You Learn From Good Businesses and Bad Trades
Plus! ByteDance, Clubhouse, China, Italy, Cloud, Facebook, more...
This is the weekly free edition of The Diff, the newsletter that tracks inflections in finance and tech. Last week’s subscriber-only posts:
Bretton Woods Revisited: to understand how the dollar works today, you need to understand the previous monetary framework, which turns out to be an incredibly baroque system that befuddled central bankers, terrified Presidents, and increased the probability of nuclear war. On the plus side, it made calculating the cost of European vacations slightly easier.
Dropbox, Information Asymmetry, and Smart Money. A good life hack when you face uncertainty is to figure out what someone with better information is up to, and proceed from there. Dropbox is a case study.
Carta: Bringing Tech Companies Back to the Futuristic 1990s: I’ve been obsessed with private company stock for a decade. Late IPOs and high transaction costs make the market appealing and challenging. In this writeup, I look at Carta’s plans for a private exchange. They’re compelling. Carta’s been playing the long game; they started out by solving the problem that other exchanges only run into when they’ve been operating for a while.
Why Does “Commoditize the Complement” Work? Supply chains approach a state where they have alternating layers of commodities and monopolies. Think blogs and Facebook, e-commerce and Google or Amazon, beige box PCs and Microsoft. This post explores the mechanics of commoditization.
Editor’s Note: Effective June 5th, the price for new Diff subscriptions will be $20/month or $220/year. Subscribers who sign up before June 5th will pay the current price of $15/month (25% off) or $150/year (32% off).
The Diff will be off for Memorial Day Monday.
You Learn From Good Businesses and Bad Trades
I’ve learned a lot from failure. Specifically, I’ve learned a lot about how it’s incredibly unpleasant and should be avoided. From the vantage point of the comfortably successful or the not-directly-participating, the cost of a graceful pivot or a quick decision to cut losses sounds like tuition. Actually experiencing it is a painful grind, which is why I’m a big fan of learning from other people’s mistakes and missed opportunities.
But which? There’s no shortage of information out there, but time is limited and you want to focus on the higher-order bits. The heuristic I’ve found is:
Good businesses are more informative than bad businesses, because judgment is correlated across domains: a business that was killed by a fatal mistake probably made several fatal mistakes.
Bad trades are more informative than good trades. While there’s an element of randomness to financial outcomes, a good trade constitutes an accurate prediction about the future, which means turning a profit on the trade has no informational content. If you heard about Covid-19 in January and bought puts—congratulations, first of all, but—you didn’t learn anything from the trade, because you figured out what would happen in advance.
Bad Businesses are Overdetermined
Annoyingly, there’s intense selection pressure in financial media for stories about bad businesses and good trades.
I’ve written about the business books: they tend to get written when there’s a source and an axe to grind, and they’re more entertaining when there’s a moral arc. A failure like Enron or Theranos is a Greek tragedy, with the hero undone by hubris. But a successful business is not a classical narrative: the standard Hero’s Journey doesn’t end with the hero as a middle-aged rich guy with a vacation home in Aspen.
Financial media report on big wins and big losses, but since the target audience for financial media is people who want to get rich, there’s more emphasis on the wins. (If you wanted to not-get-poor instead of staying rich, you’d probably focus on your day job and put your money in some kind of all-weather or 60/40 passive portfolio. If you’re reading the Journal or FT, you probably have more ambition.) One piece of evidence for this is that one of the worst trades of all time was selling credit protection against AAA-rated tranches of private-label mortgage-backed securities. And there are writeups of this trade—about the winners.
Why Success In Business is Informative
A successful companies a) has the right idea, b) executes better than the competition, and c) avoids every company-killing mistake. This is practically tautological, but it’s a sort of Pigeonhole Principle for narratives: if there were several companies trying to win in social, search, streaming video, operating systems, CRM software, etc., but only one company truly won in that space, then it probably avoided doing something its competitors did wrong.
The idea is important, too. Idea value is paradoxical, because the market value of ideas is low—you can’t sell one, and the closest you get to a market for ideas is patent litigation. But ideas are a limiting reagent. Fortunately, founder-idea fit tends to click early; a founder will try out a few disparate businesses, but eventually one of them will take off, and in retrospect it will be clear that the idea that worked was the cleanest version of the thesis behind the failures. In the very early days, Facebook focused more on a file-sharing product called Wirehog than on Facebook itself. But the point of Wirehog was to connect people online, by way of the files they had. As digital cameras got more common, the sharing loop lost a step: instead of create, save, upload, the process was create-then-share, and file-upload part was superfluous.
Google had a product before they had a business, but the core idea of the product was that the structure of the Internet had latent information about the meaning, relevance, and trustworthiness of every single page. AdWords ultimately used the same logic: a second-price auction is just a way to aggregate data on every company’s incremental pretax profit from a click, and charge them an amount as close to that as possible.
