Harry Stebbings did an awesome 20VC podcast recently where he interviewed Semil Shah, GP at Haystack and venture partner at Lightspeed, on portfolio construction, which is something I’ve really been geeking out on recently.
It was great to hear other early-stage VCs talking about how they approach this rather esoteric subject, so it inspired me to write this post. There is some fantastic material out there for those who are interested in learning more, which I reference toward the end of this article, but suffice to say that Fred Wilson, Benedict Evans and Jerry Neumann have all published phenomenal articles delving into these topics in varying degrees of technical detail.
Instead of covering the points they focus on (mainly what an “optimal” venture portfolio looks like), I thought it might be useful to illustrate what type of returns would be needed to allow a fictional European seed stage fund with a fairly typical portfolio construction to achieve a successful outcome.
Let’s assume that our little European seed stage fund has been reasonably successful for a first-time manager and has managed to raise $75M.
Most VCs target a 3x net return to LPs (their investors), i.e. you give me $1 and I will give you back $3. Management fees and recycling, which I don’t propose to go into in this post, complicate things a little bit, but broadly speaking, in order to deliver a 3x net return to LPs our fund would roughly need to achieve a 4x gross return — so in this case, $300M.
Once the fund has been raised, the GP (i.e the VC) needs to decide how to split up the $75M it has raised among its various potential portfolio company investments to maximise its chances of delivering a 3x net return to its LPs. As mentioned above, I’m not going to go into detail on some of the theory underpinning how a VC makes this decision, which rapidly gets very complicated. I’m also going to simplify things a bit and will assume that the fund can invest the full $75M — in other words, I’m going to ignore management fee drag and recycling.
For ease of illustration, we’ll just assume that the fund’s strategy is to invest in at least 25 companies at seed stage and to reserve enough capital to “double down” on the winners, i.e. the fund is targeting healthy diversification, without being spread so thinly that it prevents the VC from adding any value. We’ll also assume that average amounts raised by European startups are as follows (valuations assume a 20% dilution in each round):
Note: The data here is largely taken from Dealroom.co analysis, including the recent “Series A Landscape in Europe” report published in partnership with Local Globe (for a great summary, see Mish Mashkautsan’s blog).
When funds make new investments, they will usually hold back capital so they can invest in future follow-on rounds. The amount of capital they hold back is referred to as their “reserve ratio”. We’ll assume that our fund has a 1:1 reserve ratio, although many seed funds have reserve ratios of between 1:2 and 1:3 (or even up to 1:4). In other words, for every $1 our fund invests in new portfolio companies at seed stage, it will hold back $1 for follow-on investments.
That means our fund will have $37.5M ($75M / 2), which it will invest in 25 new seed-stage companies, i.e our fund will be writing cheques of c.$1.5M in seed rounds raising $2.5M at $10M pre-money (given assumptions mentioned above) and will be targeting a holding of c.12% on a post-money basis in the seed-stage companies in which it invests.
How will our fund decide to allocate its reserves? Well, if it’s a top quartile European fund, it will follow at Series A in 40% of its seed investments (vs. industry avg. of 19%) according to analysis by Atomico, Local Globe and Dealroom.co. We’ll extend that and assume that our fund also follows at Series B in 40% of its Series A investments. Given how much larger and more expensive Series B rounds are, we’ll also assume that the 50% held back for follow-ons is slightly weighted in favour of Series A rounds, so that 30% of the fund is deployed in follow-ons at Series A and 20% is deployed in follow-ons at Series B.
Now that we have sketched out our fictional fund’s strategy, we can show what our portfolio construction looks like. Essentially, in addition to its 25 seed-stage investments, our fund will follow-on:
at Series A in 10 of its portfolio companies, writing $2.25M tickets in $7.7M rounds at $30.8M pre-money, growing its stake from 12% to 15%; and
at Series B in 4 of its portfolio companies, writing $3.75M tickets in $18.0M rounds at $72.0M pre-money, growing its stake from 15% to 17%.
So by the time it is fully deployed, our fictional fund will have invested a total of $7.5M in each “winner” and will hold a 17% stake post-Series B. What does a “home run” now look like?
We’ll assume that the most successful portfolio companies in our fictional fund go on to raise further capital, and that since our fund is not able to participate beyond Series B, it will suffer ~50% dilution between Series C and exit. In other words, our fund will end up holding an 8.5% stake in its most successful portfolio companies just prior to exit.
To “return the fund”, one of these companies would need to exit for $887M. To deliver a 26.7x multiple of invested capital (per Benedict Evans classic post, “In Praise of Failure”), the portfolio company would need to exit for $2.4Bn. In order to “4x the fund” (i.e. return four times the total size of the fund), the portfolio company would need to exit for $3.5Bn.
I’m not saying in this post that this illustrative portfolio construction is the “optimal” portfolio for a seed-stage venture capital fund. Instead, this post is intended to illustrate what sort of returns a seed-stage VC with a fairly typical portfolio construction will be expecting every single one of its seed-stage investments to be capable of delivering, given than most will actually go on to return 1x or less.
And if you’re interested in more of my musings, you can follow me on Twitter.