The Impact of Varying Window Sizes on the Performance of a Hybrid CNN-LSTM Model A Case of Forecasting USD VS KES

Crispus Waweru Mwangi; Martin Mutwiri Kithinji; Mutua Kilai

doi:doi:10.11648/j.ijdsa.20261205.13

Research Article |

| Peer-Reviewed

The Impact of Varying Window Sizes on the Performance of a Hybrid CNN-LSTM Model A Case of Forecasting USD VS KES

Crispus Waweru Mwangi^*

, Martin Mutwiri Kithinji

, Mutua Kilai

Published in International Journal of Data Science and Analysis (Volume 12, Issue 5)

Received: 18 August 2026 Accepted: 4 September 2026 Published: 28 September 2026

Views: Downloads:

Download PDF

Share This Article

Twitter
Linked In
Facebook

Abstract

Financial time series are characterized by non-linearity, inherent volatility, and changing temporal patterns making accurate predictions very challenging. This study deploys a CNN-LSTM hybrid model to predict the USD/KES exchange rate giving particular emphasis to the effect the size of the historical input window has on the hybrid model forecasting performance. The study relied on Daily USD/KES exchange-rate data obtained from the Central Bank of Kenya to train, evaluate and validate the hybrid model. This data was prepossessed and transformed into sequential inputs using different window configurations, allowing the models to learn from varying amounts of historical information. The study combined a CNN-based local pattern extraction with an LSTM-based temporal dependency learning model to develop a hybrid CNN-LSTM model. The hybrid's model performance was then subjected to a comparison with standalone CNN and LSTM model and evaluated using Root Mean Square Error (RMSE) and Mean Absolute Error (MAE) The hybrid CNN-LSTM prediction results indicated that the model was more effective in prediction among the evaluated architecture. The choice of the input window sizes affected the model's ability to learn and capture trends in the time series data. The results from the rolling-window experiments demonstrated that shorter and moderately sized historical windows can provide a more responsive representation of recent exchange-rate movements than an expanding window, particularly during periods of pronounced market variation. These findings highlight the importance of considering both model architecture and input-window design when developing deep-learning approaches for USD/KES exchange-rate forecasting. The study provides an empirical basis for selecting appropriate historical window sizes when applying hybrid deep-learning models to volatile financial time series in the Kenyan foreign-exchange market.

Published in	International Journal of Data Science and Analysis (Volume 12, Issue 5)
DOI	10.11648/j.ijdsa.20261205.13
Page(s)	129-138
Creative Commons	This is an Open Access article, distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution and reproduction in any medium or format, provided the original work is properly cited.
Copyright	Copyright © The Author(s), 2026. Published by Science Publishing Group

Keywords

USD/KES Exchange Rate, Exchange-rate Forecasting, CNN-LSTM, Deep Learning, Time-series Forecasting, Rolling Window, Window Size, Financial Time Series

References

[1]	Sezer, O. B., Gudelek, M. U., & Ozbayoglu, A. M. (2020). Financial time series forecasting with deep learning: A systematic literature review: 2005-2019. Applied soft computing, 90, 106181.
[2]	Tang, Y., Song, Z., Zhu, Y., Yuan, H., Hou, M., Ji, J., ... & Li, J. (2022). A survey on machine learning models for financial time series forecasting. Neurocomputing, 512, 363-380.
[3]	Musyoki, D., Pokhariyal, G. P., & Pundo, M. (2012). The impact of real exchange rate volatility on economic growth: Kenyan evidence. Business and Economic Horizons, 7(1), 59-75.
[4]	Nyambariga, M. D. (2017). Effects of exchange rate volatility on imports and exports in Kenya. International Journal of Economics, 2(3), 71-84.
[5]	Zhang, G. P. (2003). Time series forecasting using a hybrid ARIMA and neural network model. Neurocomputing, 50, 159-175.
[6]	Aryal, S., Nadarajah, D., Kasthurirathna, D., Rupasinghe, L., & Jayawardena, C. (2019, December). Comparative analysis of the application of Deep Learning techniques for Forex Rate prediction. In 2019 international conference on advancements in computing (ICAC) (pp. 329-333). IEEE.
[7]	Dautel, A. J., Härdle, W. K., Lessmann, S., & Seow, H. V. (2020). Forex exchange rate forecasting using deep recurrent neural networks. Digital Finance, 2(1), 69-96.
[8]	Biswas, A., Uday, I. A., Rahat, K. M., Akter, M. S., & Mahdy, M. R. C. (2023). Forecasting the United State Dollar (USD)/Bangladeshi Taka (BDT) exchange rate with deep learning models: Inclusion of macroeconomic factors influencing the currency exchange rates. PloS one, 18(2), e0279602.
[9]	Azlan, A., Yusof, Y., & Mohsin, M. F. M. (2019). Determining the impact of window length on time series forecasting using deep learning. International Journal of Advanced Computer Research, 9(44), 260-267.
[10]	Yu, L., Wang, S., & Lai, K. K. (2005). Adaptive smoothing neural networks in foreign exchange rate forecasting. In International conference on computational science (pp. 523530).
[11]	de Prado, M. L. (2018). Advances in nancial machine learning. Hoboken, NJ: John Wiley & Sons.
[12]	Patel, J., Shah, S., Thakkar, P., & Kotecha, K. (2015). Predicting stock and stock price index movement using trend deterministic data preparation and machine learning techniques. Expert Systems with Applications, 42(1), 259268.
[13]	Bollerslev, T. (1986). Generalized autoregressive conditional heteroskedasticity. Journal of Econometrics, 31(3), 307327.
[14]	Hu, Z., Zhao, Y., & Khushi, M. (2021). A survey of Forex and stock price prediction using deep learning. Applied System Innovation, 4(1), 9.
[15]	Hochreiter, S., & Schmidhuber, J. (1997). Long short-term memory. Neural computation, 9(8), 1735-1780.

