ instead of those occurring only in continuous matching session (Tables 2, 3, 4, 5, 6, 7, 8, 9 and 10).

Table 11 Sign prediction and ideal profit ratios of different machine learning algorithms for all trades, sampled at 10 min frequency with 0.7/0.3 train/test set partition
Full size table
Table 12 Sign prediction and ideal profit ratios of different machine learning algorithms for all trades, sampled at 30 min frequency with 0.7/0.3 train/test set partition
Full size table
Table 13 Sign prediction and ideal profit ratios of different machine learning algorithms for all trades, sampled at 60 min frequency with 0.7/0.3 train/test set partition
Full size table
Table 14 Sign prediction and ideal profit ratios of different machine learning algorithms for all trades, sampled at 10 min frequency with 0.8/0.2 train/test set partition
Full size table
Table 15 Sign prediction and ideal profit ratios of different machine learning algorithms for all trades, sampled at 30 min frequency with 0.8/0.2 train/test set partition
Full size table
Table 16 Sign prediction and ideal profit ratios of different machine learning algorithms for all trades, sampled at 60 min frequency with 0.8/0.2 train/test set partition
Full size table
Table 17 Sign prediction and ideal profit ratios of different machine learning algorithms for all trades, sampled at 10 min frequency with 0.9/0.1 train/test set partition
Full size table
Table 18 Sign prediction and ideal profit ratios of different machine learning algorithms for all trades, sampled at 30 min frequency with 0.9/0.1 train/test set partition
Full size table
Table 19 Sign prediction and ideal profit ratios of different machine learning algorithms for all trades, sampled at 60 min frequency with 0.9/0.1 train/test set partition
Full size table

Rights and permissions

Reprints and permissions

About this article

Check for updates. Verify currency and authenticity via CrossMark

Cite this article

Akyildirim, E., Bariviera, A.F., Nguyen, D.K. et al. Forecasting high-frequency stock returns: a comparison of alternative methods. Ann Oper Res 313, 639–690 (2022). https://doi.org/10.1007/s10479-021-04464-8

Download citation

  • Accepted:

  • Published:

  • Issue Date:

  • DOI: https://doi.org/10.1007/s10479-021-04464-8

Keywords

Search

Navigation