Amazon started with books, because they had few wholesalers, many units, and plentiful metadata. Searching by title, author, genre, etc. was a perfect application for the web circa 1994, but Amazon knew from the beginning that books were the easiest place to start, but not the place to stop.
Successful founders like to talk about early failures and pivots, but that’s mostly a feel-good exercise. Some companies were founded by first-time founders, who may or may not have gotten lucky. Bill Gates and Mark Zuckerberg both founded companies before the ones they’re known for: Gates had a traffic analytics company called Traf-o-Data, and Zuck had a music player called Synapse. These look like failures compared to Microsoft and Facebook, but: both early failures were founded by high school students and had some measures of validation (Traf-o-Data got government contracts; Synapse turned down a million-dollar offer). If that was all they had done, they’d look like failures compared to how successful they are, but they’d have been minor legends at their high schools.
A great company is a bit like a great novel or work of art: there’s an idea, followed by a lot of effort to make the finished product reflect the original vision. Sometimes, that requires backtracking and refining. But it’s with a definite goal in mind.
By contrast, failures don’t tell you much, because too much went wrong. Enron is my single favorite failure. Here’s a list of plausible reasons they died:
Too many levered investments in low-return infrastructure projects.
A “run on the bank” from traders pushing against Enron’s futures positions.
A complicated balance sheet with too many ratings- and stock-price-based “tripwires” that made any slowdown in the business an instant funding crisis.
Volatile trading results, and a habit of smoothing that volatility by deferring loss recognition in bad quarters and deferring gain recognition in good quarters.
Willingness to adhere to the letter of the law and violate the spirit of the law when trading in the spot electricity market.
Internal political infighting.
The actual fraudulent transactions that generated totally fake income.
Importantly, any one of these problems could have killed the company. Enron had some good businesses, some bad ones, and some fake ones. Even if they’d never actively cooked the books, or done so less aggressively, the combination of leverage, growth, and intra-management conflict would have eventually done them in, or a bad quarter’s trading could have prompted a liquidity death spiral.
Good and bad businesses judgment are correlated: someone like Bill Gates, who knew that there would be a computer on every desktop decades before it happened, was also the sort of person to negotiate smart deals with commodity PC companies, and also the kind of person who would contribute code years into his tenure as CEO. And at Enron, a chairman who would countenance levered high-risk investments in politically-controversial developing world deals was also the sort of chairman who wouldn’t notice that a separate Enron division had sent California back to the dark age.
Short sellers like to say that accounting problems are like cockroaches, and they’re right. The best predictor of a big accounting scandal is a series of small ones (Enron had a smaller fraud in their trading division in the late 80s). But that means that you never know every mistake they made, just the ones where the coverup was insufficiently thorough. In an interview with Tim Ferriss, Peter Thiel describes failures as overdetermined: “You will think it failed for Reason 1, but it failed for Reasons 1 through 5. And so the next business you start will fail for Reason 2, and then for 3 and so on.” Reading about failures business failures is fun, but successes are the ones that give you new information.
Bad Trades Are Bad Luck Compounded By Cognitive Bias
The informative nature of bad trades is another proof-by-negation: you just don’t learn much from making money. A trade is a prediction, and if your prediction comes true, that’s great! You saw the future! But that also means that when you’re right, it’s because conventional wisdom converged with your view, so now you believe the conventional wisdom.
It’s a kind of strange feeling to get the trade right. You feel good, and then you immediately feel like you’re back to zero; in the double-entry sense, a profit is just shifting balance sheet entries from “valuable insight” to “cash of equivalent value.” Finance is a domain where you can sell an idea, though that’s an iterated game. In a sense, the structure of a hedge fund is a form of DRM, a way to sell IP (trading ideas) in a format that can’t be easily duplicated and shared.
Countless investors have talked about how easy the job is when they’re right
In Soros on Soros, George Soros says:
More generally, running an investment portfolio is not work in the ordinary sense of the word. It is something else. It is risk taking. The amount of work you need to do is inversely related to your success. That is to say, if you are working at what is normally considered work, if you are a salesman or a craftsman, your success is directly related to the amount of work you put in. The more you hammer away, the more goods you produce; the more customers you visit, the more orders you are likely to take. There is a direct relationship. When you are taking risks, if you make the right judgment, if you have the right insight, then you don’t need to work very hard. But if you make a mistake, and there is a divergence between your hypothesis and the actual course of events, you need to do some really serious research to find out what’s wrong. The less successful you are, the more you’re going to have to work to correct the situation. If the portfolio is doing well, you’ll have to work less. There is an inverse relationship.
Here’s Buffett: “Time is the friend of the wonderful company, the enemy of the mediocre.”