Cite This Article

Plain Text BibTeX RIS

APA Style

Mwangi, C. W., Kithinji, M. M., Kilai, M. (2026). The Impact of Varying Window Sizes on the Performance of a Hybrid CNN-LSTM Model A Case of Forecasting USD VS KES. International Journal of Data Science and Analysis, 12(5), 129-138. https://doi.org/10.11648/j.ijdsa.20261205.13

Copy | Download

ACS Style

Mwangi, C. W.; Kithinji, M. M.; Kilai, M. The Impact of Varying Window Sizes on the Performance of a Hybrid CNN-LSTM Model A Case of Forecasting USD VS KES. Int. J. Data Sci. Anal. 2026, 12(5), 129-138. doi: 10.11648/j.ijdsa.20261205.13

Copy | Download

AMA Style

Mwangi CW, Kithinji MM, Kilai M. The Impact of Varying Window Sizes on the Performance of a Hybrid CNN-LSTM Model A Case of Forecasting USD VS KES. Int J Data Sci Anal. 2026;12(5):129-138. doi: 10.11648/j.ijdsa.20261205.13

Copy | Download

@article{10.11648/j.ijdsa.20261205.13,
  author = {Crispus Waweru Mwangi and Martin Mutwiri Kithinji and Mutua Kilai},
  title = {The Impact of Varying Window Sizes on the Performance of a Hybrid CNN-LSTM Model A Case of Forecasting USD VS KES},
  journal = {International Journal of Data Science and Analysis},
  volume = {12},
  number = {5},
  pages = {129-138},
  doi = {10.11648/j.ijdsa.20261205.13},
  url = {https://doi.org/10.11648/j.ijdsa.20261205.13},
  eprint = {https://article.sciencepublishinggroup.com/pdf/10.11648.j.ijdsa.20261205.13},
  abstract = {Financial time series are characterized by non-linearity, inherent volatility, and changing temporal patterns making accurate predictions very challenging. This study deploys a CNN-LSTM hybrid model to predict the USD/KES exchange rate giving particular emphasis to the effect the size of the historical input window has on the hybrid model forecasting performance. The study relied on Daily USD/KES exchange-rate data obtained from the Central Bank of Kenya to train, evaluate and validate the hybrid model. This data was prepossessed and transformed into sequential inputs using different window configurations, allowing the models to learn from varying amounts of historical information. The study combined a CNN-based local pattern extraction with an LSTM-based temporal dependency learning model to develop a hybrid CNN-LSTM model. The hybrid's model performance was then subjected to a comparison with standalone CNN and LSTM model and evaluated using Root Mean Square Error (RMSE) and Mean Absolute Error (MAE) The hybrid CNN-LSTM prediction results indicated that the model was more effective in prediction among the evaluated architecture. The choice of the input window sizes affected the model's ability to learn and capture trends in the time series data. The results from the rolling-window experiments demonstrated that shorter and moderately sized historical windows can provide a more responsive representation of recent exchange-rate movements than an expanding window, particularly during periods of pronounced market variation. These findings highlight the importance of considering both model architecture and input-window design when developing deep-learning approaches for USD/KES exchange-rate forecasting. The study provides an empirical basis for selecting appropriate historical window sizes when applying hybrid deep-learning models to volatile financial time series in the Kenyan foreign-exchange market.},
 year = {2026}
}

