Examining and Comparing the Efficiency of MLP and SimpleRNN Algorithms in Cryptocurrency Price Prediction

Abstract

Cryptocurrencies have been widely identified and established as a new form of electronic currency exchange, carrying significant implications for emerging economies and the global economy. This research focused on the "examination and comparison of the efficiency of MLP and SimpleRNN algorithms in predicting cryptocurrency prices" using the Python programming language. Price predictions for Bitcoin, Ethereum, Binance Coin, Cardano, and Ripple were made using two deep learning algorithms (including the MLP algorithm and the SimpleRNN algorithm) over the period from 2017 to 2023. The results of cryptocurrency price prediction using deep learning algorithms were satisfactory; and the comparison of predictions across all cryptocurrencies indicated minimal differences between the algorithms studied, suggesting that they were efficient and had low error rates. Based on the obtained results regarding Bitcoin price prediction, the best algorithm was SimpleRNN; for Ethereum price prediction, the best algorithm was MLP; for Binance Coin price prediction, the best algorithm was SimpleRNN; for Cardano price prediction, the best algorithm was MLP; and for Ripple price prediction, the best algorithm was MLP.