My takeaways from the article: (1) VC is a casino for the rich and just like gamblers in casinos they have no idea what they are doing; (2) VCs have discovered that standard deviation shrinks like the square root of the sample size, however their understanding of stats seems to be just enough to run a monte-carlo simulation.
Edit: the 3rd takeaway is advice to those who're considering to join a "startup" - if VC needs 500 investments to make a 15% return on average, you need 5000 years to get the sames returns as a line worker.
> just like gamblers in casinos they have no idea what they are doing
I follow all of the VCs on Twitter and it's very clear from their many, many tweets that they have a far superior intellect than the rest of us which allows them to divinely predict the future.
And between them and their diverse network of other 40 year old, white males they have a rich, deep understand of the customer's wants and needs. And from that they 'select' the startups that best aligns with this understanding.
I even believe that one day VCs will realise that they don't need founders and can just invest in each other. Keeping the prosperity moving.
Not sure if this is sarcasm or not. I'm leaning towards sarcasm, but if there ever was a place where someone would say this with a straight face it would be HN.
It's a similar phenomenon to media celebrities who clearly have access to special information that makes them a good choice for Tv interviews on just about anything.
This is increasingly true in the Private Equity (PE) world...
A lot of recent exits have been sales to other PE funds rather than to the IPO market.
PE / VC have become an Assets under Management (AUM) game where the GPs only care about the fixed fee they get on the cash invested. With funds locked in for 10 years or more, that's a massive annuity - who needs outperformance?
And if you do need to show liquidity, let's exchange a few assets between the usual suspects...
You seem to be insinuating that a PE exit via a sale to another fund is somehow inherently bad. Many PE investors would argue the opposite - IPOs have lockups and price volatility that increase both certainty of exit and time to exit. A sale to another PE firm or corporate entity generally deliver a large onetime cash payment.
A dual track (where you run an auction to both private and public buyers and simultaneously file an S-1 to IPO) maximizes the value to the seller. This sort of disproves the premise.
> if VC needs 500 investments to make a 15% return on average, you need 5000 years to get the sames returns as a line worker
This doesn't necessarily follow. A "line worker"'s downside risk is the opportunity cost they pay for working at a startup, which has a different distribution than the investor's downside risk (the whole investment). At the extreme end I'd argue that e.g. WeWork's investors came out of it worse than the employees.
The downside distribution is more punishing on the low end. For people who need to pay rent, buy food, and pay medical bills, a 90% probability of losing half your wages is generally not worth a 10% probability of making 7x your wages.
That's because the less money you have, the more valuable a dollar is.
Wages are recurring income. So when wages fall by 50% you need to value an annuity with half the cashflows as before. The PV of your wage income stream over some time horizon can then be compared with the one time gain.
Otherwise you are comparing a stock concept (wealth) with a flow concept (income)
Correct. Another vehicle to display your point may be to consider the serial startup executive, who leapfrogs from growth stage startup to startup every few years, collecting 0.5%-2% of each in the process. For people who want to make returns from their stakes in startups, that seems like a better idea than going the VC route because it's easier to get better information and better deals as an insider.
You can only make so much money from startups by gatekeeping how other people's money is invested. That skill does not an operator make. VCs with operating experience I think operate with a different toolset and underwriting criterion than lifers.
Investors lose only a small portion of their assets and most importantly they don't have to lose time. Whether 2% of my liquid assets is more valuable than 4 years of someone's life is debatable, but I can certainly say the loss of those 2% won't make any impact on my life, while that worker won't ever recover his 4 years.
> VC is a casino for the rich and just like gamblers in casinos they have no idea what they are doing
Except that the VC expected return is positive. They just need a sufficient number and diversity of bets to lower their variance to a small value relative to their expected return, which most of them don’t do.
Correct, there isn't a distinction between positive expected value financial games, versus negative expected value financial games except cultural tolerances.
There is also the exclusively state-level regulation of negative expected value games versus federal regulation of positive expected value games.
There is also the liquidity.
But any moral distinctions are arbitrary and unnecessary, based on culture.
Today my colleagues on the federal open market committee and I made some important changes to our policy statement (brrr!)
Oh because they are just as much gambles as the other, there are several cultures that require a distinction between gambling and investing, and many cultures that don’t and won’t try to rationalize it just so they can sleep at night.
The only reason positive expected value financial games exist is because there exist certain external cases that make them profitable almost all the time, such as inflation. But if your trade isn't sticking around for that, they aren't inherently positive expected value.
