MIT researchers develop new startup investment model that consistently beats returns of top VC firms

CAMBRIDGE, MASS., July 2017 – Researchers at MIT’s Sloan School of Management have developed a new quantitative model -- using data compiled from thousands of startup companies -- that can better pick venture-capital portfolio winners and lead to lucrative exit rates that are nearly double that of top venture capital firms.

The key to the new model, as outlined in a new paper, "Picking Winners: A Framework For Venture Capital Investment," by MIT Sloan Prof. Tauhid Zaman and doctorate student David Scott Hunter, is estimating the volatility of startup investments combined with the “drift” of latent valuations as each round of funding unfolds for startups. The goal is to improve upon the current “top heavy” model of venture capital investments, in which one winning investment more than makes up for numerous other disappointing investment outcomes.

In their research of tens of thousands of startup companies from 2000 to 2015, the authors found that the more volatile a startup, the more likely its success – and the more volatile an overall portfolio, the more likely VC firms will hit winners that payoff when investors exit their investments via IPOs or acquisitions of startup companies.

“You should actually like volatility,” said Zaman. “You don’t want to play it safe. That’s what we found: Volatility is good. The bigger the risk up or down, the bigger the potential payout.”

In a way, the current VC model is similar to the business approach used by pharmaceutical companies when choosing drugs to develop or studios when selecting movies to produce: Everyone is looking for those one or two true winners among many candidates, Zaman said. So the key is trying to identify as many of those potential winners as early as possible.

Most successful venture capital firms already do a variation of this as they research every aspect of startups before investing – identifying and assessing the quality of ideas, the backgrounds of founders, the experience of startup employees, the success-rate of early stage investors, the exit rates within certain industry sectors and subsectors.

But Zaman and Hunter say they take much of the same information, assign them quantitative weights, and apply their model to the data to estimate potential investment results. They note that the very top venture capital firms hit exits of some sort in three out of 10 investments. But the Zaman-Hunter model produced six out of ten winners, or a 60 percent “winner” rate.

“We’re not trying to reinvent the investment wheel here with our model,” said Zaman. “We’re just trying to use analytics to provide better investment decisions and allow VC firms to do a better job.”