In this paper, we critically evaluate the Capital Asset Pricing Model (CAPM) and its limitations in predicting future returns using Linear Regression (LR) models. We propose an alternative approach, Bayesian Regression, which offers a more informative and accurate prediction framework. Our study compares the performance of LR and Bayesian Regression models in forecasting the returns of popular cryptocurrencies, Doge (for asset) and Bitcoin (for market). Through the use of Mean Squared Error (MSE), we demonstrate that the Bayesian Regression model outperforms the LR model in terms of prediction accuracy. The findings highlight the advantages of Bayesian methods in capturing the complex relationships and uncertainties inherent in financial markets. Our research contributes to the ongoing discourse on investment decision-making, providing valuable insights into the effectiveness of Bayesian Regression in the context of cryptocurrency investments.