Investment acumen is hard to measure. In the short term, skill is indistinguishable from luck, and if you wait until the long term has arrived, skilled investment managers won’t take your money. So asset allocators — fund LPs, heads of multi-manager funds, and portfolio managers who work with teams of analysts — have developed short-term proxies for measuring long-term skill.
Some of these are brain-dead obvious: someone who is smart and hard-working will probably outperform someone who is dumb and lazy, unless markets are perfectly efficient. (If you believe markets are perfectly efficient, close this tab — you’ll be able to instantly, costlessly find something exactly as interesting to read.)
There are some less obvious criteria. Some asset allocators look for evidence of an entrepreneurial bent, or risk-tolerance, often at a fine-grained level — Aaron Brown has written at length about the difference between traders who play poker and traders who bet on horses.
One fun indicator of investment skill is idea velocity: how often can you come up with new ideas?
This sounds like the wrong thing to optimize for. If nothing else, someone who comes up with lots of ideas will end up paying lots of commissions when they trade on those ideas, which is bad for returns. And investments tend to follow power-law distributions: a few big trades dominate your results.
Some strategies don’t appear to follow a power-law, but it’s usually because they’re being evaluated at the wrong timescale. For example, if you come up with some intraday trading signal that applies to hundreds of stocks, each trade contributes a tiny fraction of a basis point to your overall returns. But those trades are the implementation; the signal is the idea.
I don’t have a ton of data on what quant strategies are live right now, but there’s good historical data on the failures that started out as successes: Long-Term Capital Management was famously diversified — they justified their massive leverage because they had so many different trades on that it would be hard for them all to go wrong at once. But in LTCM’s first two years of operation, almost 40% of their profits came from a single strategy, a tax arbitrage in Italian bonds.
There are two ways to argue for idea velocity:
- It’s hard to judge the profitability of a strategy or trade in advance, and comparatively easy to scale it up and refine it later, so each new idea is a call option on a big idea.
- Idea velocity helps managers stay disciplined about cutting losses. You’re more willing to exit a trade if you’re confident you can replace it with something good; if you have relatively few ideas, selling out of the bad ones just leaves you less diversified.
New Ideas as Call Options on Big Ideas
Managing a portfolio is a continuous, iterative process. You start out with a sketch and fill things in as you get a better idea of what you’re doing. There are very few investors who initially size their ideas correctly, but the best tend to add to them — as Peter Thiel pointed out, in reference to a deal he missed, “Whenever a tech startup has a strong up round led by a top tier investor… it is generally still undervalued. The steeper the up round, the greater the undervaluation.” The Soros $10 billion pound short started out as a billion-dollar pound short. Most of Paulson’s mortgage short money came after his counterparties tried to get out of the trade.
The ideal investment is not just something fundamentally mispriced: it’s something that’s mispriced for a known reason. This means that subsequent price performance is part of the event path that an investor has in mind when they make a trade. For example, if you knew that CDS pricing models promised a risk-free arbitrage if held to maturity, but that many of the people writing those contracts would end up insolvent in the event of a big price move, you’d know that a long-CDS trade had a better risk-reward after the contracts in question had gone up, because your counterparties would be desperate to get out. If you’re betting on one winner in a competitive market with network effects and economies of scale, part of your return comes from losers being unable to admit that they’d lost. MySpace looked like a competitive threat for longer than it was a threat.
I think of some trades as “sticky-note trades.” Somehow having a small position, and making or losing a small amount of money every day, focuses the mind a lot more than just having a vague idea that there’s an opportunity. This doesn’t work all the time, and doesn’t work for everyone; Warren Buffett bought 100 shares of Microsoft in the 80s so he’d get the annual report, and would have benefited from owning a whole lot more. But it works for me.
Part of the psychology here is averting the aggravation of having made/lost an insignificant amount of money for not good reason. It’s like getting a gym membership so you’ll feel like a schmuck if you don’t work out — except that, for some reason, it actually works.
Idea Velocity and Loss-Cutting Discipline
It’s received wisdom among traders that you should cut positions when they move against you, and add when they work. There’s some academic evidence that this is true — momentum predicts returns across many asset classes, in different countries and over different timescales. But a momentum strategy is easy to automate, so you’d expect this trader wisdom to get arbed away.
Why do traders still talk about cutting losses? Why do the best of them seem to do it? Why is it that even people who would slap you if you called them a trader with good stop-loss discipline — classic value investors, say, or VCs — still make most of their returns by holding on to the good stuff for a long, long time?
I think the answer is psychological, not financial. Let’s suppose you’re an investor with some kind of edge: you really do have an eye for which currencies will appreciate, which startups are worth the post-Demo Day premium, which defaulted bonds will pay out, whatever. You may be smart, but you’re still human, and human beings are absolutely terrible risk managers. When we’re right, we skittishly reduce our exposure; when we’re wrong, we construct increasingly elaborate theories for why we’re more right than ever.