I have pity for the people that feel they need to split hairs over this.... while paying for insurance.
If you have high enough wealth and the reliable ability to replicate your bet a sufficient number of times, then you might be able to reduce the variance enough to see the benefit: right.
If you're not in that position, then you're gambling just the same, whether you have -EV or +EV.
one doesn't work at startups to maximize your earnings. you work at a startup because the pay is sufficient for your goals and its work you want to do (ex: because you really believe in the product, you feel you can grow in ways you couldn't in large companies / have impact or control in ways you can't in lage companies or perhaps others).
If either of those 2 things aren't true, you probably shouldn't choose to work at a startup over a large established company.
I disagree. VCs are great; they are the backbone of our economy and are responsible for creating millions of jobs. A good VC is a huge value-add to any startup and provides valuable advice to aid the value-creation process and streamline the value chain.
What does a good VC add that would be hard to get otherwise? The obvious counterpoint to your argument is that good VCs play primarily passive roles in comparison to the good operators they invest in, by a very wide order of magnitude.
(I'm a VC.) I'm skeptical of simulations like this one because they typically assume that every startup has the same statistical distribution of returns, and whether you pick 15 or 500 startups, each startup will have the same expected value and variance. But in practice it doesn't work like that.
A full time VC might see 2000 pitch decks per year, meet with a few hundred of those companies, and end up investing in 5. So they are picking what they perceive to be the top .25% of all pitches they see. If they invested in 20 companies, it would be the top 1%. If they invested in 200 companies, it would be the top 10%. But because these companies are ranked, the expected outcome distribution of the VC's #1 company should be a little better than the #10 company, and a lot better than the #200 company. The further down the list you go, the worse the expected value -- even though some of the lower ranked companies may turn out to be amazing.
So you can't just 40x the number of investments you make and say "expected value is the same but variance is now much lower!!1!" because you would be lowering your bar a lot by expanding the # of investments. And your variance would go down considerably, but so would your expected value.
(PhD Finance here) Even classical 70s-era finance is against this article:
Rich investors don't need their VC to be super diversified. They can just invest into multiple VC funds and get diversification that way.
Also, it doesn't matter how much risk is in a given VC fund, what matters is how investing part of my porfolio in that VC fund affects the risk of my portfolio!
This was the first thing that occurred to me. Why do they need the VC firms to change from an investment perspective? Just buy the asset class as a whole.
The lesson is VC funds are more like individual stocks than index funds from an investment perspective. Which is fine. In fact VC funds as a class are like a great class of individual stocks, from an investment perspective.
The problem isn't from an investment perspective at all. It's from the VC perspective. VC's which invest in small numbers of companies are trusting the future of their fund on the role of a dice. That is risk they might want to eliminate for their own good.
Your explanation makes sense but is predicated on the VC being able to reliably rank these companies along expected returns.
The perfect VC with perfect foresight might be able to pick just 1 investment and get the best returns from that.
If returns are distributed on a power-law distribution (with the top performer returning a multiple of the second one and so on), then any deviation between the VC ranking and the outcome distribution of returns is very costly. What if I only invest in my predicted top 5, but the real top 1 is ranked 6 in my estimate?
If what matters is to reliably capture the top performers, then the perfect VC would get the top 5 with only 5 investments. But for imperfect VCs, it might be worth it to invest in 50 just to approach 100% chance of capturing these 5 top performers, the 45 others are just the cost of doing that.
> But for imperfect VCs, it might be worth it to invest in 50 just to approach 100% chance of capturing these 5 top performers, the 45 others are just the cost of doing that.
Yep! This is exactly how most funds operate. They invest in 30 or 40 companies over 3-4 years with the hope that 30-40 is enough to hit 1-2 big winners. The more confident a VC is in their picking skills, the more likely they are to invest in 15 or 25 companies in a fund, instead of 30+.
Right. It seems commonly misunderstood that the universe of investable startups represents a zero-sum game for investors. It isn't as simple as "invest in more companies"; one fund's investment represents another fund's forgone investment. (Not many good companies want a five-fund Series A where each fund buys 20%.)
Having the capital and desire to make an investment is the easy part of investing. Actually earning that allocation is difficult.
Yet, most conceptions of investing seem to be shaped by public markets and their ubiquity. That is to say, public market float is taken for granted - when in reality it is missing from every asset class (practically or literally) save for G7 (+China) secondary-market public equities (and now cryptocurrencies).
