STOCK PRICE PREDICTION WITH DEEP LEARNING MODELS: COMPARATIVE ANALYSIS OF LSTM, GRU, RNN, MLP MODELS

Öz

Making predictions about the stock market is challenging and requires extensive research into data patterns. In the era of big data, financial analysts often resort to deep learning techniques in stock price forecasting. These techniques can improve the accuracy of forecasts, allowing investors to make more informed decisions. However, stock price forecasting remains one of the most challenging and unpredictable tasks in financial forecasting due to the complexity and dynamic interactions of the stock market. Deep learning technology has been widely used in the financial industry, primarily to improve financial time series forecasting based on stock prices. This paper proposes a stock price prediction model based on deep learning algorithms to solve the problem of low fit and poor accuracy in traditional stock price prediction models. This study analyzes the stock price movements of Nike (NKE), one of the most established companies in the financial markets, using modern deep learning approaches. We use dataset of daily opening, high, low, and closing prices and trading volumes of Nike stock over 31 years from 1993 to 2024. For this purpose, four different deep learning models, namely Long-Short Term Memory (LSTM) Gated Recurrent Unit (GRU) Recurrent Neural Networks (RNN) Multilayer Perceptron (MLP), were used. According to the analysis results, the training metrics show that the LSTM model exhibits the most successful training performance with the lowest MAE (1.606) and RMSE (0.821) values and explains a large portion of the variance in the dataset with an R² value of 0.998. The GRU model, on the other hand, has slightly higher error metrics (MAE: 1.009, RMSE: 1.190), but maintains its strong prediction ability with an R² value of 0.996. In contrast, the RNN and MLP models showed higher error rates with RMSE values of 1.827 and 1.786, respectively, and were insufficient to capture complex dependencies in time series data. The results highlight the advantages of LSTM and GRU models in financial time series forecasting and show that using these models can produce reliable results, especially in long-term analysis.

Anahtar Kelimeler

• Deep Learning
• Stock Forecasting
• LSTM
• GRU
• RNN
• MLP

DERİN ÖĞRENME MODELLERİ İLE HİSSE SENEDİ FİYAT TAHMİNİ: LSTM, GRU, RNN, MLP MODELLERİNİN KARŞILAŞTIRMALI ANALİZİ

Öz

Borsa hakkında tahminlerde bulunmak oldukça zordur ve veri kalıplarına ilişkin kapsamlı bir araştırma gerektirir. Finans analistleri, büyük veri çağında hisse senedi fiyat tahminlerinde genellikle derin öğrenme tekniklerine başvurmaktadır Bu teknikler, tahminlerin doğruluğunu artırarak yatırımcıların daha bilinçli kararlar almasına olanak tanıyabilmektedir. Ancak, hisse senedi fiyat tahmini, borsa piyasasının karmaşık yapısı ve dinamik etkileşimleri nedeniyle finansal tahminler alanındaki en zorlu ve öngörülemez görevlerden biri olarak kalmaktadır. Derin öğrenme teknolojisi, öncelikle hisse senedi fiyatlarına dayalı finansal zaman serisi tahminini iyileştirmek için finans sektöründe yaygın olarak kullanılmaktadır. Geleneksel hisse senedi fiyat tahmin modellerindeki düşük uyum ve zayıf doğruluk sorununu çözmek için bu makale, derin öğrenme algoritmalarına dayalı bir hisse senedi fiyat tahmin modeli önermektedir. Bu çalışmada, finans piyasalarının en köklü şirketlerinden biri olan Nike'ın (NKE) hisse senedi fiyat hareketleri, modern derin öğrenme yaklaşımları kullanılarak analiz edilmiştir. 1993'ten 2024'e uzanan 31 yıllık süreçte, Nike hissesinin günlük açılış, yüksek, düşük, kapanış fiyatları ve işlem hacimlerini içeren bir veri seti kullanılmıştır. Bu amaçla, Uzun-Kısa Süreli Bellek (LSTM) Kapılı Tekrarlayan Birim (GRU) Yinelemeli Sinir Ağları (RNN) Çok Katmanlı Algılayıcı (MLP) olmak üzere dört farklı derin öğrenme modeli ele alınmıştır. Analiz sonuçlarına göre, eğitim metrikleri, LSTM modelinin en düşük MAE (1.606) ve RMSE (0.821) değerleriyle en başarılı eğitim performansını sergilediğini ve 0.998 R² değeriyle veri setindeki varyansın büyük bir kısmını açıkladığını göstermektedir. GRU modeli ise biraz daha yüksek hata metriklerine (MAE: 1.009, RMSE: 1.190) sahip olmakla birlikte, 0.996 R² değeriyle güçlü tahmin yeteneğini korumaktadır. Buna karşılık, RNN ve MLP modelleri, sırasıyla 1.827 ve 1.786 RMSE değerleri ile daha yüksek hata oranları göstermiş ve zaman serisi verilerindeki karmaşık bağımlılıkları yakalamada yetersiz kalmıştır. Sonuçlar, finansal zaman serisi tahminlerinde LSTM ve GRU modellerinin sağladığı avantajları vurgularken, bu modellerin kullanımının özellikle uzun vadeli analizlerde güvenilir sonuçlar üretebileceğini göstermiştir.

