Decision Analysis and Portfolio Construction

In most asset classes, portfolio construction is an exercise in risk reduction. In early-stage venture, however, portfolio construction can be a powerful tool for increasing returns, not just decreasing risk

In most asset classes, portfolio construction is an exercise in risk reduction. In early-stage venture, however, portfolio construction can be a powerful tool for increasing returns, not just decreasing risk.

Portfolio construction should be designed in context. It should support a fund’s strategy and provide guidance for individual investment decisions. Ulu’s context is represented by the following decision hierarchy.

At the top of the hierarchy is Ulu’s Strategy which sets the constraints within which portfolio construction decisions should be made. If Ulu changes strategy, to include late stage investments for example, then Ulu’s portfolio construction will change as well.

In the middle of the hierarchy are three key Portfolio Construction decisions:
Size (i.e. the number of investments in the portfolio). Ulu’s answer is 70+, unusually large for a seed fund.
Reserves (the amount of capital held back for follow-on investments). Ulu’s answer is $0.50 for every dollar initially invested in a seed round. This is unusually small for a seed fund.

Mix (the mix of stage, sector, geography, risk and other key characteristics in the individual investments in a fund). We believe our strategy and large portfolio size naturally lead to sector diversification.
Allowing for a small amount of experimentation at the pre-seed stage, Ulu does not plan to diversify with respect to stage or geography. In other words, mix is not a meaningful constraint on our individual investment decisions.

Picking Winners is a Myth, The Power Law is Not

The annual Midas List celebrates the top 100 VCs each year. These are VCs who have “the magic touch,” reinforcing the widely held but erroneous belief that great VCs can pick winners. The Power Law demonstrates otherwise.

Research from Cambridge Associates shows that as an industry, VCs pick winners only 2.5 percent of the time. More than a decade of data reveals that out of more than 4,000 VC investment rounds annually, the top 100 deals generate between 70 and 100 percent of industry profits. Horsley Bridge data[2] indicates that, among the top performing VCs, 4.5 percent of invested capital generates 60 percent of their funds’ returns.

Thus, returns in early stage venture capital are distributed according to a Power Law with the lion’s share of returns earned from a small number of investments.

The blue dashed line above represents a Normal distribution with a bit of upside skew. It effectively characterizes returns of most asset classes including public equities, bonds, commodities, buyout, distressed debt, real estate, and even late stage venture capital[3]. For these assets, making the claim investors can pick winners has more credibility.

To outperform consistently, a manager only needs to pick assets in the top half of the distribution. Small portfolios highlight skill and a large portfolio creates regression to the mean, “watering down” the advantages of skill.

At the extreme, a portfolio of all assets in a normally-distributed asset class, eliminates all advantages of skill and generates industry average returns. Chance plays a different role in early-stage returns and must be factored into portfolio construction.

The solid green line above represents a “Power Law distribution.” It effectively characterizes returns in early stage venture.

The vast majority of investments return little to nothing and a small number of outliers drive the majority of the industry’s profit. Like other asset classes, a skilled VC is able to identify investments in the top half of the curve. Unlike other asset classes, however, most investments in the top half of a power law curve still generate uninteresting financial results.

Early-stage venture is an extreme outlier business. Even the best firms have many more misses than hits. At most, skill improves the odds of picking a winner from 2.5% to 4.5% with any one investment.

At Ulu, we don’t believe that VCs can reliably pick winners, but we do believe VCs can construct portfolios that have the potential to consistently generate great returns.

To illustrate the point, an index fund of early-stage venture investments over the last 40 years would have generated a mean return of 22% as compared to the median return of only 5.6%[4]. In a Power Law world, regression to the mean increases expected fund returns, it does not decrease them.

70+ Target Investments

Based on probabilistic reasoning, we believe a portfolio of at least 70 companies gives a 96% chance of investing in at least one outlier with corresponding superior risk adjusted returns and best balances diversification of early stage risk with practical considerations of deal flow and support.

The goal of such a large portfolio is to increase the chance of having at least one “outlier” in a portfolio, which typically means overall portfolio returns will be strong.

For purposes of the following analysis, an “outlier” is an investment that generates a 50x or better return. According to Correlation Ventures, who analyzed all available venture-backed company exits between 2004 and 2014, 2.5% of seed rounds and 0.8% of seed dollars invested generated 50x returns or better.[5] However, this represents the minimum outlier return. The high end can be multiples of 1,000x or greater.

The “Chance of an Outlier” chart below shows the chance a venture fund will have at least one outlier investment (y-axis) as a function of the number of investments in the portfolio (x-axis) and of the capability of the VCs (red and blue dashed lines).

To better understand the chart, let’s explore a few points.

The red dashed line represents the chance an average VC has of having an outlier in their portfolio. This is calculated using a 2% chance of an outlier with every investment, which comes from reducing the Cambridge Associates industry average of 2.5% a bit to remove top tier investors.

Looking at a few key points (the vertical gold dashed lines), an average VC with a concentrated portfolio of 20 investments has a 33 percent chance of investing in at least one outlier and an average VC with a large portfolio of 70 investments has a 76% chance of investing in at least one outlier.

