ANALYSIS OF FINANCIAL RETURN PREDICTABILITY USING A HYBRID CNN- LSTM NEURAL ARCHITECTURE
Keywords:
financial time series, return forecasting, neural networks, CNN-LSTM, deep learning, volatility, S&P 500Abstract
Thispaper investigatesthepredictabilityoffinancial time seriesreturnsusingtheS&P500 index as an empirical example. Financial markets are characterized by high stochasticity, weak autocorrelation of returns, and volatility clustering, which significantly complicates the construction of reliable forecasting models. The objective of the study is to empirically test the hypothesis of the existence of an extractable predictive signal in return dynamics using a hybrid CNN-LSTM neural architecture.
The empirical dataset consists of daily observations of the S&P 500 index covering the period from 2005 to 2025. Logarithmic returns and a set of derived features describing short-term price dynamics, intraday volatility, and trading activity are constructed from the raw data. The forecasting model is based on a hybrid architecture combining convolutional neural networks for extracting local temporal patterns and LSTM layers for modeling sequential dependencies.
Experimental evaluation is conducted in regression, classification, and probabilistic forecasting settings. The obtained results allow assessing the presence of predictive structure in return dynamics and identifying the limitations of deep neural networks when applied to financial time series forecasting.
References
Тимиркаев Д.А. Моделирование волатильности многомерных финансовых временных рядов // Экономический анализ: теория и практика. 2010. № 8. С. 25-31. EDN: KZUDSX.
Campbell J.Y., LoA. W., MacKinlay A.C. The Econometrics of Financial Markets. Princeton: Princeton University Press, 1997. 611 p.
Sezer O.B., Gudelek M.U., OzbayogluA.M. Financial Time Series Forecasting with DeepLearning: A Systematic Literature Review // Applied Soft Computing. 2020. Vol. 90. DOI: 10.1016/j.asoc.2020.106181 EDN: UHBFUI.
Hochreiter S., Schmidhuber J. Long Short-Term Memory // Neural Computation. 1997. Vol. 9. no. 8. P. 1735-1780.
Fama E.F. Efficient Capital Markets: A Review of Theory and Empirical Work // The Journal of Finance. 1970. Vol. 25. no. 2. P. 383-417.
Engle R.F. Autoregressive Conditional Heteroskedasticity with Estimates of the Variance of United Kingdom Inflation // Econometrica. 1982. Vol. 50. no. 4. P. 987-1007.
Кузнецова Л.Г. Экскурс в теорию случайных блужданий и её использование для оценки стоимости финансовых активов // Финансы и кредит. 2005. № 28. С. 2-10.