Copy | Download

TY  - JOUR
T1  - The Impact of Varying Window Sizes on the Performance of a Hybrid CNN-LSTM Model A Case of Forecasting USD VS KES
AU  - Crispus Waweru Mwangi
AU  - Martin Mutwiri Kithinji
AU  - Mutua Kilai
Y1  - 2026/09/28
PY  - 2026
N1  - https://doi.org/10.11648/j.ijdsa.20261205.13
DO  - 10.11648/j.ijdsa.20261205.13
T2  - International Journal of Data Science and Analysis
JF  - International Journal of Data Science and Analysis
JO  - International Journal of Data Science and Analysis
SP  - 129
EP  - 138
PB  - Science Publishing Group
SN  - 2575-1891
UR  - https://doi.org/10.11648/j.ijdsa.20261205.13
AB  - Financial time series are characterized by non-linearity, inherent volatility, and changing temporal patterns making accurate predictions very challenging. This study deploys a CNN-LSTM hybrid model to predict the USD/KES exchange rate giving particular emphasis to the effect the size of the historical input window has on the hybrid model forecasting performance. The study relied on Daily USD/KES exchange-rate data obtained from the Central Bank of Kenya to train, evaluate and validate the hybrid model. This data was prepossessed and transformed into sequential inputs using different window configurations, allowing the models to learn from varying amounts of historical information. The study combined a CNN-based local pattern extraction with an LSTM-based temporal dependency learning model to develop a hybrid CNN-LSTM model. The hybrid's model performance was then subjected to a comparison with standalone CNN and LSTM model and evaluated using Root Mean Square Error (RMSE) and Mean Absolute Error (MAE) The hybrid CNN-LSTM prediction results indicated that the model was more effective in prediction among the evaluated architecture. The choice of the input window sizes affected the model's ability to learn and capture trends in the time series data. The results from the rolling-window experiments demonstrated that shorter and moderately sized historical windows can provide a more responsive representation of recent exchange-rate movements than an expanding window, particularly during periods of pronounced market variation. These findings highlight the importance of considering both model architecture and input-window design when developing deep-learning approaches for USD/KES exchange-rate forecasting. The study provides an empirical basis for selecting appropriate historical window sizes when applying hybrid deep-learning models to volatile financial time series in the Kenyan foreign-exchange market.
VL  - 12
IS  - 5
ER  -

Copy | Download

Author Information

Crispus Waweru Mwangi

Department of Mathematics,Kirinyaga University, Kerugoya, Kenya

Contact Email

http://orcid.org/0009-0005-0275-4257
Martin Mutwiri Kithinji

Department of Mathematics,Kirinyaga University, Kerugoya, Kenya; Department of Mathematics, Pan African Insitute of Basic Science, Technology and Innovation, Nairobi, Kenya

Contact Email

http://orcid.org/0000-0002-0449-3776
Mutua Kilai

Department of Mathematics,Kirinyaga University, Kerugoya, Kenya

Contact Email

http://orcid.org/0000-0001-6290-9728

Download PDF

Submit an Article

Sections

Plain Text BibTeX RIS

APA Style

Mwangi, C. W., Kithinji, M. M., Kilai, M. (2026). The Impact of Varying Window Sizes on the Performance of a Hybrid CNN-LSTM Model A Case of Forecasting USD VS KES. International Journal of Data Science and Analysis, 12(5), 129-138. https://doi.org/10.11648/j.ijdsa.20261205.13

Copy | Download

ACS Style

Mwangi, C. W.; Kithinji, M. M.; Kilai, M. The Impact of Varying Window Sizes on the Performance of a Hybrid CNN-LSTM Model A Case of Forecasting USD VS KES. Int. J. Data Sci. Anal. 2026, 12(5), 129-138. doi: 10.11648/j.ijdsa.20261205.13

Copy | Download

AMA Style

Mwangi CW, Kithinji MM, Kilai M. The Impact of Varying Window Sizes on the Performance of a Hybrid CNN-LSTM Model A Case of Forecasting USD VS KES. Int J Data Sci Anal. 2026;12(5):129-138. doi: 10.11648/j.ijdsa.20261205.13

