An article TechCrunch published on December 21st highlights a new venture capital fund in Berlin, Fly Ventures, that is utilizing machine learning to deal source new companies.  In the past, I have heard of venture funds utilizing mathematical modeling to run Monte-Carlo simulations that estimate the necessary portfolio size to be profitable with a random set of investments (bigger portfolio averages return), but I have never seen a fund utilize machine learning for sourcing new investments. Essentially, they have created a machine learning algorithm that scrapes the internet (blogs, job boards, accelerators, CrunchBase, etc.) and then ranks them. The article states that the program presents the results, “in a Tinder-like interface that lets [Fly Ventures] quickly decide whether [they] want to speak to a company.” The company then proceeds to swipe yes or no, and ideally (the article does not clarify) the machine learning algorithm learns from their preferences and tunes the future list to show it more-likely investments first.

Although this fund was founded recently in 2016, the results seem promising. According to the article, 60% of the startups they speak to are approached cold. This means that they had no prior connection to the company, and the algorithm independently found it online. This is a very high statistic. Typically, deal sourcing comes from a connection through a venture capital firm’s network or the founders of a startup company come looking for funding. By using machine learning, Fly Ventures has turned the paradigm around, and they are on the “hunt” looking for investments that no one else has found yet. This likely gives them a competitive advantage because Fly Ventures can get a better deal on startups that are very early in their lifecycle that also fit their investment thesis. It was tough to find very thorough reporting on their fund, but I am interested in keeping up to date on their successes in the future.

Article Citation:

O’Hear, Steve. “Fly Ventures, a Berlin-based VC using machine learning to find its next deal, closes $41M fund.” TechCrunch. December 21, 2017. Accessed January 16, 2018.