Venture capitalists usually try to avoid statistics, arguing that picking startups is more art than science. When asked why they don’t use statistical decision tools, most VCs say there either isn’t enough data to build reliable models, or picking startups is simply too intangible and complex to be reduced to math. There’s some truth to both arguments… but they’re getting weaker every day as machine learning, AI and data science continue to blossom. Thanks to data science, math is elbowing its way to the venture capital table.
We’re still in the early days, but that’s what’s exciting. According to the National Venture Capital Association, there were 1,562 venture capital funds in the US in 2016. The real number is bigger since some funds aren’t reported by the NVCA, but whatever the total, fewer than 1% of venture capital funds make any serious use of statistical tools to guide their decisions.
This makes the venture capital industry an open target for anyone with the actuarial discipline and technological skill to gain an edge. As a Partner at WR Hambrecht Ventures, one of the first quantitative VC funds, we’ve found mind-bending advantages and information asymmetries by using data science to look at deals. Other quant-oriented funds such as CapitalG (Google), Correlation Ventures and Link Ventures are enjoying similar perks.
It’s perhaps even more interesting to note these funds (CapitalG, Correlation, Hambrecht, Link) use wildly different techniques. As it turns out, there’s more than one way to predict the future. Rather than seeing data science in venture capital as some monolithic, one-way-to-do-it path, in reality the field is evolving with delightful variety and creativity. Each model emphasizes different aspects of a startup, draws from different data and combines it in different ways. Science, it seems, can be a bit of an art form too.
Multinationals other than Google are also using statistical tools to pick which startups, acquisitions and new product launches to pursue. For example, Fortune reported companies including Intel, 3M and Cray using specialized prediction engines to figure out the statistical odds of success for products before deciding whether or not to launch them into the market.
This is exciting for more reasons than just making money. Yes, it’s great to be part of a top performing VC fund, but the real story is bigger. Venture capital puts a lot of money into startups ($58.8 billion invested in the US in 2015)[i], but it’s a tiny fraction of the approximately $2.5 trillion in global private equity assets under management.[ii] A big reason venture capitalists don’t have more money is the industry’s high failure rate. One study found 75% of venture-backed firms in the US didn’t even return their investors’ capital.[iii] While this study has been debated, venture capital is notorious for at least failing more often than it succeeds. As a result, money managers relegate venture capital to a small “high risk” portion of their portfolios rather than investing more. This puts less money in venture capital funds, which in turn leaves less money for startups that might otherwise solve more problems, hire more people, and make bigger contributions to the real economy.
Actuarial science and statistics have been around since at least the 1600s. Today they shepherd more than an estimated $4 trillion in insurance capital assets, and trillions more in private capital assets held by banks. It’s therefore sharp irony that, despite its status as the bedrock of risk management, actuarial science been largely ignored by venture capital – one of the world’s highest risk asset classes.
Venture capitalists, startups and society as a whole stand to benefit if VCs get better at picking winners. The industry has traditionally been based on intuition, which has failed more often than it’s succeeded. Now, with big data and big analytics, the industry has a historic opportunity to improve. It’s also an opportunity for a changing of the guards, as a new breed of investors challenge the status quo and continue to demonstrate how science can be combined with art to paint a better picture.
[i] National Venture Capital Association, $58 Billion in Venture Capital Invested Across U.S. in 2015, According to the MoneyTree Report, January 15, 2016. http://nvca.org/pressreleases/58-8-billion-in-venture-capital-invested-across-u-s-in-2015-according-to-the-moneytree-report-2/
[ii] ValueWalk, Private Equite Assets Under Management Approach $2.5 Trillion, January 31, 2017 http://www.valuewalk.com/2017/01/private-equity-assets-management-approach-2-5-trillion/
[iii] Deborah Gage, The Wall Street Journal, The Venture Capital Secret: 3 Out of 4 Start-Ups Fail, September 20, 2012 https://www.wsj.com/articles/SB10000872396390443720204578004980476429190#articleTabs_comments%3D%26articleTabs%3Darticle