A Return to Venture Skewness (& Serial Persistence)

Earlier this week I posted some data showing that the median return of top quartile venture funds was well ahead of the median return of second-quartile funds. At the same time, the gap between second-, third-, and fourth-quartile funds was nowhere near as large as that between first-quartile and second-quartile funds. While I didn’t come right out and say it, I implied that this was an interesting and unusual characteristic of the venture market.

Well, venture guy Bill Burnham wrote in to make a cogent and articulate case for why there was less to the quartile performance data than meets the eye. It was, he argued, an artifact of venture industry structure, nothing specifically to do with, say, super-investors at the best funds.

What is it about the venture industry’s structure that made Bill say this? It is that downside is limited — you can’t (easily) lose more money than you put in — while upside is unlimited — now and then you’ll have a Google on a relatively small investment. The upshot of this market structure, Bill rightly argued, is a one-tailed performance distribution with many firms delivering mediocre returns, and a few firms delivering outsized numbers. That, of course, must lead to a big gap between first– and second-quartile funds, but for no other reason than mere statistics of a one-tailed distribution.

I agree, which brings us to Bill’s second point. He says that the more interesting question is whether quartile barriers are “rigid” in venture markets. In other words, just because you are first quartile this year, does that mean you will remain so next year (or the year after)? Rigidity or non-rigidity has enormous implications, as the mutual fund industry has well demonstrated, where top performers this period are highly likely to underperform next period.

Is the same thing true in venture markets? No, and this has been demonstrated in at least one public study, as well as in my own and other non-public analyses with which I’m familar. The studies have all shown approximately the same thing: that the best predictor of first-quartile venture fund performance next period is first-quartile performance last period. (Similarly, and perhaps more entertainingly, the best predictor of bottom quartile performance next period is bottom-quartile numbers last period.)

So, if serial persistence exists, at least in first– and fourth-quartile venture funds, what about those poor bastards in the messy middle? Are second– and third-quartile funds trapped there forever, as if in some sort of financial caste system?

Good question, but one without a clear answer. There is some class mobility in venture performance, but it’s mostly a random walk between the middle quartiles. Then again, a few well-known funds have climbed from middling to top-tier performance; but having done a Benchmark, say, and ascended to the first quartile on the back of an eBay or equivalent, does not necessarily mean that said fund will stay first-quartile. The data shows it’s more complicated than that.

So where do we end up? There is a caste system in venture fund performance, with a marked gap between first– and all other quartile funds, and the size of that gap can be explained as a statistical artifact of the venture marketplace.

Arguably more importantly, however, at least from a limited partner investor’s point of view, there is serial persistence in venture fund performance. So Yale/Harvard/UC/etc. are not being irrational by trying desperately to get into those top performers (where the same behavior would be irrational in investing in top mutual funds).

The deepest question, however, of whether funds can break out of the mid-market malaise and ascend permanently to KP-esque status remains open. Yes, it happens, but it doesn’t happen often, and there is no recipe for doing it.


  1. Fantastic! Great post Paul.
    That persistance thing is definetely dynamite.