Thiel vs. AngelList. Who is Right?
U.S. venture is a hits driven business, and we’ve previously published data quantifying how right-skewed the distribution of outcomes is. What are the implications for the optimum portfolio size for a U.S. venture fund? Peter Thiel of Founders Fund and Abe Othman of AngelList arrive at very different conclusions.
Highly successful investor Peter Thiel advised VCs to only invest in “seven or eight promising companies for which you think you can get a 10X” return. Abe Othman, Head of Data Science at AngelList, countered in his excellent blog post here using AngelList data that the math behind the “power law” instead suggested VCs should have much larger portfolios to make sure they have a sampling of the winners.
Who is right?
We thought you might find interesting analyses we’ve conducted over the years to help answer this question. We’ve built the most complete database of U.S. venture financings and outcomes in the industry. To create the graph below, we created tens of thousands of portfolios by randomly selecting investments for each vintage since 2000 with portfolio sizes ranging from 1 to 100 companies (a “Monte Carlo” analysis). For the portfolios at each portfolio size, we calculated both the average and median (50th percentile) return.
A larger portfolio alone does not impact average, or expected, returns. However, the median (50th percentile) returns for a fund, which a fund is more likely to achieve in practice, does increase with portfolio size. This is a mathematical result.
Median returns nearly converge with mean returns by the time a venture portfolio reaches one hundred companies. Median returns increase as a function of portfolio size because the distribution of venture returns is right-skewed. For those interested in learning more about the math behind this convergence, it results from the Central Limit Theorem.
The above graph supports Abe’s argument that the power law leads mathematically to the benefits of larger portfolios, all else equal. However, there’s more to the distribution of venture returns than just mean and median returns. The entire band of potential outcomes shrinks as portfolio size increases. To illustrate, the graph below plots the probabilities of portfolios at each portfolio size having a gross return of 5X or greater (the green line) and less than 1X (the red line).
For example, at Thiel’s recommended portfolio size of eight, the probability of the overall fund generating a loss is 28% versus only 6% for a portfolio of 100. However, the probability of Thiel’s smaller portfolio generating an overall return of 5X+ is 10% versus just 2% for the larger portfolio.
In other words, for a venture portfolio of eight companies, the probability is five times greater both that it will return less than 1X, and that it will return greater than 5X, versus a portfolio with one hundred companies.
While smaller venture portfolios have a higher probability of an overall loss, they also have a higher probability of hitting the ball out of the park.
The above conclusions assume that the venture portfolios were selected and managed by an “average” VC, and there is no difference in the average quality of the investments as a function of portfolio size. This later assumption is almost certainly not true for most venture funds. Instead, there will be an inverse correlation between the quality of each investment and portfolio size. There are only so many high quality opportunities most VCs can find and manage.
Implicit in Peter Thiel’s recommendation is that a small, concentrated portfolio helps a VC to increase the odds that each of their few investments is one of the winners. The greater the “edge” the VC is assumed to have over the industry average, the more the above curves will shift.
So what would these statistics look like, for example, if a VC believes that each investment they make has twice the odds of generating a 10X+ return when compared with investments made by an “average” VC? The probability of this “twice-as-good” VC with a portfolio of eight companies having a less than 1X fund declines from 28% to 21%, while the probability of them having a 5X-or-greater fund increases from 10% to 18%.
So who is right: Thiel or AngelList? It depends.
The expected returns are the same for VC portfolios of different sizes. The risk/return profiles are simply different. One is not necessarily better than the other. It depends on the goals and risk tolerance of the investor, and of course how confident the VC is that they can beat the odds and select and/or create winners above the industry average across whatever portfolio size they choose.
At Correlation Ventures, we’ve designed our funds with the above statistics in mind. We believe that our proprietary selection model enables us to increase the odds that each investment will be one of the winners, consistent with Thiel’s advice. The scalability of the selection model, however, also enables us to take advantage of the benefits of larger venture portfolios, as recommended by AngelList. We create large diversified portfolios of selected co-investments in order to mathematically reduce downside risk and maximize the likelihood that our overall portfolio realizes its attractive expected return.
If you are entrepreneur raising or working with a team that is closing a round, we would value any introductions where we might be helpful as a co-investor.
Correlation Ventures is the predictive analytics pioneer in the venture capital industry and the industry’s leading co-investor. With more than $475 million under management, we’re one of the most active U.S. venture investors. Since inception, we’ve invested in nearly 400 companies. Sample portfolio companies include: AlienVault (ACQ:AT&T), Astra (ASTR), BlueVine Capital, Gabi (ACQ: Experian), IonQ (IONQ), Imperfect Foods, Janux (JANX), Lemonaid (ACQ: 23andMe), Lever, Manticore Games, Personal Capital (ACQ: Empower), PowerVision (ACQ: Alcon), Synthorx (ACQ: Sanofi), and Upstart (UPST). Correlation offers a dramatically better option for lead investors, syndicates, and companies seeking additional venture capital to fill out a round. We offer the most rapid, low hassle, and helpful source of co‐investment capital in the industry; for example, typically making investment decisions within days. Correlation is backed by leading institutional investors.
More Stories by Correlation Ventures: