logo

SCIENCE CHINA Information Sciences, Volume 64 , Issue 9 : 192102(2021) https://doi.org/10.1007/s11432-020-3064-4

Quantifying the effects of long-term news on stock markets on the basis of the multikernel Hawkes process

More info
  • ReceivedMar 12, 2020
  • AcceptedJun 3, 2020
  • PublishedJul 27, 2021

Abstract


Acknowledgment

This work was supported by National Key Research and Development Program of China (Grant No. 2018AAA0101901) and National Natural Science Foundation of China (Grant Nos. 61976073, 61702137). We thank the anonymous reviewers for their constructive comments.


References

[1] Ding X, Zhang Y, Liu T, et al. Knowledge-driven event embedding for stock prediction. In: Proceedings of the 26th International Conference on Computational Linguistics: Technical Papers, 2016. Google Scholar

[2] Hu Z, Liu W, Bian J, et al. Listening to chaotic whispers: a deep learning framework for news-oriented stock trend prediction. In: Proceedings of International Conference on Web Search and Data Mining, 2018. 261--269. Google Scholar

[3] Wang W Y, Hua Z. A semiparametric Gaussian copula regression model for predicting financial risks from earnings calls. In: Proceedings of the 52nd Annual Meeting of the Association for Computational Linguistics, 2014. 1155--1165. Google Scholar

[4] Xie B, Passonneau R, Wu L, et al. Semantic frames to predict stock price movement. In: Proceedings of the 51st Annual Meeting of the Association for Computational Linguistics, 2013. 873--883. Google Scholar

[5] Ding X, Zhang Y, Liu T, et al. Using structured events to predict stock price movement: an empirical investigation. In: Proceedings of the 2014 Conference on Empirical Methods in Natural Language Processing, 2014. 1415--1425. Google Scholar

[6] Ding X, Zhang Y, Liu T, et al. Deep learning for event-driven stock prediction. In: Proceedings of 24th International Joint Conference on Artificial Intelligence, 2015. 2327--2333. Google Scholar

[7] Xu Y, Cohen S B. Stock movement prediction from tweets and historical prices. In: Proceedings of the 56th Annual Meeting of the Association for Computational Linguistics, Melbourne, 2018. 1970--1979. Google Scholar

[8] Hawkes A G. Spectra of some self-exciting and mutually exciting point processes. Biometrika, 1971, 58: 83-90 CrossRef Google Scholar

[9] Duan J, Zhang Y, Ding X, et al. Learning target-specific representations of financial news documents for cumulative abnormal return prediction. In: Proceedings of International Conference on Computational Linguistics, 2018. Google Scholar

[10] Lima R, Choi J. Hawkes process kernel structure parametric search with renormalization factors. 2018,. arXiv Google Scholar

[11] Ogata Y. Seismicity analysis through point-process modeling: a review. In: Proceedings of Seismicity Patterns, Their Statistical Significance and Physical Meaning, 1999. 471--507. Google Scholar

[12] Mishra S, Rizoiu M A, Xie L. Feature driven and point process approaches for popularity prediction. In: Proceedings of the Conference on Information and Knowledge Management, 2016. 1069--1078. Google Scholar

[13] Gupta A, Farajtabar M, Dilkina B, et al. Discrete interventions in hawkes processes with applications in invasive species management. In: Proceedings of International Joint Conference on Artificial Intelligence, 2018. 3385--3392. Google Scholar

[14] Cao Q, Shen H, Cen K, et al. Deephawkes: bridging the gap between prediction and understanding of information cascades. In: Proceedings of the Conference on Information and Knowledge Management, 2017. 1149--1158. Google Scholar

[15] Du N, Dai H, Trivedi R, et al. Recurrent marked temporal point processes: embedding event history to vector. In: Proceedings of KDD, 2016. Google Scholar