Some investors are just wired differently, and correctly respond to new evidence. Warren Buffett isn’t putting in trailing stop-loss orders; he just admits it when he made a bad call. But for the rest of us, it’s useful to impose some kind of discipline, and loss-cutting is the single best way to do that. Anyone who has ever overheard a couple fighting knows how quickly we dig in, reason be damned, when someone tells us we’re wrong. And markets do that all the time; asset prices fluctuate, and on a short enough timescale 50% of the time they move in the wrong direction.
Loss cutting applies to other fields, too. Faulkner allegedly said “In writing, you must kill all your darlings,” and it’s completely true. Whenever I’ve worked on a collaborative project, the stuff I’ve fought hardest to keep has turned out to be the most self-indulgent. I’ve actually considered an editing process where I print up each post, read through it, underline my very favorite line, and ruthlessly delete it.
But… that was easy for Faulkner to say. (Assuming he said it. Apparently it was actually some other writer.) You can only be confident enough to kill bad ideas when you know you’ll have more ideas. That’s where idea velocity comes in. Force yourself to develop new opinions frequently and you’ll be happy to discard the tarnished ones when the time is right.
To get on a regular schedule of generating new, actionable ideas, you need a few big themes that can spin off more specific theses. My favorite example of this is “the peace dividend of the smartphone war.” It’s the kind of thesis you invent if you spend years investing in companies that benefit from the shift to mobile, until you realize — you’re not getting many good ideas. The action is elsewhere. So you start to ask: if smartphones and apps are both getting monopolized, what are the side effects? Monopolies beget commoditized supply chains, but some parts of the smartphone supply chain can contribute to other products instead.
Charlie Munger talks about accumulating mental models, and finding cases where they compound. He explains the history of Coca-Cola in terms of branding, Pavlovian conditioning, forming positive associations, making the product ubiquitous, commoditizing the complement, etc. In other talks, he hammers home the point that incentives powerfully shape behavior — don’t ask what people are supposed to do, or what they say they’ll do; ask what they’re rewarded for and punished for and you’ll know how they’ll behave.
These frameworks add up. Coca-Cola, for example, wasn’t just a bet on the product — it was also a bet that a management team compensated by ROI rather than profit would return capital to shareholders rather than build a bigger but lower-return empire for themselves.
Nassim Taleb has built a whole career on the observation that people use normal distributions in situations where the underlying distribution is likely to have fat tails. That’s just one claim, but it ends up giving him ammunition to talk about financial markets, history, politics, and diet
Idea Velocity in Practice
Idea velocity comes from public market investors because it’s easiest to implement in public markets. Suppose you have a portfolio of, say, ten longs and fifteen shorts. That’s 25 positions. If you have one decent idea every trading day, you can measure that in terms of turnover; you should be turning over your entire portfolio a little more often than monthly.
That’s a grueling, punishing schedule. Not only do the commissions add up fast, but it’s hard to have an original idea every trading day. So people cheat a little bit: they come up with a set of overlapping heuristics that can add up to a big idea. For example, a long/short investor who cares about fundamental and technical factors might have a list of rules like:
- Any company that misses guidance twice in a row has problems, and
- Any company that dials back its conference schedule has big problems, and
- Any time a company is at the peak of its guided leverage and has been buying back stock, a tiny hiccup in the business will force them to suddenly stock their buyback, and
- The default explanation for a stock underperforming on no news is that someone smart is convinced it’s a dog
That gives you plenty of material. You might have twenty stocks on your “missed guidance” list, a handful on your “suddenly shy” list, a few more on your “leverage multiple is getting dicey” roster, and when a company on all three lists drops on an up day, you’re ready for action.
A purely fundamental investor would have a different list, based on factors like which kinds of margin compression are transient and which are permanent, or the early signs of a company finally winning a competitive advantage. And an investor focused on early-stage startups as a different list of heuristics — I pay close attention to what Y Combinator says they look for, because they have a bigger sample size than anybody, and start earlier, too; if you’re starting a company, they’re better-equipped than anyone else to quickly assess your odds of success.
In practice ,pursuing idea velocity is tiring. But like any other discipline, it gets easier if you just do it relentlessly until it’s a habit. With enough time, you’ll start to get a sense of which ideas are worth pursuing early. In my investing, I’ve found that I lose money on complicated theories with no immediate catalyst. Complexity is okay, but I need some way to know I’m wrong ASAP. In writing, I have learned that nobody wants to read book reviews (or, at least, they don’t want to read mine). This is all useful information. Nassim Taleb notwithstanding, you’ll make better decisions with a bigger sample size.
Superficially, idea velocity leads to the risk that you’ll ignore your biggest and most important ideas in favor of the newest and shiniest, but I’ve found that the opposite is the case: big ideas are also the most fertile. One of Y Combinator’s big ideas is that companies led by founders who can code will have better results. That’s just one idea, and it’s one that they’ve had for a decade-plus. But every round of applications, it spawns a bunch of team-specific ideas, and, on a lag, those ideas make people rich.