Wide diversification is only required when investors do not understand what they are doing.
The venture capital business is a 100% game of outliers- it’s extreme exceptions
[Epistemic status: simplified but kind of true]
Zeitgeist farming is a notion I’ve had in my mind for a long while. It’s the idea of picking a theme well, and doing no other work, and getting rich! It does seem to pop up again and again, and each time it seems to get dismissed by those who know better as a historical accident. So I thought this worth a look.
I've had a dichotomy of investment styles in my head ever since I started investing a decade plus ago. On the one extreme, there's the idea of new-drug investment. You put in billions of dollars and, if you're lucky, it works. Then you get a couple decades to make bank. If your drug works on, say, Alzheimer's or breast cancer, the only question really is how much to charge, since you already know goddamn well that people want it.
If you take PGs admonition to make something people want, this seems like a foregone conclusion. In other words, the business side of the question is easy. It's solved. Just try cure cancer or male pattern baldness. Drug works = make money. But can it work? You have no idea.
The other end of the spectrum is what I call the venture strategy. Imagine Tesla. Not Elon’s Tesla, the Nikola version. If you make AC current, what's that worth? What's the business model? Do people want it? No idea. What if you made a Tesla coil or induction motor or wireless power? No real clue. But there could be an outsized outcome if the tech and business problems were figured out.
In modern day terms, what's the benefit if someone made, say, a new material that supposedly was 10x more thermally conductive compared to existing semicon? If you put gallium nitrate on a substrate of diamond? How would you market it? Sell it? Price it? Compare to the question of what if you made the next best version of a customer service chatbot. The latter is so much easier, because the uncertainty is so much lower. Customer wallet sizes are clearer, purchasing processes are more transparent, the implementation is easier.
Thing is, Silicon Valley was pretty damn good at the second bit of this. They existed to invest in the second bit. To invest in things too risky and unproven and difficult to understand, much less underwrite, as Marc has said. The first strategy requires an understanding of technological feasibility. The second an understanding of business feasibility alongside it.
And whenever the business feasibility is well understood though, you can underwrite the risks easier. (And vice versa).
So if you wanted to start a Chinese restaurant in Mayfair, that's a well understood business problem. The model is clear. You make sensible assumptions on footfall and menu prices and server salaries and so on and the answer pops out.
Same with drug companies, except with a bunch more zeroes. If you make a cure for dementia there's a population and a propensity to pay and reach and you get a reasonable range of answers.
However the natural trend for software in the last decade, after Marc wrote his famous article, has been that the business model risk has been massively reduced or (depending on whom you ask) eliminated. While there's always been tremendous pull from the market for specific solutions to their problems, rarely before have they been so overwhelmed with choice.
Simultaneously, we learnt that in a large percentage of software companies, the success relies on wholesale speedy adoption, which in turn depends on a great product (of course!) but also the creation of a network effect. This isn't new. Microsoft had network effects, famously, decades before. As did any old telephone companies.
What's new was that the network effects could now be created. With judicious application of marketing dollars, you could propel a company from being a pretty good solution to an enormous success. You could create a positive feedback loop of good solution leading to tons of money leading to hiring tons of talent to build an even more amazing solution.
What this meant was that suddenly the Tesla strategy becomes a biotech strategy. If you're sufficiently convinced that a product is good enough, you can throw them a quarter of a billion dollars to go win the market. Because in biotech investing, if the probability of success is assured, the payoffs are magnificent!
In software, it's as if growth investors were getting a direct line of sight into investing in Phase III stage drugs for Phase I prices. Even if inflated slightly to Phase II, the probability of success being semi-predictable makes all the damn difference in the world.
This is also why the best trade in the last half decade has been the monkey see monkey do ape-in trades that required either stupidity or extreme courage. For instance, even after the selloff, if you had:
Bought every software IPO blindly
Blindly followed the meme-investment of the season
Invested at any valuation into the “top” company in any sector (top in quotes because see above the marketing spend)
Buy every crypto L1 ICO
Applied the 1980s algorithms to automated crypto trading
If you're smarter you might have missed these. (Yes, I'm trying to make some of you feel better.)
More importantly, the skill this required is a different one. Not one of Buffett style company analysis but reading the macroeconomic zeitgeist tealeaves. It's moving to analysing things one level up in the aggregation. It's finding that group of no-loss portfolio bets vs that perfect investment. If you want to be fancy, it's about turning your fallibility into an investing asset.
And that's kind of why all become biotech investors. Quality of diligence, that old chestnut, became “good enough” and not “wow!” Because you didn't need to anymore. It's not about secret knowledge or “building conviction” because in an index strategy the conviction is in the strategy, not the company. As John Bogle said:
Don't look for the needle in the haystack. Just buy the haystack!
(Lest this be misconstrued, I'm not saying this is bad. The point of investing is returns, and things that predictably returned 4x in aggregate shouldn't be shooed aside.)
As a corollary, it's also foolishness to praise yourself for some kind of fortitude for missing out on this crazy growth because you believed in a different style of investing. That's an error of omission too, in not realising that circles of competence aren't permanent and fungible. There's no room for patting yourself on the back for not making returns.