[16] Duan J, Ding X, Liu T. Learning sentence representations over tree structures for target-dependent classification. In: Proceedings of the Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies, 2018. Google Scholar

[17] Rocktäschel T, Grefenstette E, Hermann K M, et al. Reasoning about entailment with neural attention. 2016,. arXiv Google Scholar

[18] Hochreiter S, Schmidhuber J. Long Short-Term Memory. Neural Computation, 1997, 9: 1735-1780 CrossRef Google Scholar

[19] Peters M, Neumann M, Iyyer M, et al. Deep contextualized word representations. In: Proceedings of the Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies, 2018. 2227--2237. Google Scholar

[20] Dozat T, Manning C D. Deep biaffine attention for neural dependency parsing. 2016,. arXiv Google Scholar

[21] Li J, Luong M t, Jurafsky D. A hierarchical neural autoencoder for paragraphs and documents. In: Proceedings of the 53rd Annual Meeting of the Association for Computational Linguistics, 2015. 1106--1115. Google Scholar

[22] Bahdanau D, Cho K, Bengio Y. Neural machine translation by jointly learning to align and translate. 2014,. arXiv Google Scholar

[23] Etesami J, Kiyavash N, Zhang K, et al. Learning network of multivariate hawkes processes: a time series approach. In: Proceedings of the 32nd Conference on Uncertainty in Artificial Intelligence, 2016. 162--171. Google Scholar

[24] Wilson A, Adams R. Gaussian process kernels for pattern discovery and extrapolation. In: Proceedings of International Conference on Machine Learning, 2013. 1067--1075. Google Scholar

[25] Hwang Y, Tong A, Choi J. Automatic construction of nonparametric relational regression models for multiple time series. In: Proceedings of International Conference on Machine Learning, 2016. 3030--3039. Google Scholar

[26] Kingma D P, Ba J. Adam: a method for stochastic optimization. In: Proceedings of the 3rd International Conference on Learning Representations, 2015. Google Scholar

[27] Lavrenko V, Schmill M, Lawrie D, et al. Mining of concurrent text and time series. In: Proceedings of KDD-2000 Workshop on Text Mining, 2000. 37--44. Google Scholar

[28] Taylor S J, Xu X. The incremental volatility information in one million foreign exchange quotations. J Empirical Finance, 1997, 4: 317-340 CrossRef Google Scholar

[29] Andersen T G, Bollerslev T. Intraday periodicity and volatility persistence in financial markets. J Empirical Finance, 1997, 4: 115-158 CrossRef Google Scholar

[30] Atsalakis G S, Valavanis K P. Forecasting stock market short-term trends using a neuro-fuzzy based methodology. Expert Syst Appl, 2009, 36: 10696-10707 CrossRef Google Scholar

[31] Taylor S J. Modelling Financial Time Series. World Scientific, 2008. Google Scholar

[32] Li L, Leng S, Yang J. Stock Market Autoregressive Dynamics: A Multinational Comparative Study with Quantile Regression. Math Problems Eng, 2016, 2016: 1-15 CrossRef Google Scholar

[33] Rather A M, Agarwal A, Sastry V N. Recurrent neural network and a hybrid model for prediction of stock returns. Expert Syst Appl, 2015, 42: 3234-3241 CrossRef Google Scholar

[34] Sun X Q, Shen H W, Cheng X Q. Trading Network Predicts Stock Price. Sci Rep, 2015, 4: 3711 CrossRef ADS Google Scholar

[35] Chen D, Zou Y, Harimoto K, et al. Incorporating fine-grained events in stock movement prediction. In: Proceedings of the 2nd Workshop on Economics and Natural Language Processing, Hong Kong, 2019. 31--40. Google Scholar

[36] Qin Y, Yang Y. What you say and how you say it matters: predicting stock volatility using verbal and vocal cues. In: Proceedings of the 57th Annual Meeting of the Association for Computational Linguistics, Florence, 2019. 390--401. Google Scholar

