Background: With the development and change of the surrounding world, earning and trading have also transformed, so Bitcoin and other similar cryptocurrencies can be an alternative to the common currencies of the world in the future. Therefore, its price prediction will be essential in investing in these cryptocurrencies. Researchers have proposed various methods to improve the accuracy of Bitcoin price prediction. Examining the results of these approaches shows that no significant result has been achieved in predicting the price of Bitcoin.

Objective: In this research, relying on machine learning and data mining, new models have been proposed to predict the price of Bitcoin.

Methodology: The core of the proposed models is the XGBoost algorithm, which is in the category of meta-learning algorithms. To increase the efficiency of the XGBoost parameter setting model in the form of an optimization problem, the definition and values of the target parameters were searched using the Harris hawks algorithm. The second proposed model is also based on the LSTM algorithm, which combines the XGBoost algorithm with the LSTM algorithm.

Findings and Originality: The first proposed model reaches the highest prediction accuracy with an R2 index of 99.32 for 30 falcons. The second proposed model with 70 LSTM blocks reaches the highest level of accuracy with an R2 index of 74.86. The first proposed model is more accurate in price prediction than the second proposed model and other models, so with the RMSE index, the error rate in the first proposed model reaches 0.0086.