Anahtar Kelimeler

• Derin Öğrenme
• Hisse senedi tahmini
• LTSM
• GRU
• RNN
• MLP

Kaynakça

1. Akita, R., Yoshihara, A., Matsubara, T., and Uehara, K. (2016,). Deep learning for stock prediction using numerical and textual information. In 2016 IEEE/ACIS 15th International Conference on Computer and Information Science (ICIS) (pp. 1-6). IEEE. https://doi.org/10.1109/ICIS.2016.7550882
2. Akşehir, Z. D., and Kilic, E. (2022). How to handle data imbalance and feature selection problems in CNN-based stock price forecasting. IEEE Access, 10, 31297-31305. doi: 10.1109/ACCESS.2022.3160797
3. Albayrak, E., ve Saran, N. (2023). İstatistiksel ve Derin Öğrenme Modellerini Kullanarak Hisse Senedi Fiyat Tahmini. Türkiye Bilişim Vakfı Bilgisayar Bilimleri ve Mühendisliği Dergisi, 16(2), 161-169. https://doi.org/10.54525/tbbmd.1031017
4. Al-Tamimi, H. A. H., Alwan, A. A., and Abdel Rahman, A. A. (2011). Factors affecting stock prices in the UAE financial markets. Journal of Transnational Management, 16(1), 3-19. https://doi.org/10.1080/15475778.2011.549441
5. Bao, W., Yue, J., and Rao, Y. (2017). A deep learning framework for financial time series using stacked autoencoders and long-short-term memory. PloS one, 12(7), e0180944. https://doi.org/10.1371/journal.pone.0180944
6. Baek, Y., and Kim, H. Y. (2018). ModAugNet: A new forecasting framework for stock market index value with an overfitting prevention LSTM module and a prediction LSTM module. Expert Systems with Applications, 113, 457-480. https://doi.org/10.1016/j.eswa.2018.07.019
7. Bentoumi, M., Daoud, M., Benaouali, M., & Taleb Ahmed, A. (2022). Improvement of emotion recognition from facial images using deep learning and early stopping cross validation. Multimedia Tools and Applications, 81(21), 29887-29917.
8. Booth, G. G., Martikainen, T., Sarkar, S. K., Virtanen, I., and Yli-Olli, P. (1994). Nonlinear dependence in Finnish stock returns. European Journal of Operational Research, 74(2), 273-283. https://doi.org/10.1016/0377-2217(94)90096-5