[37] Luss R, D'Aspremont A. Predicting abnormal returns from news using text classification. Quantitative Finance, 2015, 15: 999-1012 CrossRef Google Scholar

[38] Schumaker R P, Chen H. Textual analysis of stock market prediction using breaking financial news. ACM Trans Inf Syst, 2009, 27: 1-19 CrossRef Google Scholar

[39] Feldman R, Rosenfeld B, Bar-Haim R, et al. The stock sonar--sentiment analysis of stocks based on a hybrid approach. In: Proceedings of the 23rd International Association of Arson Investigators Conference, 2011. Google Scholar

[40] Etzioni O, Fader A, Christensen J, et al. Open information extraction: the second generation. In: Proceedings of 22nd International Joint Conference on Artificial Intelligence, 2011. Google Scholar

[41] Fader A, Soderland S, Etzioni O. Identifying relations for open information extraction. In: Proceedings of the Conference on Empirical Methods in Natural Language Processing, 2011. 1535--1545. Google Scholar

[42] Gao J, Shen H, Liu S, et al. Modeling and predicting retweeting dynamics via a mixture process. In: Proceedings of the 25th International Conference Companion on World Wide Web, 2016. 33--34. Google Scholar

  • Figure 1

    (Color online) The influence of historical news decays with the increase of time. On January 14th, Reuters reported that Samsung had offered to buy BlackBerry for as much as 7.5 billion, which stimulated the stock price of BlackBerry to soar on that day. But in following days, both companies denied they were in talks with respect to any possible takeover, thus the influence of the good news was time-decaying and the stock price continued to fall. We notice that although the acquisition event was later denied, it will still have a certain positive effect on the subsequent stock price of BlackBerry.

  • Figure 2

    (Color online) Framework of stock prediction in combination with the Hawkes process. It omits part of the lines in case of chaos. $t_{n-5}$ has three financial news pieces, whereas $t_{n-3}$ has no news about specific firm $c$.

  • Figure 3

    (Color online) Quantification of time-decaying influence on historical consecutive days in an interval of 6 days (29/04/2015–04/05/2015) using multikernel Hawkes process.

  • Figure 4

    (Color online) Accuracy of prediction with different time intervals (from 2 days to 30 days). The Hawkes process-based models first rise and then fall as the time interval increases; they reach a peak when the time interval is 6 days.

  • Figure 5

    (Color online) Accuracy of prediction with different number of Gaussian kernels (from 1 to 10). Hawkes process based model first rises and then falls as the number of Gaussian kernels increases. They reach a peak when the number of Gaussian kernels is 5.

  • Table 1  

    Table 1Statistics of datasets

    Training Development Test
    # records 18042 996 2003
    Time interval 22/10/2006–14/04/2014 15/04/2014–01/09/2014 02/09/2014–26/08/2015
  • Table 2  

    Table 2Experimental accuracy results of stock movement prediction$^{\rm~a)}$

    Method Accuracy (%)
    short-term (Duan et al.[9]) 52.32
    model-no-weight 53.25
    model-CNN (Ding et al. [6]) 53.72
    model-temporal-attention (Hu et al. [2]) 54.12
    model-temporal-attention (Xu et al. [7]) 54.57
    model-exp (ours) 55.17
    model-pow (ours) 55.32
    multik (exp+pow) (ours) 55.97
    multik (Gaussian) (ours) 56.02
    multik (exp+pow+Gaussian) (ours) 56.12

    a

  • Table 3  

    Table 3Return compared with different methods$^{\rm~a)}$

    Method Return (%)
    AMEX composite index $-21.65$
    NYSE composite index $-9.59$
    NASDAQ composite index 2.16
    short-term $-34.81$
    model-no-weight $-28.81$
    model-temporal-attention (Hu et al. [2]) $-18.82$
    multik (exp+pow+Gaussian) (ours) 8.82

    a

qqqq

Contact and support