Math Formula Picks Startups Worth Funding
For multimillion-dollar projects, the ideal co-investor lets a computer program make its picks—at least according to Correlation Ventures, a venture capital firm that uses predictive software to help choose startups to fund. The company claims it's the first to do so.
The decisions aren't entirely robotic. Startups looking for Correlation funding send in five basic documents that companies always have, according to the venture firm's website. Correlation feeds information from those documents into its computer model, which gives the startup a score, similar to a credit rating. Top scorers earn interviews with the venture firm that's leading the deal. Correlation still depends on a partnering firm to perform background checks and make other human decisions, reported MIT's Technology Review.
Correlation Ventures has invested in 26 companies so far, but it's too soon to know if its approach works, Technology Review reported. In an interview and on its website, the ventures firm says its algorithm can "tilt the odds" in its favor.
This data-driven investing means some changes in how deals work. For one, they happen much faster. Correlation Ventures' website says it makes decisions within two weeks, when high-tech deals usually take months.
The company's algorithm depends on its proprietary database, which includes details on 20 years of financial deals, including each deal's terms, investors, board of directors and outcomes. That kind of information isn't publicly available, so Correlation Ventures spent years contacting ventures firms and companies individually.
High-tech industries usually love "big data," so it's surprising to learn that the million-dollar deals that fund high tech didn't use more mathematical models before. But it might have been impossible to build these models before, because the venture capital industry didn't have long enough of a history, Harvard Business School professor and Correlation Ventures advisor Matthew Rhodes-Kropf told Technology Review.