Back to Blog

Startup Investor Guide: Diversification

By Alex Mittal and Christopher Steiner  •  May 19, 2016

Most investors opting into startup investing should limit startups to less than 15% of their total portfolio, probably much less, and should make at least 10 investments, probably much more. Diversification can reduce the impact of negative events within your startup portfolio, but excessive diversification can also reduce the likelihood of your portfolio outperforming the asset class. Finally, since returns are driven by top-performing outliers, diversifying across randomly-selected startups will not save your portfolio; you must aim to diversify across top startups, which is hard work.

Diversification is one of the simplest and most important tenets of prudent investing. Whether it's a 401k, an IRA or a large family fund, smart investors seek this trait within their portfolios. The idea is to avoid having a single investment—or a group of correlated investments—that inflict excessive damage to one's overall net worth. By spreading investment exposure around different classes of investments, from commodities to real estate to equities and more, investors can maximize long-term gain while also limiting risk.

The best way to invest in startups requires embracing similar ethos when it comes to diversification. But diversifying startup investments isn't the same exercise as diversifying the rest of your portfolio. The biggest difference is that startup investments are a class of asset all by themselves. And, as a class, they're uniquely volatile. Compared with stocks or bonds, a single startup investment stands a far higher chance of being marked to zero, of returning nothing, ever. In fact, according to data gathered by the Kauffman Foundation’s Angel Investor Performance Project, 52% of early startup investments will fail to return 1x. Horsley Bridge, a fund of funds that has made significant investments in venture capital for more than 30 years, reports similar numbers, saying that 62% of its funds’ early venture investments return less than 1x.

The magnitude of risk here runs more than a few standard deviations above that of blue-chip stocks. With greater risk should come greater potential rewards, of course, and that is indeed the case with startup investing. According to the Kauffman data, 7% of exits brought returns of 10x or more, which supplied 75% of the total volume of returns overall. Buttressing that, Horsley Bridge’s data show that early-stage VC investments are 60% more likely to return 10x or more compared with mid-stage or late-stage investments.

The Kaufmann dataset aggregates the investing results of 539 angels with investments spanning two decades and a total number of 1,137 exits, both positive and negative. A closer examination of the data by professors Robert Wiltbank and Warren Boeker of Willamette University and the University of Washington yielded a study that supplies some of the date here. Horsley Bridge published its own data covering three decades and, through its fund partners, more than 5,500 investments.. Neither data set perfectly reflects the sector. The Kaufmann data is self-reported, although Kaufmann believes it is free of significant self-selection bias based on their specific analysis of that issue. The Horsley Bridge data draws from sophisticated VC fund portfolios, which may not reflect one’s own ability to invest in the startup asset class. Regardless, these datasets provide reasonable approximations of reality that lend themselves well to drawing actionable conclusions.

Startup investing, as a class, needs to be diversified in two distinct ways:

No. 1: An investor needs to decide how much of her portfolio should be allocated to venture capital. Just as real estate or commodities are a class of investment, so, too, are startups, and an investor's overall exposure to them should be limited. Most investors with VC experience agree that the total shouldn’t comprise more than 5% to 15%, depending on an investor’s goals, the size of the portfolio, and what an investor’s liquidity requirements are year-to-year. Investors placing money with longer horizons in mind (~20+ years) may want to consider allocating toward the top end of that range. Investors who may require much of their capital back more quickly (within ~10-15 years) should allocate toward the lesser end.

Yale University’s endowment, for one, has a 2016 goal of placing 14% of its capital in venture capital vehicles. This is an increase compared with earlier years as Yale’s VC investments have performed very well over the past 30 years.  During the last 10 years, Yale’s VC portfolio has delivered an 18% annual return, the best of any asset class in its wide-reaching portfolio. During the equivalent time period, the S&P 500 delivered a 10.1% annual return.

That brings us to No. 2: Investments in startups, because of the wide distribution of their returns, need to be spread out across a sufficient number of companies. One investment isn't enough. The trick to investing in startups is to realize that the bulk of returns come from a small slice of the total number of companies. Chris Dixon, a general partner at Andreessen Horowitz, describes the Babe Ruth Effect, where swinging harder produces more strikeouts, but also more home runs.

So the top 10% of companies hold a disproportionate share of the total returns for the portfolio. But it's also true, and this is the key, that the top 10% of the top 10% contributes a disproportionate share of those large returns, and the top 10% of that slice contributes a disproportionate share of those returns and so on. Kevin Dick at Right Side Capital Management did a deep statistical dive on this and describes this conundrum well. Paul Graham describes it even better in his essay, Black Swan Farming, which lays out the dilemma: the biggest hits, like YC’s own Airbnb, don’t at all look like big hits early on.

Kevin Dick's statistical assessment concludes that, as the number of startup investments increase, the average return coalesces around 4.05x. Assuming a overall portfolio hold time of six years (perhaps not a valid assumption in today’s environment, where the path to liquidity could take longer), that yields an internal rate of return of 26%.

With all that said, investors who possess deep domain knowledge or who otherwise believe they have an unfair advantage in predicting the future may feel comfortable placing fewer bets, as they may be able to suss out the truly revolutionary companies. Having more hits with fewer investments, of course, greatly increases returns. Those investors who are newest to the space and who don’t feel they yet have a knack for venture investing may be comfortable placing more bets, which limits the downside.

The rather simple bottom line for startup investors: Limit your portfolio’s exposure to venture capital to some number less than 15%, depending on liquidity needs and time horizons. And place multiple bets within the startup space. If you’re surer of your success for whatever reason, you may be comfortable placing fewer bets, e.g. ~10-15 (But good luck! Venture math is not on your side). Otherwise, it’s better to err toward 20+, if not 50+ investments, and to also diversify across time. Most top VC firms will place 20-75 companies within each fund, which typically deploy capital across three-year periods. This means that over a decade, most VCs are investing across dozens, and in many cases, hundreds, of startups.


This all assumes that investors will be performing an equivalent amount of due diligence on each investment. The Kaufman study showed a close correlation between the time an investor spent on diligence work for a particular company and the return that company eventually generated. The overall return for investments made with a high amount of diligence—defined as more than the median of 20 hours, was 5.9x. Compare that with returns of 1.1x for investments where less than the median amount of diligence was performed. Even here, the top 25% shines even more—as investments that carried with them diligence work of greater than 40 hours returned an average of 7.1x. But it’s important to note that diligence is just one factor that can greatly affect returns. Among the others: geography, sector and stage/age of company.

Clearly, diligence is good and more diligence is better. But taking on that kind of workload can get onerous when an angel might be aiming for 20 or more startup investments. This is why venture capital LPs like leaving this work to VCs: they're counting on the VCs expert diligence and the diversification effect of being in a larger fund to keep their overall returns elevated.

While investors should always conduct their own diligence, individual angels and startup investors can benefit by investing through platforms that perform a large amount of diligence on the startups that eventually appear as investment opportunities. At FundersClub, for instance, fewer than 2% of companies who are reviewed eventually clear the vetting and diligence process and are made available for investment.