EUVIMEDEUVIMED entdecken
Crossref

LSTM-BASED FORECASTING IN VOLATILE AND DATA-SCARCE ENVIRONMENTS

B. S. Aitbekov, M. Zhartybayeva

Bulletin of Toraighyrov University. Physics & Mathematics series · 30.06.2026 · (2, 2026) · 17-29

DOI: 10.48081/bgqf2030 ↗

Kurz gefasst

Fachartikel in der Zeitschrift Bulletin of Toraighyrov University. Physics & Mathematics series, online erschienen am 30.06.2026. Verfasst von B. S. Aitbekov und 1 weiteren Autor:innen.

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

This research addresses the challenge of exchange rate forecasting in volatile emerging markets. It focuses on the rate between Kazakhstani Tenge (KZT) and the United States Dollar (USD), a market characterized by high volatility and limited data availability. LSTM networks are used often in time series prediction, which includes financial data. While LSTM networks are widely used for time series prediction, their performance under such unstable and data-scarce environments, such as in Kazakhstan, remains unclear. Therefore, a comparative study that focuses on three models: a standalone LSTM model, a CNN LSTM model and a CNN LSTM model with residual correction is proposed. All models are trained on daily KZT and USD exchange rates and Brent oil prices from 2015 to 2024. This period includes major economic events, which increase volatility and provide a strict test for forecasting models. The empirical results show that the standalone LSTM model achieves the best performance. RMSE equals 4.8837, and MAE equals 3.7555. These results are significantly better than the errors of the hybrid models. The CNN LSTM model shows an RMSE of 8.2236, which indicates weaker performance. The residual correction model shows an RMSE of 5.3163, which places it between the two models. The results suggest that simple models may perform better in volatile and data constrained environments. This may be because higher complexity captures noise rather than useful patterns. In other words, carefully tuned simpler architectures may provide more reliable predictions in emerging market conditions. Keywords: LSTM, Time Series Forecasting, Exchange Rate Prediction, KZT/USD, Emerging Markets.

Zitieren

APA

Aitbekov, B. S., & Zhartybayeva, M. (2026). LSTM-BASED FORECASTING IN VOLATILE AND DATA-SCARCE ENVIRONMENTS. Bulletin of Toraighyrov University. Physics & Mathematics series, 17-29. https://doi.org/10.48081/bgqf2030

Vancouver

Aitbekov BS, Zhartybayeva M. LSTM-BASED FORECASTING IN VOLATILE AND DATA-SCARCE ENVIRONMENTS. Bulletin of Toraighyrov University. Physics & Mathematics series. 2026:17-29. doi:10.48081/bgqf2030

BibTeX
@article{Aitbekov2026,
  author = {Aitbekov, B. S. and Zhartybayeva, M.},
  title = {LSTM-BASED FORECASTING IN VOLATILE AND DATA-SCARCE ENVIRONMENTS},
  journal = {Bulletin of Toraighyrov University. Physics & Mathematics series},
  year = {2026},
  number = {2, 2026},
  pages = {17--29},
  doi = {10.48081/bgqf2030}
}
Kostenlos starten →