What Predicts Venture Capital Investment and What Should? A Machine Learning Approach
Diag Davenport, Behavioral Science PhD student
Venture capital is an important institution that deploys billions of dollars each year to startups around the world. Thus, researchers are justifiably interested in both positive and normative explorations of VC investment decisions. Prior work has shown that VCs have attractively high return rates, suggesting that there is some efficiency in how capital is allocated. More recent work suggests that VC decisions are highly influenced by the quality of the management. However, it isn’t obvious that firms should be taking signal from the management when making allocation decisions. Using a novel dataset from Pitchbook, I will build two flexible machine learning models that will: (1) identify predictors of VC investment decisions and (2) identify predictors of startup success. By comparing the sets of predictors, I hope to gain insight on the extent to which investors mis- weight startup cues.