Can an Algorithm Predict if You Will Start a Successful Company?The Algorithm is Still Working On It...
Algorithms are the new Nostradamus. They are predicting all sorts of things these days — whether you might quit, election results using social media feeds, and whether you are pregnant.
So it’s no surprise that in the age of startups the latest predictive algorithm comes from Bloomberg Beta (we love that their manual is hosted on github!), an early stage fund founded by the large media / financial information company. In an article on Medium, they discussed the algorithm they used to predict who will start a venture backed company. It’s a fascinating read.
They have been using the algorithm for two years, with the following results:
Last year, the algorithm identified 350 potential future founders in the San Francisco area and New York
In their estimation, the group was more diverse than the stereotype of a typical VC-backed founder. In their words, “the data don’t understand discrimination”. That said, in this year’s class, only 1 in 5 predicted future founders are women
8 of these potential founders started companies and 3 are venture backed (including the popular tech curation site Product Hunt)
Half of these candidates have changed roles this year and 40% changed companies
Overall, they believe their method is 50x better than random at predicting who will start venture-backed companies.
Bloomberg isn’t the only one in the startup prediction game. The company Growth Science, claims to have an algorithm that does Business Model prediction, allowing “innovators simulate new businesses in highly predictive virtual worlds”, and effectively claiming to predict which startups or models will succeed or fail.
The Ironic Thing
One of Growth Science’s empirical findings is that corporate incubators are not that effective. On their blog, they claim that 70-90% of corporate incubator projects fail and when they do succeed, they are usually base hits rather than home runs. Perhaps that’s why Bloomberg Beta is very careful to distinguish what they do from Bloomberg Ventures, their sibling corporate incubator.
Our Take
Bloomberg doesn’t specify the prediction duration. So although only 3 out of the 350 predicted founders have actually founded a venture-backed startup a year later, it’s entirely likely that several more will surface in the coming months. Tapwage’s founders alas did not feature in their prediction list (we are devastated), but even if it is 50x more accurate than random, it’s still far from a proven predictive success.
Our anecdotal view is that people who do want to start companies, tend to gravitate towards the startup scene, especially given the continued proliferation of entrepreneurs and capital. So if being a founder is your dream, but you don’t feel ready for that type of commitment, maybe check out one of our many startup channels with curated jobs from leading startups globally. You may as well get started.