Copy | Download

@article{10.11648/j.ijdsa.20261205.13,
  author = {Crispus Waweru Mwangi and Martin Mutwiri Kithinji and Mutua Kilai},
  title = {The Impact of Varying Window Sizes on the Performance of a Hybrid CNN-LSTM Model A Case of Forecasting USD VS KES},
  journal = {International Journal of Data Science and Analysis},
  volume = {12},
  number = {5},
  pages = {129-138},
  doi = {10.11648/j.ijdsa.20261205.13},
  url = {https://doi.org/10.11648/j.ijdsa.20261205.13},
  eprint = {https://article.sciencepublishinggroup.com/pdf/10.11648.j.ijdsa.20261205.13},
  abstract = {Financial time series are characterized by non-linearity, inherent volatility, and changing temporal patterns making accurate predictions very challenging. This study deploys a CNN-LSTM hybrid model to predict the USD/KES exchange rate giving particular emphasis to the effect the size of the historical input window has on the hybrid model forecasting performance. The study relied on Daily USD/KES exchange-rate data obtained from the Central Bank of Kenya to train, evaluate and validate the hybrid model. This data was prepossessed and transformed into sequential inputs using different window configurations, allowing the models to learn from varying amounts of historical information. The study combined a CNN-based local pattern extraction with an LSTM-based temporal dependency learning model to develop a hybrid CNN-LSTM model. The hybrid's model performance was then subjected to a comparison with standalone CNN and LSTM model and evaluated using Root Mean Square Error (RMSE) and Mean Absolute Error (MAE) The hybrid CNN-LSTM prediction results indicated that the model was more effective in prediction among the evaluated architecture. The choice of the input window sizes affected the model's ability to learn and capture trends in the time series data. The results from the rolling-window experiments demonstrated that shorter and moderately sized historical windows can provide a more responsive representation of recent exchange-rate movements than an expanding window, particularly during periods of pronounced market variation. These findings highlight the importance of considering both model architecture and input-window design when developing deep-learning approaches for USD/KES exchange-rate forecasting. The study provides an empirical basis for selecting appropriate historical window sizes when applying hybrid deep-learning models to volatile financial time series in the Kenyan foreign-exchange market.},
 year = {2026}
}

Copy | Download

TY  - JOUR
T1  - The Impact of Varying Window Sizes on the Performance of a Hybrid CNN-LSTM Model A Case of Forecasting USD VS KES
AU  - Crispus Waweru Mwangi
AU  - Martin Mutwiri Kithinji
AU  - Mutua Kilai
Y1  - 2026/09/28
PY  - 2026
N1  - https://doi.org/10.11648/j.ijdsa.20261205.13
DO  - 10.11648/j.ijdsa.20261205.13
T2  - International Journal of Data Science and Analysis
JF  - International Journal of Data Science and Analysis
JO  - International Journal of Data Science and Analysis
SP  - 129
EP  - 138
PB  - Science Publishing Group
SN  - 2575-1891
UR  - https://doi.org/10.11648/j.ijdsa.20261205.13
AB  - Financial time series are characterized by non-linearity, inherent volatility, and changing temporal patterns making accurate predictions very challenging. This study deploys a CNN-LSTM hybrid model to predict the USD/KES exchange rate giving particular emphasis to the effect the size of the historical input window has on the hybrid model forecasting performance. The study relied on Daily USD/KES exchange-rate data obtained from the Central Bank of Kenya to train, evaluate and validate the hybrid model. This data was prepossessed and transformed into sequential inputs using different window configurations, allowing the models to learn from varying amounts of historical information. The study combined a CNN-based local pattern extraction with an LSTM-based temporal dependency learning model to develop a hybrid CNN-LSTM model. The hybrid's model performance was then subjected to a comparison with standalone CNN and LSTM model and evaluated using Root Mean Square Error (RMSE) and Mean Absolute Error (MAE) The hybrid CNN-LSTM prediction results indicated that the model was more effective in prediction among the evaluated architecture. The choice of the input window sizes affected the model's ability to learn and capture trends in the time series data. The results from the rolling-window experiments demonstrated that shorter and moderately sized historical windows can provide a more responsive representation of recent exchange-rate movements than an expanding window, particularly during periods of pronounced market variation. These findings highlight the importance of considering both model architecture and input-window design when developing deep-learning approaches for USD/KES exchange-rate forecasting. The study provides an empirical basis for selecting appropriate historical window sizes when applying hybrid deep-learning models to volatile financial time series in the Kenyan foreign-exchange market.
VL  - 12
IS  - 5
ER  -

Copy | Download