An Initial Coin Offering (ICO) is a method of raising funds for digital currency projects. Investors purchase these coins at a very low initial price before they are released. These coins are then listed on the trading platform, and their prices may increase rapidly if the currency performs well. After six months of release, ICO evaluation is the expected time for investors to profit. A dataset consisting of 109 ICOs was constructed from reputable websites after data preprocessing. Correlation analysis of 12 inputs revealed issues of multicollinearity, leading to biased regression model results. Overfitting occurred when using the regression model. To address these limitations, the Ridge regression method resolved the issues with the ICO data. An artificial neural network model addressed the complex nonlinear relationships between inputs and ICO prices. By adjusting parameters to achieve the best performance according to the Root Mean Square Error, R-squares, and Mean Absolute Error metrics, the results showed that the Ridge regression algorithm with a test set of three ICOs achieved accuracy ranging from 63% to 92% of ICO prices, while the artificial neural network model predicted with 98% accuracy depending on the metric used.