Also best ideas, op-risk cat bonds and unicorn valuations.
World Cup predictions.
The World Cup kicks off today, minutes after this newsletter publishes, which means that there will be no finance news for the next month. That will make this newsletter a bit of a slog, sorry.
So I guess let’s talk about the World Cup? There is a long tradition of investment banks publishing research notes in which they predict the winner of the World Cup using some sort of half-baked statistical method. These predictions are always pretty dumb, which makes sense, because the analysts are in the business of predicting investment returns—which is hard enough!—and not in the business of predicting soccer results.
But of course the analysts are not really in the business of predicting investment returns. They are in the business of producing content that will improve their banks’ relationship with their customers, giving the customers a reason to trade with the bank, giving the banks’s salespeople a good excuse to call the customers. The World Cup reports fit perfectly with that business. If you go write a report about a clever interest-rate options strategy, and send it out today, and your salespeople follow up by calling clients tomorrow afternoon, no one will answer the phone because, come on, Spain will be playing Portugal. But if you go write a report about who will win the World Cup, your salespeople can call up customers and be like “hey bro, Brazil, good stuff, good stuff,” and the clients will be like “oh yeah man, true, true.” Everyone is going to be wasting a month on the World Cup anyway; you might as well waste some of it building client goodwill.
Of course this only works (I mean, “works,” whatever) in a world in which banks give away investment research for free as a relationship-building tool, and monetize it in the form of commissions on trades. The Mifid II regime is shifting European research away from that model and toward one where customers have to pay for research. Customers are not going to pay for World Cup research. If European regulators kill sell-side World Cup research… honestly it will be fine, no one will miss it all that much, but it’ll be a little bit sad. If you make research more transactional and professional and value-oriented, then it will just be so transactional and professional and value-oriented. The old murkier business models preserve some inefficiencies, but here and there they also preserve some charming quirks.
Anyway here’s a roundup of this year’s predictions, and there’s some of the usual dumb stuff:
Blah.
But then there is this from Goldman Sachs:
This is a little different: It’s not just using the analysts’ financial expertise to predict soccer results (why?); it’s using their machine-learning expertise to predict soccer results. The thing about machine learning in finance is that domain expertise always seems so optional. I am constantly reading stories about biology researchers or hobbyist coders, who barely know what a stock is, building machine-learning models that outperform the experts. (“It’s a little embarrassing, no,” I once wrote: “That investing is best understood by people who don’t understand investing? That it’s a trivial application of broader data-science principles, best addressed by people who were trained on harder and more interesting applications?”) This is the same line of thinking but in reverse: Goldman’s analysts know enough about soccer to throw in an awkward “stay on the bench” metaphor, but it is not exactly the subject of their Ph. D. research. (I hope.) But if you can build a machine-learning model, you can build a machine-learning model; what it is learning doesn’t much matter.
What if Goldman’s model works? I mean, I doubt it; I suspect there are not nearly enough data points for a machine-learning model to get good at predicting the World Cup. (Also: “Goldman has predicted a Brazilian victory for the last three World Cups, and has been wrong every time.”) But if it does? On the one hand, the magic of machine learning would have allowed some random finance types to outperform the experts at predicting the World Cup. On the other hand, that same magic should allow random … biology, or physics, or internet-advertising, or sports-analytics … types to outperform the experts at predicting stock returns. If the World Cup is as easy as that, maybe finance is too.
Elsewhere: “Argentina prisoners call hunger strike to get TV fixed in time for World Cup. ”
Best ideas.
“Hedge Funds’ Best Ideas? Those Are Just Stocks They’re Dumping,” reads the headline here, but the story—by Bloomberg’s Sarah Ponczek, based on a paper called “Talking Your Book: Evidence from Stock Pitches at Investment Conferences” by Patrick Luo of Harvard—is actually rather heartening. What you learn about hedge funds’ “best ideas” that they pitch at investment conferences is:
That’s all good! Famous hedge-fund managers have some skill at picking stocks. They have some metacognitive skills that allow them to recognize which of their ideas are particularly strong. And when they tell you to buy a stock, you can believe them. It’s all sort of sweet and charming. On the other hand they’re not doing it out of pure altruism:
I mean, they’re former best ideas, not future best ideas, but you can’t have everything.
Op-risk cat bonds.
Once upon a time, the big global investment banks were wild places, where creative bankers competed to find novel and aggressive ways to slice up risks and sell them to buyers who were hungry for attractively packaged risks. Then the global financial crisis happened, and the banks lost a lot of money, and the attractively packaged risks stopped looking so attractive, and the banks started paying multibillion-dollar fines every 15 minutes, and the regulators stepped in to rein in the worst excesses of the risk-repackaging business, and the whole thing got a lot less wild. And if, like me, you get a lot of aesthetic enjoyment out of crazy financial products, then you will find less aesthetic enjoyment in banking these days. It’s all, like, ooh, zero-commission mobile trading, whatever.
But Credit Suissse AG’s operational risk catastrophe bonds are among my very favorite post-crisis financial products, not only because they are pretty wild, but also because they are wild in such a specifically post-crisis way. After the crisis, regulators started imposing big fines and monitoring banks more closely for misbehavior and fraud. And Credit Suisse looked around and said: Yes, sure, we can sell that.
And so they bought operational-risk insurance from Zurich Insurance Co. Ltd., which would pay out if Credit Suisse lost money due to rogue traders or fraud by its employees or computer problems or whatever. And Zurich sold catastrophe bonds to reinsure that risk, so that the capital markets would ultimately own (some of) the risk that Credit Suisse would commit fraud. “What if Credit Suisse is committing fraud when it sells the bonds?,” I asked . “Can the bondholders sue Credit Suisse for fraud? If they win, do they have to pay Credit Suisse back?” (I also suggested that the obvious buyer for these bonds would be Credit Suisse’s bonus pool: Why not make the employees own the sliced-up risk that they will misbehave?)
Anyway they are at it again, or at least, Zurich is selling new catastrophe bonds that Artemis reports will cover Credit Suisse:
In the decade since the financial crisis, the focus at investment banks has shifted from the creative slicing of risk to more mundane stuff like regulatory compliance and cybersecurity. And Credit Suisse is creatively slicing that mundane stuff and selling it. It is a real triumph of the human spirit in the face of adversity, or of boredom anyway.
People are worried about unicorns.
It feels like it’s been a while since I have read any dire warnings about overstretched valuations of private technology companies and the imminent crash of the unicorn bubble. I guess everything is fine? “E-scooter company Bird is seeking to raise around $200 million in new funding at a $2 billion valuation,” reported Dan Primack on Tuesday, “ just weeks after it raised $150 million at a $1 billion valuation, and only three months after raising at a $300 million valuation.
