Course project on Dota2 for CSE255 under Prof. Julian McAuley This is a survey of existing simplistic Regression methods for predicting the success of a DotA2 match, a wildly successful RPG game based off a custom Warcraft III map created almost a decade ago. It is played by millions of players worldwide and has regularly hosted million dollar prize pool tournaments, courtesy of Valve Corporation. A novel genetic algorithmic approach is also applied to this prediction task to determine the "successful" combination of heroes that may lead to a higher probability of victory and factor this into our regresion model to see if it performs better.