What I find absolutely fascinating is that there is a skill here that we consistently seem to underrate, even when it works. When we think of statements like “oh she just got lucky” or “oh they just threw money into every company willy nilly”, we’re internalising the idea that for great results you need to undergo great pains. Whereas what it requires is to know a secret and to choose your level of aggregation.
In the public markets we admonish those who think they can daytrade, and rightly so because index funds are safer. Its a knowledge about our inability to know much about individual companies as much as its about our ability to know something about the world overall. Its silly to feel bad about the ignorance while forgetting to acknowledge the insight.
Looked at in this vein, it's worth thinking what the future will hold for us. A few choices.
Return of the fundamentals. I think this is broadly unlikely because there is too much capital supply, more than demand anyway, and flight to quality is real. What this is likely to mean is a small multiple compression at the top and 100% reduction in deals done at the bottom. Essentially if you're not sure the company will win (back to Phase I trials, in other words), then the chances are you'll not invest.
Return of the OG venture capital strategy. With new types of companies emerging who are harder to understand technologically and still work in progress, the original investment thesis of VCs will come back. This holds as much for portable fusion, new materials, consumer medical devices, or more. And since the large pools of capital that exist today don't really want to invest in this strategy, it remains called VC but stays a separate bucket.
Software continues to eat the world. Large funds and crossover investors pile into potential winners in known sectors, continuing to pay up for optionality to win the market, and this continues unabashedly long as nobody looks too closely at the V part of VC.
We’re seeing some of all three. There are already hints of investors getting pickier, which will create an even more bifurcated market. There are also hints of investors doing really well investing in things heretofore considered esoteric - whether that’s defense systems (Anduril), prolonging human life (Altos Labs), low cost satellites (Astranis), drug discovery (Alpha Bio), just to stick with the As. There’s a lot more building to be done ahead of us.
Regime changes are notoriously unstable. And this regime change will no doubt be frustrating and chaotic. Whether it's interesting times or not, I'm looking forward to seeing this play out!
Interesting article! I have a couple of questions:
How is Zeitgeist Farming different from Factor analysis? Or is A a subset of B?
On my newsletter, I have been writing about investment strategies which I called "alternative" for the lack of a better word, but many of them would fit into the framework of Zeitgeist farming - For example, you suggested in this article that investing in all tech IPOs at inception might be a good idea, and I actually wrote an article analyzing whether you could make money off of buying IPOs (https://marketsentiment.substack.com/p/can-you-make-money-from-ipos - The results weren't as favorable as you might think!). Other ideas of mine included "an automated trading strategy for crypto" and "investing in the most reputable brands in the world" which correlated with some ideas here. Would be interesting to know your thoughts on how you distinguish between an arbitrary factor and a "zeitgeist" factor.
My issue here is, as George pointed out earlier, that a) It's hard to identify the wave when you're in it, and b) It's hard to predict how long the opportunity will last. (Though my analyses are based on years of data, past returns are no indicator of future performance possibilities and figuring out how to account for that is a problem I'm grappling with. Any thoughts on this appreciated!)
In principle I agree on this, but in practice I think it's much harder.
There's the situation where in 2016 you buy BTC and it jumps and you are happy.
But there's also the situation in which you buy in 2017 and you have to wait until now to make your money back 3x, if you actually manage to sell at the peak, or loss a lot if you panic, and you loss on the upside of just having bought an index in 2017 which by now would have been 1.5-5x, with the maximum downside having been maybe 0.6-0.7x (outside of a global-war situation, in which case money starts working differently anyway)
Or there's the one situation in which you just buy "Long Blockchain Corp" in 2017 instead and now you've just lost a whole lot of money.
The interesting thing with longevity companies, for example, is that the funders are the same but they like company hoping. I think anyone in their right mind knows right now that rapalogs or some form of NR/NMN are going to be big, like 1000x big compared to today. But if the business model of the companies selling is to go public when they are sensing immediate loss, and if there are 1000x companies trying to do rapalogs, you might well still lose on the one insider-traded company that makes it and just get losses from buying all the IPOed failures.
The problem with zeitgeist farming, in other words, is that you might be overfiting on the winners and forgetting how the zeitgeist "felt like" back when the winners were on the market for cheap.
Which is not to say in principle I don't agree, but it doesn't seem easy.
One anecdote: I have a significant (20%) chunk of my savings into tobacco comapnies. Why? Because I've seen how vapes sell on school yards around the world, and I've tried them myself and monitored what happens and they are as damn side-effect-free as it gets given how good they are. Am I zeitgesit farming or being insane? Is my read of highschoolers good here? What about regulatory action that could kill the whole industry or at least place it under government monopoly? Maybe I am reading this correctly but vape-focused business models won't be enough to replace the revenue from tobacco once cheaper models hit the market. Maybe an addictive drug replace nicotine entirely.
Another framing: Silicon Valley style software is amazing right up until the point where NLP models can write good software and someone with enough compute open-sources alternatives to every single piece of proprietary software. This is precisely the kind of action nobody can predict that can entirely de-value the tech segment to 0 (while making the world better, so we are not in a "don't invest in worlds you wouldn't want to live in" type scenario) -- And there's totally a zeitgeist among AI researchers where that will happen, and we should be banking on that world instead.
But I digress,
On the whole I'm bookmarking the article as a memento, I think it's a great idea and very well put.