And Phil Fisher: “[F]inding the really outstanding companies and staying with them through all the fluctuations of a gyrating market proved far more profitable to far more people than did the more colorful practice of trying to buy them cheap and sell them dear.”
And Edwin Lefèvre, probably paraphrasing Jesse Livermore:
It never was my thinking that made the big money for me. It always was my sitting. Got that? My sitting tight! It is no trick at all to be right on the market. You always find lots of early bulls in bull markets and early bears in bear markets. I’ve known many men who were right at exactly the right time, and began buying or selling stocks when prices were at the very level which should show the greatest profit. And their experience invariably matched mine—that is, they made no real money out of it. Men who can both be right and sit tight are uncommon.
So there you have it. Of the four great investing traditions: macro, value, growth, and degenerate gambling, everyone agrees that holding on to winners is a good idea, which means that getting rid of bad investments is the hard part.
It’s hard for typical cognitive-bias reasons: most people are averse to taking losses, and anxious to get to breakeven. And past a certain point, more information leads to higher confidence rather than better judgment. Combine these two factors, and someone who makes a bad trade is psychologically primed to justify it and factually well-prepared to do so.
At the start of this year, I had a large (for me) position in European dividend futures. It did not go well, and I exited the trade at a substantial loss. I did a fair amount of research beforehand, got reasonably confident that it was a sensible risk/reward decision, and plunged in. The futures represent a bet on a single year’s dividends from an index of European stocks; when the future matures, it pays out at that dividend—so while it can be mispriced, you at least know when the gap between expectations and reality will go away.
For a while, things went well: the general cadence of the trade is that the dividend futures gradually appreciate over time, but drop by more than they fundamentally should when the market declines. So I could follow a fairly lazy strategy: buy and hold, and buy the dip. I listed a bunch of specific risks to the trade in my writeup, but there was one risk I alluded to as a side comment and totally ignored in practice: “A lot more can go wrong in Europe by year-end 2021 than year-end 2020.” As it turns out, this was not just an obvious point about the fact that time moves in one direction: it was also an admission that a single catastrophic event could cause dividends in one year to suddenly collapse.
So I learned something from that trade: think about the exact scenario you’re betting on, and think about the precise ways that scenario could change. “Dividend futures” are an abstraction, but any one year’s contract is a bet on what happens in that specific year. Now that we live in 2020, The Year Of Rescaled Y Axes, I’m aware of that.
A more concrete lesson is to use stop-loss orders. A stop-loss is an order to sell something you own if it drops or to buy something you’re shorting if it rises. I’m deeply philosophically opposed. If you thought the stock was cheap at $20, and you precommit to automatically selling your position at $18, that’s totally irrational. What I’ve realized is that stop losses are irrational in a fixed, known quantity; you’re committing to buy high and sell low. But anyone with a losing position is arbitrarily irrational. A stop loss order is a put option on your own self-destructive insanity.
A stop-loss is not a portfolio-management tool; it’s a psychological management tool. What you should do with every position is ask yourself if you’d size up the trade at the current price, and exit if the answer is no. In practice, that’s a hard question to answer accurately when you’re down. Many of the best investors do sell their losers, but usually because the stock has dropped in response to fundamentals and they’ve recognized this. (Anchoring bias is real, and momentum effects persist: stocks are a little more likely to underreact than overreact to important news, so the stock that looked cheap at $20 might be expensive at $18.) Value investors generally don’t use stop losses, but the difference between bad ones and great ones is that the great ones don’t keep adding to positions when they don’t work out; a bad value investor is someone who ends up with 100% of their money in Kodak, which had a low P/E all the way down to zero.
Looking at my trading records, the really surprising thing to me is that I mostly didn’t do anything as futures prices plummeted. It would have been more intellectually honest to buy some, and lose even more money, but I didn’t do that. So that’s a lesson I’ll be remembering: for every trade I have on, I should ask myself if, given an unrelated cash windfall, I’d be adding to it right now. If the answer is no, time to exit the trade.
That’s easy to type because I’ve heard it before many times. But now I really know it.
A Word From Our Sponsors
What’s the one thing in every hedge fund titan’s portfolio that you’re probably not investing in? A-R-T. In fact, 84% of ultra-high-net-worth individuals collect art according to a 2019 Deloitte survey. It makes sense—art has outperformed the S&P by over 180% since 2000, with little of the downside risk. And with the total art market expected to balloon from $1.7T to $2.6T by 2026, it’s no wonder that the price of paintings have skyrocketed.
One New York startup is at the center of it all: Masterworks. They’ve fractionalized multimillion-dollar masterpieces by Warhol, Basquiat, Banksy, and more—and you can be a part of it. They’ve allowed our readers to skip the 25,000 waitlist, so do yourself a favor and check it out.
