We F****d the Power Law
For the last month I’ve been debating with one of my friends about the benefit/drawback of a concentrated portfolio approach. I can tell him that mathematically spray and pray funds underperform concentrated portfolios, with multiple case studies, but I couldn’t mathematically explain why the law of large numbers wouldn’t apply. In other words, more startup investments should hypothetically get you closer to obeying a theoretical probability distribution, a theoretical power law. Basically, you’d be guaranteed to be more likely to find 4 or 5 unicorns, and if you maintained ownership at an even level across the portfolio you’d be more likely to generate top quartile returns. Even more confusing because this strategy certainly used to work ten or fifteen years ago looking at some of the older funds that have lost credibility now.
After reflecting on my math, logic, and venture understanding I’ve come up with two explanations, one of which I’m more inclined to believe than the other, which is that we f*****d up the power law by rapidly overexpanding the VC ecosystem.
For posterity’s sake though, I’m still going to explain my other line of reasoning, which is that the power law isn’t real at all, and why I don’t think it’s (strictly) true. Without further ado:
Is the Power Law a Lie?
This is the easy explanation for why so many larger venture funds underperform the market, why accelerators and incubators are notoriously hard to build, and why spray and pray doesn’t work. The power law must be wrong or the distribution is more linear than we predicted. If the distribution is funkier and VCs manage to find one extreme of the distribution by investing in highly risky companies, or just some VCs never find that end of the distribution, then obviously you’d approach whatever that distribution looks like. Maybe the power law only applies in certain circumstances. Maybe VC as an asset class is just one giant web of lies.
Two issues though, first, I couldn’t come up with another distribution that would create a pattern that looks anything like what we really have in venture backed companies. Sure, you can fatten the middle of a distribution. You can make it look weird at the tail and the head. You can do all sorts of things, but it’s hard to replicate what VC looks like with anything but a power law.
And the more important the issue though, spray and pray does work. Sometimes. There are a lot of large, less concentrated, fund strategies that generate excellent returns. A16, Soma, Lightspeed, and many others. Which means there’s only one other explanation…
We F***** the Power Law
The only other mathematical explanation I could contrive (and I’m sure I’ve missed many, but feel free to flame me in the comments for that) to generate the returns that we see is simply that we seriously screwed up the power law. In other words, we made the probability of finding unicorns so rare relative to the size of the ecosystem that even a portfolio of 500+ startups will not represent a sample size large enough to approach even average returns unless those startups are picked intentionally and exceptionally.
How did this happen? Really simple actually. Venture capital as an asset class grew too quickly with too little attention to who we were letting write checks. Lots of dumb money disproportionately led to investments in lots of dumb startups (and many good startups that were just not timed correctly for the market).
Because of this, instead of expanding the long tail, we only grew the tall head. Which means that spray and pray won’t work because the theoretical probability of randomly encountering a unicorn is far lower.
So why can some spray and pray funds bypass this? Simple actually — be better. Or moreso, be as good as investors used to be. With a diligence process you can easily bias out the tall head of really bad investments. Unfortunately, most VC investors are not known for their great ability to do simple diligence. More on that later. Large funds, people with great processes and automation, can conduct real diligence processes and then index on the remainder to generate top tier returns (which let’s be real, still aren’t that great). You can similarly index off “pre-vetted” ecosystems like the YC ecosystem and within that diligence a little, you can end up on the right side of the new power law curve.
This also means though, that as a small it is impossible to properly execute a large portfolio strategy. If you need to write 50 checks out of a $10MM fund it becomes impossible to perform the level of diligence that you need to in order to properly underwrite most rounds. You’re far better off writing 15 or 20 and having a much more thorough DD process. Meet founders in person instead of watching them play league of legends over Zoom, it’s not that hard.
Similarly with a large portfolio strategy you will likely struggle inherently to get access to deals that are pre-vetted or have characteristics that bias them to larger outcomes. This is why smaller accelerators, small spray and pray funds, and small incubators often struggle significantly. There’s no way to beat the new power law if you don’t have the ability to do the work or have an advantage.
Conclusion
There’s one more grim detail I’ll leave you with, which is there’s realistically no way to change this in the near term. Checks have been written. In some cases when institutional LPs have committed to a fund I they’re almost guaranteed to commit too a fund II. The checks have been written, and there’s only one way out: time.
The happy ending? Time does tend to course correct the venture industry… We’ll see.