9. Breidt, F. J., Crato, N., and De Lima, P. (1998). The detection and estimation of long memory in stochastic volatility. Journal of Econometrics, 83(1-2), 325-348. https://doi.org/10.1016/S0304-4076(97)00072-9
10. Chai, T., and Draxler, R. R. (2014). Root mean square error (RMSE) or mean absolute error (MAE)? – Arguments against avoiding RMSE in the literature. Geoscientific Model Development, 7(3), 1247-1250. https://doi.org/10.5194/gmd-7-1247-2014
11. Cho, K. (2014). Learning phrase representations using RNN encoder-decoder for statistical machine translation. arXiv preprint arXiv:1406.1078. https://doi.org/10.48550/arXiv.1406.1078
12. Çevik, O. (2002). İstanbul Menkul Kıymetler Borsası Endeksinin Box-Jenkins Yöntemı İle Modellemesi. Afyon Kocatepe Üniversitesi İktisadi ve İdari Bilimler Fakültesi Dergisi, 4(1), 17-32.
13. Çoban, Ç., ve Hayat, E. (2023). Hisse Senedi Piyasası Analizinde Farklı Derin Sinir Ağı Modellerinin Karşılaştırılması. Adnan Menderes Üniversitesi Sosyal Bilimler Enstitüsü Dergisi, 10(2), 120-139. https://doi.org/10.30803/adusobed.1402228
14. Dalkıran, İ., ve Ozan, M. (2022). Derin Öğrenme Teknikleri Kullanılarak Borsadaki Hisse Değerlerinin Tahmin Edilmesi. Avrupa Bilim ve Teknoloji Dergisi, 39, 143-148.
15. Erden, C. (2023). Derin Öğrenme ve ARIMA Yöntemlerinin Tahmin Performanslarının Kıyaslanması: Bir Borsa İstanbul Hissesi Örneği. Yönetim Ve Ekonomi Dergisi, 30(3), 419-438. https://doi.org/10.18657/yonveek.1208807
16. Fischer, T., and Krauss, C. (2018). Deep learning with long short-term memory networks for financial market predictions. European journal of operational research, 270(2), 654-669. https://doi.org/10.1016/j.ejor.2017.11.054
17. Gamboa, J. C. B. (2017). Deep learning for time-series analysis. arXiv preprint arXiv:1701.01887.
18. Gavcar, E., ve Metin, H. M. (2021). Hisse senedi değerlerinin makine öğrenimi (derin öğrenme) ile tahmini. Ekonomi ve Yönetim Araştırmaları Dergisi, 10(2), 1-11.
19. Gökbulut, R. İ., and Pekkaya, M. (2014). Estimating and forecasting volatility of financial markets using asymmetric GARCH models: An application on Turkish financial markets. International Journal of Economics and Finance, 6(4), 23-35. http://dx.doi.org/10.5539/ijef.v6n4p23
20. Haykin, S. (2009). Neural networks and learning machines, 3/E. Bangalore:Pearson Education
21. Heaton, J. B., Polson, N. G., and Witte, J. H. (2016). Deep learning in finance. arXiv preprint arXiv:1602.06561. https://doi.org/10.48550/arXiv.1602.06561
22. Hewamalage, H., Bergmeir, C., and Bandara, K. (2021). Recurrent neural networks for time series forecasting: Current status and future directions. International Journal of Forecasting, 37(1), 388-427.
23. Hochreiter, S., and Schmidhuber, J. (1997). Long short-term memory. Neural Computation, 9(8), 1735-1780. https://doi.org/10.1162/neco.1997.9.8.1735
24. Hundman, K., Constantinou, V., Laporte, C., Colwell, I., and Soderstrom, T. (2018, July). Detecting spacecraft anomalies using lstms and nonparametric dynamic thresholding. In Proceedings of the 24th ACM SIGKDD international conference on knowledge discovery & data mining (pp. 387-395).
25. Karakoyun, E. Ş. (2018). Derin öğrenme ile zaman serilerinin gerçek zamanlı tahmini. Yüksek Lisans Tezi, Necmettin Erbakan Üniversitesi.
