n this modern era of technology, the more secured ways are needed to
deal with financial investments or transactions. Cryptocurrency can be
named as one of the solutions for this concern. Cryptocurrency is a
digital payment system that doesn’t rely on banks to verify transactions.
A digital payment system called cryptocurrency doesn’t rely on banks
to validate transactions. Anyone can send and receive funds using this
method. Payments made using cryptocurrencies only exist as digital
records in an online database that detail specific transactions. This new
sort of investment is providing vast areas for research to the
researchers. By predicting its price this can be as more efficient asset
for investment. Much research is going on in this area. This paper
proposes two different recurrent neural network (RNN) algorithms to
predict prices of cryptocurrency namely Bit coin and they are Long
short-term memory (LSTM) and Gated Recurrent Unit (GRU). the
measures being used in this paper to assess the accuracy of the used
algorithms are mean squared error (MSE), Mean Absolute Percentage
Error (MAPE), Root Mean Squared Error (RMSE), and Mean
Absolute Error (MAE), are also used to assess different prediction
algorithms. Comparisons are carried out on the basis of three datasets
training, testing, and validation. The loss and evaluation functions are
based on the mean squared error. The model performs better the lower
the value. Based on findings the GRU model outperforms the LSTM
algorithm in terms of accuracy and reliability in predicting
cryptocurrency prices, but both algorithms produce excellent outcome.