Algorithm-Driven Venture Capital

VC is dominated by probabilities. In the most famous instance, venture returns aren’t evenly distributed but instead fall victim to the Power Law. We make investing decisions accordingly. We expect that most of our investments will won’t generate a meaningful return, but some will generated an outsized return.

In the face of that certainty, most VCs make investment decisions subjectively. Even if you conduct interviews and thoroughly understand the market, there’s an opportunity to be far more rigorous.

While not proven to be successful, relying on algorithms to make investment decisions can’t hurt your chances of success in a world where probabilities are king, especially when it comes to later-stage startups.

A Startup’s Probability of Success

Certain characteristics of a startup point to increased probability of success. Correlation Ventures, a VC firm in California, ascribes to this notion — they choose firms that have already caught the eye of larger firms, like Sequoia Capital, Accel Partners, Battery Ventures, InterWest, and Canaan Partners. Correlation is an algorithm-driven VC fund:

“Correlation’s secret algorithms weigh heavily on the track records of the entrepreneurs, investors and other advisors. Correlation believes that reputations aren’t random. Also highly prized is a company’s ability to spend money efficiently. Correlation has a good idea how much it costs a startup to pass through its stages of development. A prospect rolling ahead on a tight budget looks appealing. A spendthrift rival, not so much.“

Correlation’s goal is to analyze data based on proprietary algorithms, and then to “pick investments via pattern-matching software.” They have made about 85 algorithm-based investments in startups since they opened in 2011.

While I do think that algorithm-based investment make sense when investing in more mature startups, the jury is still out when it comes to the seed stage. Early on, I think that it’s important to understand and known the management team. Algorithms can’t do that (yet).

When it comes to series B deals and later, algorithms can and should be used. According to Fred Wilson in an interview with Dan Primack: ‘after Series B, it’s just a check.’ In other words, investors are not going to move the needle with advice, network, and connections. You need to rely on past data to predict the future.

How Algo VC Investing Works

Thomas Thurston explains that his “think tank,” Growth Science, is “trying to do the science, build the tools, and do the research all around this one question: How can we better predict when innovations will survive or fail, both for startups and when corporations launch new products or do acquisitions?” They use algorithms to predict the success of business models, and they avoid all human subjectivity, basing their decisions on “a discrete yes or no.”

Google Ventures also uses algorithms to make decisions on investments: “The firm feeds its algorithms data gleaned from academic literature, past experience and due diligence about start-ups and their founders.” This seems similar to the methods used by Correlations to predict whether a startup will be successful.

Use Algorithms for Diversification

Since Ben Graham’s Intelligent Investor, diversification has become an integral part to your stock portfolio, but it’s difficult to achieve in VC. In the public markets, you can buy an S&P 500 index fund or a diversified basket of ETFs.

Obviously, this is not the case in venture. There’s no ETF for a 2011 Vintage early-stage startups. However, picking your investments based on numbers rather than subjective interests can help you to diversify, which will only improve your chances of generating a successful return for your investors. If you use algorithms to guide decision making — ranking several criteria offers a less subjective selection process — you’ll hopefully achieve diversification and top quartile results.