26. Kılınç, H. Ç., ve Öztürk, Y. (2022). Hibrit Gri Kurt Optimizasyonu ile Kapılı Tekrarlayan Birim Modeli Kullanılarak Günlük Akım Tahmini. Avrupa Bilim ve Teknoloji Dergisi, (35), 259-267. https://doi.org/10.31590/ejosat.1062777
27. Kim, H. Y., and 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. https://doi.org/10.1016/j.eswa.2018.03.002
28. Längkvist, M., Karlsson, L., & Loutfi, A. (2014). A review of unsupervised feature learning and deep learning for time-series modeling. Pattern Recognition Letters, 42, 11-24.
29. LeCun, Y., Bengio, Y., and Hinton, G. (2015). Deep learning. Nature, 521(7553), 436-444. https://doi.org/10.1038/nature14539
30. Makridakis, S., Wheelwright, S. C., and Hyndman, R. J. (1998). Forecasting: methods and applications. NewYork City: John Wiley & Sons.
31. Makinde, A. (2024). Optimizing Time Series Forecasting: A Comparative Study of Adam and Nesterov Accelerated Gradient on LSTM and GRU networks Using Stock Market data. arXiv preprint arXiv:2410.01843.
32. Montgomery, D. C., Peck, E. A., and Vining, G. G. (2012). Introduction to linear regression analysis, NewYork City: John Wiley & Sons.
33. Munkhdalai, L., Munkhdalai, T., Park, K. H., Lee, H. G., Li, M., and Ryu, K. H. (2019). Mixture of activation functions with extended min-max normalization for forex market prediction. IEEE Access, 7, 183680-183691
34. Nabipour, M., Nayyeri, P., Jabani, H., Mosavi, A., and Salwana, E. (2020). Deep learning for stock market prediction. Entropy, 22(8), 840. https://doi.org/10.3390/e22080840
35. Nelson, D. M., Pereira, A. C., and De Oliveira, R. A. (2017, May). Stock market's price movement prediction with LSTM neural networks. In 2017 International Joint Conference on neural networks (IJCNN) (pp. 1419-1426). Ieee. https://doi.org/10.1109/IJCNN.2017.7966019
36. Öztürk, C., and Karacı, A. (2024). An Empirical Analysis of Stock Price Prediction Using Deep Learning Methods: LSTM, GRU, GAN, and WGAN-GP. Gazi Journal of Engineering Sciences (GJES)/Gazi Mühendislik Bilimleri Dergisi, 10(3).
37. Praveen, P., Shravani, S., Srija, R., and Tajuddin, M. (2023, June). A model to stock price prediction using deep learning. In 2023 International Conference on Sustainable Computing and Smart Systems (ICSCSS) (pp. 242-252). IEEE. https://doi.org/10.1109/ICSCSS57650.2023.10169558
38. Prechelt, L. (2002). Early stopping-but when?. In Neural Networks: Tricks of the trade. Berlin, Heidelberg: Springer Berlin Heidelberg.
39. Pranolo, A., Setyaputri, F. U., Paramarta, A. K. A. I., Triono, A. P. P., Fadhilla, A. F., Akbari, A. K. G., and Uriu, W. (2024). Enhanced Multivariate Time Series Analysis Using LSTM: A Comparative Study of Min-Max and Z-Score Normalization Techniques. ILKOM Jurnal Ilmiah, 16(2), 210-220.
40. Rahman, M. O., Hossain, M. S., Junaid, T. S., Forhad, M. S. A., and Hossen, M. K. (2019). Predicting prices of stock market using gated recurrent units (GRUs) neural networks. Int. J. Comput. Sci. Netw. Secur, 19(1), 213-222.
41. Raşo, H. ve Demirci, M. (2019). Predicting the Turkish Stock Market BIST 30 Index Using Deep Learning. Uluslararası Mühendislik Araştırma ve Geliştirme Dergisi, 11, 7-8.
42. Rezaei, H., Faaljou, H., and Mansourfar, G. (2021). Stock price prediction using deep learning and frequency decomposition. Expert Systems with Applications, 169, 114332. https://doi.org/10.1016/j.eswa.2020.114332
