The cryptocurrency market is increasingly exerting influence on the global economy, however, the prevalence of price volatility serves as a hindrance to the entry of new participants into its commercial dynamics. This article undertakes an assessment of the efficacy of predictive models, specifically various machine learning algorithms, including Extreme Learning Machine, Support Vector Regression, Gradient Boosting Machine, Extreme Gradient Boosting, and Random Forest. The focus of existing literature predominantly centers on well-established cryptocurrencies such as Bitcoin and Ether, thereby neglecting a comprehensive examination of alternative assets as the cryptocurrency of this study, AXS. Performance evaluation in the test set involved the utilization of key metrics, namely the root mean square error, the coefficient of determination squared, the mean absolute error, the mean absolute percentage error, and the standard deviation. Within forecast horizons ranging from 1 to 4, the Extreme Learning Machine model demonstrated notable accuracy, yielding error percentages of 3.82%, 5.18%, 6.53%, and 7.49%, respectively. In horizon 5, the Gradient Boosting Machine emerged as the optimal model with a percentage error of 8.22%, while a less precise prediction was observed in horizon 10, where the lowest percentage error reached 10%. Conversely, the Support Vector Regression model exhibited sub-optimal performance among the machine learning algorithms, registering a percentage error of 12% in the initial forecasting horizon. This underscores the challenges inherent in employing certain models for predicting cryptocurrency values, emphasizing the need for a nuanced understanding of the strengths and limitations of each approach.