I recorded another podcast with Erik Torenberg of Village Global. Topics include: MMT, institutional trust, journalism, pensions, and Bitcoin. And on Coindesk, some thoughts on Paul Tudor Jones' Bitcoin trade.
The Information has a great writeup on TikTok parent company ByteDance, with a focus on their advertising plans. ByteDance has been an incredible platform for creators:
And that growth comes from ably matching content to viewers—exactly the sort of skill that applies directly to targeting ads.
Meanwhile, the US can still produce novel social apps. Clubhouse has a tiny userbase unless you measure it by Twitter followers or assets under management. It looks like a goofy novelty, but it might be something more important: there’s been a persistent pattern in social media of tighter feedback loops between posting content and getting feedback. Static sites have a very slow feedback loop; blogs with subscription tools are faster; Twitter is faster than that, but Snapchat is close to instantaneous since it makes sense for users to get a push notification every time there’s something new for them to see. Clubhouse found a way to make it even faster, by making it real-time only. You can get last night’s Snap or last year’s tweet, but Clubhouse content expires immediately.
This also has privacy implications, at least for the everyday threat model of having something you said taken out of context long after the fact. Unless someone else in the room is recording you, your Clubhouse comments vanish as soon as you’re done talking.
What’s China Up To?
It’s always sad when a beloved compounder pulls guidance: China is not setting a GDP growth target for 2020. This makes perfect sense. There’s no reason for them to stretch to hit the number when much of what they need to do won’t positively impact GDP.
That said, there are some interesting datapoints coming out of China recently. For example, the FT reports that China is streamlining its iron ore import process. Iron is an input into infrastructure, and infrastructure spending is China’s classic approach to stimulating the economy during demand shortfalls. They used a lot of rebar in 2009 to keep unemployment low.
But in post-industrial-revolution terms, they still have catching up to do: Semiconductor Digest notes that China is well behind its Made In China 2025 goals for semiconductor production. And, more damning: China-headquartered firms are only responsible for 39% of semiconductor production in China. This is an ironic reversal of Shenzhen’s enduring importance. China is hard to displace as a manufacturing hub because their human and institutional capital in that area is tied to big machines that are hard to move; if you duplicated every factory in Shenzhen and put them somewhere else, you’d have no hope of competing, because you wouldn’t have the managers, purchasing agents, and engineers who keep the system optimized. The same force may make it hard for China to succeed in semiconductors: they can certainly supply the capital, but if the right engineers are in Hsinchu and Hillsboro, that’s just a stranded asset.
At a time when capital and free time are abundant, and out-of-home entertainment is hard to come by, cloud computing is a very good bet. The FT has an overview of cloud competition in Asia. One notable takeaway: outside of China, four of the top five cloud computing companies in Asia are American. Within China, four of the top five are Chinese, and one is an Amazon joint venture.
Cloud economics are ideal for the panic-then-relax cadence of Covid-19. They’re the fastest way to get to arbitrary scale for a surge in demand—but a forgotten S3 bucket or EC2 instance is basically an annuity. Cloud spending is a complement to engineering spend, but has the added benefit that when layoffs happen, companies lose the one person who knows whether the data they’re paying to store is essential or not. So cloud spending is not just a bet on Internet usage, but a bet on volatility within the Internet sector.
Facebook Goes More Remote
Facebook has announced that it’s adding hubs in cities like Dallas and Atlanta, and that the company could be 50% remote in a decade. I maintain that residential real estate in superstar cities will be less impacted than other cities by this: it’s harder for two-earner couples to leave, and proximity to the main office still matters even if it’s a bit further away.
Facebook also plans to adjust salaries based on cost of living. Since Facebook is still hiring, they’re a price-setter for engineering talent, meaning that tech workers will be teleported back about half a century, to a time when your standard of living was the same whether you lived in the Midwest or California, but you could choose CA because the weather was better.
Italy’s Financial Engineering
The government of Italy plans to make equity investments in major Italian companies. Can they afford to? In a sense: “Under European accounting rules, such financial holdings have no impact on countries’ budget deficits.” As it turns out, the EU’s rules set countries up for some very interesting rogue trading. They can fund themselves by borrowing from retail investors, whose savings are up due to typical recession saving patterns, and up more because so many spending venues are unavailable. They can transfer this money to companies, which will keep workers on payroll, and those workers will continue to save a larger share of their incomes, which they can reinvest in more government debt. The payoff function for the Italian government is favorable: if Italy needs a bailout, the size of the bailout will be based on the size of the fiscal hole they’re in. If they don’t need a bailout, the investments will appreciate. Looking at the transactions, this policy is funded by Italian savers. But looking at the outcomes, it’s funded by France and Germany. (Who are both willing to cut checks.)