Leo - as other folks have noted it is a major assumption to think that VCs are good at calculating the expected values of companies in which they make investments. The fundamental randomness is so high here that any ranking a VC made would likely be garbage - the sample size of investments is not nearly high enough to come to a precise viewpoint on expected value of one good deals vs another. As an analogy, if this was a regression the standard errors would be massive, so much so that each investment would not be statistically significantly different from one another in terms of expected return. I think it’s plausible that they’d be able to distinguish best from absolute worst, but best from slightly less good? No way.
I think a more realistic “model” of a VCs behavior is one of thresholding - set the bar somewhere and invest in any company that seems to be above that quality bar, with some stochasticity as to which deals you actually win in the end.
> it is a major assumption to think that VCs are good at calculating the expected values of companies in which they make investments.
It's not quite the same as proving VCs can calculate expected values well, but VC is one of the few asset classes with persistence of returns -- meaning having a top quartile fund is correlated with the next fund also being top quartile. In most investment classes, being top quartile across funds is basically uncorrelated. So good VCs tend to be reliably good and not good just because of pure chance.
> The fundamental randomness is so high here that any ranking a VC made would likely be garbage
A ranking doesn't have to be perfect, just better than random. E.g. let's say you give me a bunch of companies and ask me to rank them, and I put 60% of the actual top 1% and 40% of the next 1% into my top 1%. I'm making mistakes, but that's still pretty good. As long as I'm directionally right on average, I should do well on the investing side.
Go backwards. How many IPOs have not had VC backing? Obviously maybe they grow slower, or maybe IPO bound tech firms are oversubscribed so it’s only an individual VCs prediction ability that’s wrong not the asset class as a whole, but still.
How do you decide the number of companies you invest in each year? Is this simply a function of the available capital or do you take future capital requirements of the companies you want to invest in into account?
Let's say you only pick the top 0.25%, do you do any statistical analysis afterwards to see if this percentage gave the best ROI? Because I would assume that the top 1% at least gets funded by other VCs so it should be possible to do this? Although at the same time I wonder if outsized returns can really be made if there is a VC consensus of the top 1%, my understanding is that the best returns are made in firms that people disagree on.
Good returns can come both from consensus and nonconsensus companies. For example, Robinhood has publicly spoken about the large number of investors that passed on their seed round, while Flexport has been inundated with interest during every funding round.
One big constraint on the # of investments per year for a fund is the amount of time that the fund's partners have. Investors want to differentiate themselves with the help they can provide, and help takes time. As a result, a Series A investor/board member can manage 1-2 new investments per year, while a seed investor might be able to manage 3-8 new investments per year. So # of partners * ~5 investments per year is a good estimate of how many investments a hands on seed fund will do.
There are funds that invest in a lot more companies, but they tend to write small checks and are less hands on.
This was my initial thought too but it is tempered by the "deals we passed on" meme that is prevalent in VC circles. A VCs perception of the best companies is not based on reality. Like a gambler a VC has no idea whether their bet is the right one until much later. Sure you can de-risk it based on previous exits of founder etc but the reality is every VC probably passes on more great companies than they invest in.
Agreed that VCs pass on way more good companies than they invest in. But if a VC invests in .5% of companies they look at, and there are 3 great companies in that .5% and 15 great companies in the other 99.5%, that suggests the VC's ranking is pretty good on a relative basis.
It's impossible to know as getting VC investment increases the odds of success or at least surviving to the point where you can even attempt to have a go. Clearly some are better than others at choosing winners. Part of me thinks that this could be driven by network more than competence though.
>> I'm skeptical of simulations
Yeah, no kidding. Once you see a monte carlo simulation your eyes should roll. This is like all financial predictions that have a bell curve outcome hidden in them. Dice are easy to predict when you roll them many times. Finance is just like dice, just ask the Nobel Prize winners at Long Term Capital Management.
They could've done an analysis on the variance of returns categorized by # of investments per fund. That would illuminate if # of investments is a big correlational/causative factor as they claim.
A simulation alone isn't going to cut it as it involves too many assumptions though it can give rise to reasonable hypothesis (not conclusions).
This is a good point, but I think the point being, you'd 5x your staff and 5x the size of your funnel, thereby ostensibly keeping the same ratio of 'good deals'.
Obviously, at some scale, you literally tap out of opportunities ... but not at just 2K pitches.