43. Rosenblatt, F. (1958). The perceptron: a probabilistic model for information storage and organization in the brain. Psychological review, 65(6), 386.
44. Rumelhart, D. E., Hinton, G. E., and Williams, R. J. (1986). Learning representations by back-propagating errors. Nature, 323(6088), 533-536. https://doi.org/10.1038/323533a0
45. Sakarya, Ş., ve Yılmaz, Ü. (2019). Derin öğrenme mimarisi kullanarak BİST30 indeksinin tahmini. European Journal of Educational and Social Sciences, 4(2), 106-121.
46. Samii, A., Karami, H., Ghazvinian, H., Safari, A., and Dadrasajirlou, Y. (2023). Comparison of DEEP-LSTM and MLP models in estimation of evaporation pan for arid regions. Journal of Soft Computing in Civil Engineering, 7(2), 155-175.
47. Saud, A. S., and Shakya, S. (2020). Analysis of look back period for stock price prediction with RNN variants: A case study on banking sector of NEPSE. Procedia Computer Science, 167, 788-798. https://doi.org/10.1016/j.procs.2020.03.419
48. Selvin, S., Vinayakumar, R., Gopalakrishnan, E. A., Menon, V. K. and Soman, K. P. (2017). Stock price prediction using LSTM, RNN and CNN-sliding window model. 2017 International Conference on Advances in Computing, Communications and Informatics (ICACCI), Udupi, India, pp. 1643-1647, doi: 10.1109/ICACCI.2017.8126078.
49. Shen, G., Tan, Q., Zhang, H., Zeng, P., and Xu, J. (2018). Deep learning with gated recurrent unit networks for financial sequence predictions. Procedia computer science, 131, 895-903. https://doi.org/10.1016/j.procs.2018.04.298
50. Si, J., Harris, S. L., and Yfantis, E. (2018). A dynamic ReLU on neural network. In 2018 IEEE 13th Dallas Circuits and Systems Conference (DCAS) (pp. 1-6). IEEE.
51. Siami-Namini, S., Tavakoli, N., and Namin, A. S. (2019). A comparison of ARIMA and LSTM in forecasting time series. 17th IEEE International Conference on Machine Learning and Applications (ICMLA), 1394-1401.
52. Şişmanoğlu, G., Koçer, F., Önde, M. A., and Sahingoz, O. K. (2020). Derin öğrenme yöntemleri ile borsada fiyat tahmini. Bitlis Eren Üniversitesi Fen Bilimleri Dergisi, 9(1), 434-445. doi: 10.17798/bitlisfen.571386.
53. Tay, F. E., and Cao, L. (2001). Application of support vector machines in financial time series forecasting. Omega, 29(4), 309-317. https://doi.org/10.1016/S0305-0483(01)00026-3
54. Viadinugroho, R. A. A., and Rosadi, D. (2021, March). Long short-term memory neural network model for time series forecasting: case study of forecasting IHSG during Covid-19 outbreak. In Journal of Physics: Conference Series (Vol. 1863, No. 1, p. 012016). IOP Publishing. https://doi.org/10.1088/1742-6596/1863/1/012016
55. Willmott, C. J., and Matsuura, K. (2005). Advantages of the mean absolute error (MAE) over the root mean square error (RMSE) in assessing average model performance. Climate Research, 30(1), 79-82. . https://doi.org/10.3354/cr030079
56. Yan, X., Weihan, W., and Chang, M. (2021). Research on financial assets transaction prediction model based on LSTM neural network. Neural Computing and Applications, 33(1), 257-270. https://doi.org/10.1007/s00521-020-04992-7
57. Zhang, G., and Zhang, X. (2016). ARMAD-GARCH Stock Price Forecasting Model Based on Differential Information. Systems Engineering-Theory & Practice, 36(05), 1136-1145.
58. Zhang, Q., Yang, L. T., Chen, Z., and Li, P. (2018). A survey on deep learning for big data. Information Fusion, 42, 146-157. https://doi.org/10.1016/j.inffus.2017.10.006