In this research, the prediction power ofthe two popular machine learning models, random forest model and gradientboosting model is compared in the case of forecasting of cryptocurrency price. Therandom forest model and gradient boosting model are well known for accurate predictionon structure data. For this research, the data of XBTUSD, which is XBT/USDperpetual contract listed on BITMEX, one of the largest cryptocurrencyexchanges in the world; XBTUSD is a derivative product that closely followsBitcoin price in the spot market. In order to compare the prediction power, thetwo models are designed to be trained with the daily data of XBTUSD price from January1, 2017 to December 31, 2021 for the first prediction. The training processincludes cross validation and grid search. The data used for the trainingincludes open, close, low, close, trading volume, and RSI (relative strengthindex) data. After the training, the two models make predictions on the next-dayprice of XBTUSD for the period from January 1, 2022 to March 31, 2022. RMSE (RootMean Square Error) and MAPE (Mean Absolute Percentage Error) are used to testthe accuracy of the two models. For the period from January 1, 2022 to March31, 2022, RMSE and MAPE for the random forest model are 2348.46 and 4.19%, andthose for the gradient boosting model are 2136.78 and 3.95%. RMSE and MAPE fromthis research suggest that the gradient boosting model has higher predictionpower than the random forest model in the case of prediction of price in cryptocurrencymarket.