The currency exchange rate is a crucial link between all countries related to economic and trade activities. The non-linear and non-stationary nature of financial time series data poses significant challenges for standalone statistical and neural network methods. While modeling in finance often focuses on volatility, there is a notable lack of research on modeling actual prices, particularly in the Kenyan exchange rates. The research used daily USD/KES and EUR/KES exchange rate data from the CBK ranging from November 10, 2008 to March 1, 2024 totaling 5409 entries.The research employs the GARCH model to extract statistical properties, which are then combined with historical daily exchange rate prices and fed into LSTM, and Transformer models leading to GARCH-LSTM, GARCH-Transformer hybrid models. Results indicate that hybrid model GARCH-Transformer, outperform the standalone models.This integration of GARCH with Transformer model offers a more robust framework for modeling actual prices.
Chamakany et al. (Thu,) studied this question.