Given Ulu’s experience over more than a decade of seed stage investing, we believe the blue dashed line, representing a top-tier investor’s odds, is the better model for Ulu.

The right portfolio size is still a judgment call based on risk tolerance, opportunity set, support model, and deal flow access. At Ulu, we have built a team and process we believe can source and support four to five high quality deals per quarter. Using a 4-year investment period, typical of a venture fund investing horizon, this translates to approximately 70 investments.

Some VCs contend this analysis doesn’t apply to them, pointing to their concentrated portfolio and great performance as proof. However, funds can make bad portfolio decisions and still have a good outcome thanks to chance.

The average VC with a concentrated portfolio of 20 investments only has a 33 percent chance of investing in an outlier. If there are 100 such funds, however, odds suggest 33 of them will have an outlier success! However, the odds also indicate that only 4% of such concentrated funds will invest in an outlier in three funds in a row.

Using the same logic, Ulu has an 88% chance of finding an outlier in three consecutive funds. While the Managers can make no guarantees[6], they believe smart portfolio construction can dramatically improve the odds of finding an outlier and delivering consistent top tier performance.

Small Reserves – Invest Aggressively Upfront

Ulu believes in a strategy we describe as “investing aggressively upfront” to maximize capital deployed in early stages when risk-adjusted returns are most attractive. As a result, Ulu typically reserves about $0.50 for every $1 in initial funding compared to most seed stage funds that reserve $1-$3 for every $1 in initial funding[7].

For more traditional seed and early stage funds, the typical reasons for large reserves are as follows: 1) large reserves are needed to maintain your ownership percentage in follow-on rounds. Funds that do not have the capital to follow-on will suffer meaningful dilution in their best companies; 2) early funds have an “information advantage” in follow-on rounds as they have been able to closely watch a company operate compared to any new investor. A large reserve fund is needed to capitalize on this “information advantage.”

These two arguments are sometimes summarized by saying a large reserve fund is valuable as it allows you to “double down on your winners.”

While the “double down” strategy sounds compelling initially, it comes at a cost.

Using Ulu’s historical follow-on statistics, a strategy of “investing aggressively upfront” substantially outperforms a “double down” strategy.

Here’s why.

Conceptually, “doubling down on winners” means investing in progressively later stages with the related later stage risk/return patterns. An early stage fund with a large reserve pool effectively becomes a late stage fund on a dollar-weighted basis. In addition, valuations associated with follow-on “winner” rounds have the generally unwanted effect of dampening the overall return of the total investment.

Ulu’s strategy is conceptually orthogonal as we seek to invest in companies before they become winners when the risk-adjusted returns are the most attractive.

To help illustrate the difference in approach, Ulu built a model to compare these two strategies on the dimensions of portfolio returns and ownership in “winners.” On ownership in “winners” we find the following:

  • The green line in the chart above illustrates a model portfolio of a well-performing seed fund following the “invest aggressively upfront” strategy. It shows 100% of capital invested in the seed round for an ownership percentage of 12.5%. Nothing is held in reserves for follow-on rounds, and for those that survive to raise a Series A, the ownership position is diluted down to 8.9%. Likewise, for those that survive to the Series B and C rounds, the “invest aggressively upfront” strategy is diluted down to 6.1%.
  • The red line in the chart above illustrates a model portfolio of a well-performing seed fund following a “double down on winners” through Series C strategy. In this scenario, a seed fund reserves $3 for every $1 invested. This is enough capital to meet all pro rata funding requirements through Series C for those companies that survive. In the seed round, this strategy would lead to purchasing 3.3% of the startup’s equity. And at each of the Series A, B, and C the fund maintains their 3.3% equity, suffering no dilution.

At the end of the day, the “invest aggressively upfront” strategy will have roughly doubled the ownership in the winners, the companies who survive to Series C and beyond. This leads to roughly a doubling of fund performance as well. If the Red Line, the “double down on winners” fund, generates a 3.4x net multiple then the green line, investing in exactly the same companies and under exactly the same terms, generates a 6.4x multiple.

We are not arguing that a strategy of doubling down is always wrong. There are a few key reasons where it may make sense for an early-stage VC to double down.

  • When the fund size is too large to be fully invested in the early-stages. Large funds must invest in later rounds to put all their cash to work.
  • When entrepreneurs have not yet garnered enough traction to convince follow-on investors to invest. These are not “winners” so are still reasonably priced. It’s worth stressing that most potential winners — such as Ulu investments SoFi or Palantir — tend to have more investors willing to invest capital than needed once they’ve demonstrated traction. In the case of potential winners, early investors’ continued financial support is rarely needed as a signal of value.
  • When follow-on capital is scarce and good startups would not survive unless the early-stage investors continued to support them financially.
  • When a fund’s initial investments perform poorly. If only a small percentage of a fund’s investments receive follow-on financing, follow-on checks may have a better risk/return profile than initial checks.

On balance, for most early-stage venture capital firms working with a limited supply of cash, investing more in the Seed and Series A rounds is the smartest way to boost returns. Early-stage funds with large reserves effectively become late stage funds on a dollar-weighted basis and will tend to deliver lower returns than funds whose dollars are concentrated earlier in the life cycle.