This thesis consists of the creation of a reliable model to predict the success of Startups located in U.S. Previous researches have been focused either on the accuracy of different algorithms, without investigating the real effect of the risk and success factors, or on explaining the effects of those factors for Startup success/failure, such as the education of the entrepreneurs, financing methods, and timing. Another difference is that this research is focused only on Startups located in the U.S. This research combines the findings of many studies of the determinants of startup success and models focused on comparing different predictive methods, like Logistic Regressions, Linear Discrimination Analysis and Machine Learning Algorithms. The final output is a reliable and predictive model that could help Angel Investors and Venture Capitalists to build more consistent and measurable startup portfolios.