Proceedings of the 19th Conference on Computer Science and Intelligence Systems (FedCSIS)

DOI: http://dx.doi.org/10.15439/2024F3560

Citation: Proceedings of the 19th Conference on Computer Science and Intelligence Systems (FedCSIS), M. Bolanowski, M. Ganzha, L. Maciaszek, M. Paprzycki, D. Ślęzak (eds). ACSIS, Vol. 39, pages 755–759 (2024)

Full text

Abstract. Predicting stock price trends is a challenging puzzle. The immediate price of a stock is affected by an uncountable number of factors. Thus there is essentially no way to accurately predict short-term stock price due to dynamic, incomplete, erratic, and chaotic data. However, by analyzing key financial indicators, it is possible to gain an accurate understanding of a company's operations, make a quantitative assessment of its value, and thus make a reasonable prediction of the long-term trend of its stock price. In this FedCSIS 2024 Data Science Challenge, participants are asked to predict the trends of the stocks which are chosen from the Standard \& Poor's 500 index. In this paper, we apply a wrapper feature selection method that tightly combines the steps of feature selection and model building to result in better prediction models, and provide insight into the indicators. After selecting the best set of features, we train two kinds of gradient boost machine: multi-classification model and regression model for class and risk-return performance prediction respectively. Finally a high confidence voting strategy is used to determine the kind of trading action (buy, sell, or hold). Experimental and competition results demonstrate the effectiveness of the methodology in this paper.

References

1. M.Leippold, Q.Wang, W.Zhou. Machine-learning in the Chinese Stock Market. Journal of Financial Economics, 2022, 145(2): 64-82. http://dx.doi.org/10.1016/j.jfineco.2021.08.017.
2. S.Wu, O.Irsoy, S.Lu, V.Dabravolski, M.Dredze, S.Gehrmann, P.Kambadur, D.Rosenberg, G.Mann. BloombergGPT: A Large Language Model for Finance. https://arxiv.org/abs/2303.17564v3 [cs.LG]. https://doi.org/10.48550/arXiv.2303.17564.
3. https://knowledgepit.ai/fedcsis-2024-challenge/.
4. Aleksandar M. Rakicevic, Pavle D. Milosevic, Ivana T. Dragovic, Ana M. Poledica, Milica M. Zukanovic, Andrzej Janusz, Dominik Slezak: “Predicting Stock Trends Using Common Financial Indicators: A Summary of FedCSIS 2024 Data Science Challenge Held on KnowledgePit.ai Platform”. In: Proceedings of FedCSIS 2024 (2024).
5. C. Lin. Predicting Frags in Tactic Games using Machine Learning Techniques and Intuitive Knowledge. 2023 IEEE International Conference on Multimedia and Expo Workshops (ICMEW), Brisbane, Australia, 2023, pp. 11-15, https://doi.org/10.1109/ICMEW59549.2023.00008.
6. C.Lin. Tackling Variable-length Sequences with High-cardinality Features in Cyber-attack Detection. Proceedings of the 18th Conference on Computer Science and Intelligence Systems, Vol.35, pages 1295-1299 (2023). https://dx.doi.org/10.15439/2023F2385.
7. A.Rakićević, A.Poledica, B.Petrović. A Novel IBA-DE Hybrid Approach for Modeling Sovereign Credit Ratings. Mathematics 2022, 10, 2679. https://doi.org/10.3390/math10152679.
8. A.Janusz, A.Jamiołkowski, M.Okulewicz. Predicting the Costs of Forwarding Contracts: Analysis of Data Mining Competition Results. Proceedings of the 17th Conference on Computer Science and Intelligence Systems, ACSIS, Vol.30, pages 399-402 (2022). https://dx.doi.org/10.15439/2022F303.
9. M.Czerwinski, M.Michalak, P.Biczyk, B.Adamczyk, D.Iwanicki, I.Kostorz, M.Brzeczek, A. Janusz, M.Hermansa, L.Wawrowski, A.Kozlowski. Cybersecurity Threat Detection in the Behavior of IoT Devices: Analysis of Data Mining Competition Results. Proceedings of the 18th Conference on Computer Science and Intelligence Systems, ACSIS, Vol.35, pages 1289-1293 (2023). https://dx.doi.org/10.15439/2023F3089.