Even if we agree that a VC is able to rank investments by high EV and low (non-zero) variance, it's still the case that a small number of investments poses a higher risk than optimal.
> How big is diversified enough? Classical portfolio theory says that in public equities, reasonable diversification effects can be expected when one combines 20 to 30 stocks, while more detailed recent studies set this number at 40 to 70.
Lol, what. No. Diversification is about correlation, not stock count. If you buy 40 different tech companies, you are not diversified. If you buy 3 etfs, you may be extremely well diversified.
> Analysis shows that this handful of successful deals is responsible for around one-fifth of the total cash returned by the industry. To reliably access the excess returns generated by one in 250 deals, a fund size of more than 500 investments is needed. Anything less risks having a portfolio without any mega-winners.
This only matters from the perspective of the fund manager, not an investor. From the perspective of the investor, investing in 1 big fund with 500 companies is equivalent to investing in 2 small funds with 250 companies.
> This counterintuitive result is further validated by a recent Kauffman Fellows study, which says that at the seed stage, “indexing beats 90 to 95 percent of investors picking deals.” In VC as a whole, it seems from Figure 2 that indexing beats 50 to 75 percent of investors picking deals, since indexing gives second-quartile result
When will people stop saying this trash? Indexing beats lots of things. Indexing beats cash, it beats bonds, it beats CDS. But all those things still exist. And they exist because they are uncorrelated asset classes, just like venture capital.
Diversifying your portfolio across uncorrelated asset classes elevates your long term CAGR assuming each returns stream has positive expectancy. This is why people invest in things that have lower expected returns than stocks - those things have less risk, and more importantly, uncorrelated risk. All of this is finance 101, and yet, publications calling themselves 'institutional investor' still don't seem to understand portfolio theory. I guess then they can't write clickbait articles about the great mystery of venture capital.
There is no mystery in VC. Investors diversify across VC funds. If you are thinking about investing in a VC firm, what you care about is uniqueness of the return stream. You want to be able to pick and choose manages who will themselves make diversified bets, so that your overall portfolio is diversified both across companies, and across perspectives. The system as constructed makes complete sense.
> Is venture capital or private equity really uncorrelated to the general stock market?
Somewhat, yes. It's not that it's completely uncorrelated, but it is less correlated than other public companies tend to be.
> It is also difficult to evaluate the volatility of an asset if the price is evaluated infrequently.
Yep, you're absolutely right. This is sometimes called the 'liquidity premium'. People tend to pay less for illiquid assets than liquid assets, all else equal. Which means that investing in illiquid assets should, all else equal, produce better risk-adjusted returns on average over time, especially if you are selling them after they become liquid.
I find that VC investors tend to have more of a "we invest in bull and bear markets" mentality, especially because all investments are long term investments.
I think this creates a smoothing effect. For example, during the COVID flash crash, an investor sold their stake in a VC fund, to another investor at the NAV price.
I realize this is a bit off topic, but no, the parent poster should not. It's a mish-mash of poorly explained introductory probability material. There are numerous other textbooks on statistical aspects of heavy-tailed random variables that would be far more insightful.
Didier Sornette, Critical Phenomena in Natural Sciences: Chaos, Fractals, Selforganization and Disorder. Springer, 2nd edition.
This covers the mathematics of heavy-tailed distributions and a lot of fun applications, and also touches on some other interesting, Taleb-esque topics (fractals, renormalization group).
For financial stuff in this vein (heavy tails), Bouchaud has some books with various coauthors (e.g. Theory of Financial Risk and Derivative Pricing: From Statistical Physics to Risk Management). Admittedly I have only browsed them, but what little I've read has been good.
Also, on a tangent which may interest you, Charles Martin and Mike Mahoney (of UC Berkeley) have some recent work which uncovers heavy-tailed laws in neural networks. You can get started here and follow the citations: https://arxiv.org/abs/1901.08278.
On the upside, that means a lot more founders are given the chance to test out their ideas. That should hopefully lead to more experimentation and faster iteration of ideas in general.
I think this is pretty much what PG and YC have discovered from reading their essays. They're not really concerned about the idea of a startup, if anything they would like there to be more startups to invest. The exponential nature of growth means averaging over large numbers seem to still provide decent returns and all they have to filter for are teams that are really serious about getting something done.
Edit: the 3rd takeaway is advice to those who're considering to join a "startup" - if VC needs 500 investments to make a 15% return on average, you need 5000 years to get the sames returns as a line worker.