Researchers at the Massachusetts Institute of Technology (MIT) have developed an algorithmic scheme for venture capital investment that pairs the unpredictabilty of Brownian motion with large datasets about startup founders, investors, and performance.
MIT's David Scott Hunter and Tauhid Zaman accumulated data on more than 83,000 companies culled from startup databases Crunchbase and Pitchbook, and then correlated it on LinkedIn. The data left them with a set of startup properties or features that could be fed into a predictive model.
A study found the model can achieve 60% exit rates on optimized investment portfolios, which is about twice the rate of top venture capital firms.
"Our analysis showed that the things [venture capitalists] look for are the things we found that are the keys to success," Zaman says.
He thinks the model's success in prediction is keyed to how it measures factors such as founder experience, versus how investors measure it in the real world.
View Full Article
Abstracts Copyright © 2017 Information Inc., Bethesda, Maryland, USA