Cliff Asness on Momentum, Value Investing, and Market Efficiency

Cliff Asness — co-founder and CIO of AQR Capital Management, a quantitative investor who wrote his Chicago dissertation on momentum under Eugene Fama — joins Tyler Cowen in Ep. 5 to explain why two boring rules (buy recent winners, buy what is cheap) have beaten the market for a century, why nobody can quite agree whether the reason is risk or human folly, and why he reserves the word bubble for something that genuinely cannot work out. Recorded live at George Mason University.

Key ideas

1. Momentum and value are the two great anomalies, and they pull in opposite directions. Momentum — buying the third of stocks that rose most over the past 6 to 12 months and shorting the third that fell most — earns roughly 100–125 basis points (a basis point is one-hundredth of a percentage point) of excess return a year in large US stocks, and 250–300 in small ones. Value — buying what is cheap on price-to-earnings, price-to-book or price-to-sales — works on a 3-to-10-year horizon. Crucially the two are negatively correlated: a good year for one is often a bad year for the other, so running both together is far more powerful than either alone.
2. Markets are ‘efficiently inefficient’ — almost right, not perfectly right. Even Fama, the godfather of the efficient markets hypothesis (the claim that prices already reflect all available information), tells his Chicago class that markets are ‘almost certainly not perfectly efficient’. Asness sits in the same place: prices are good enough that beating them is hard, wrong enough that disciplined systematic strategies can. The anomalies survive precisely because they live in a sweet spot — good enough to matter, painful enough that most investors cannot stick with them.
3. Why the factors work is genuinely unresolved: risk versus behaviour. A factor is a shared, repeatable driver of returns across many securities. Any persistent pattern, Asness argues, exists for one of three reasons — pure data-mining accident, a risk premium (you are paid because the strategy hurts when it hurts most), or a behavioural error someone else is making. Momentum has two leading behavioural stories that embarrassingly contradict each other: underreaction (prices move too little on news, then drift) and overreaction (people chase prices too far). Both can be true at once.
4. A bubble is something that cannot work out — not merely an expensive market. Asness keeps a high bar: a bubble exists only when no plausible, rational set of future assumptions justifies the price. By that test late-1999 equities qualified; 2015 bonds, despite yielding less than in 90 per cent of recorded history, do not — one word, Japan, supplies a scenario in which they pay off. He calls a diversified stock-and-bond portfolio merely the most expensive thing on offer, not a bubble.
5. Running a quant fund is mostly the management of discipline and fees. Strategies that work statistically — two years in three for a century, with brutal multi-year droughts — survive only because most investors flee at the wrong moment. Hence Asness’s would-be economic law: ‘There’s no investment process so good that there’s not a fee high enough that can’t make it bad.’ For ordinary savers his advice is to trade far less and, failing real conviction, just buy a low-cost index from Jack Bogle.

Content

Momentum: the ‘insane proposition’ that works

Asness opens by calling momentum ‘the rather insane proposition that you can buy a portfolio of what’s been going up for the last 6 to 12 months, sell a portfolio of what’s been going down … and you beat the market. Unfortunately for sanity, that seems to be true.’ He is careful about the word works: he means it in the ‘cowardly statistician’ sense — large t-statistics, small p-values, a positive average return over a hundred years — not that it pays every year. ‘If your car worked like this, you’d fire your mechanic.’ The strategy endures, he argues, not despite its bad streaks but because of them: enough investors cannot live through the droughts and abandon it at the bottom, which is exactly why the edge is not arbitraged away.

The effect is larger in small-caps (small-capitalisation stocks, those with low total market value) — 250–300 basis points against 100–125 in large-caps — but so is the risk, because small stocks swing more. Almost every regularity in investing, Asness notes, is stronger in small stocks; the candidate explanations include thinner analyst coverage and greater inefficiency, but the pattern is near-ubiquitous.

Why the anomalies do not disappear

Cowen presses the obvious objection: if this is real, why doesn’t everyone do it until the return vanishes? Asness’s answer is that the factors live in a sweet spot — ‘good enough to be really important if you can follow discipline, not so good enough that the world looks at it and goes, this is easy.’ He lived the extreme case during the technology bubble, when his value discipline both nearly destroyed AQR and ultimately vindicated it: a partner told him he ‘looked like Lincoln before and after the Civil War’. Value, momentum, quality and low-risk investing all share the same structure — if everyone ran them yesterday they would be gone, but the droughts are excruciating enough that few can.

Risk, behaviour, or accident

The episode’s intellectual spine is the contest over why the factors pay. Asness lays out three possibilities for any empirical regularity. The first is accident — he checked 63 things in his 1990 dissertation, so a big t-statistic is ‘a bit of a lie’; momentum, though, has survived 200 out-of-sample tests across time and asset classes, lowering that probability without ever driving it to zero. The second is behavioural: someone is making an error you exploit. The third is risk: in a rational world momentum would pay only if recent winners were genuinely riskier than losers — and ‘they don’t seem to be’.

For momentum the two behavioural stories — underreaction and overreaction — are ‘over- and underreaction’, a pairing Asness admits is awkward but insists need not be mutually exclusive (‘Remember the movie Highlander? … there could be multiple explanations’). For value he finds a risk story more plausible than for momentum, because something priced to a low long-term level might plausibly carry a hidden risk; value has, after all, suffered in the Great Depression and the global financial crisis. He parts company with Fama and French here: he thinks value is partly behavioural, whereas Fama leans toward all-risk, and he cannot see why momentum is not their model’s missing sixth factor, given how much return it adds.

The inscrutability of risk

Cowen, reading the cutting-edge risk literature — coskewness, U-shaped pricing kernels, ‘the volatility of the volatility of volatility’ — concludes that the field knows almost nothing about risk and that these are Ptolemaic epicycles. Asness partly agrees: he thinks the fifth moment and plain skewness are ‘bad stuff’, but allows that coskewness (whether a thing’s worst losses coincide with everything else crashing) at least makes sense as a real risk factor. His plain-English test for whether something is genuinely risky: ‘Does it hurt you when it hurts to be hurt?’ Momentum has a bad left tail — a Talebian black swan — but its crashes have historically struck in strong markets, not weak ones, which is why he loses less sleep over it than the geeky maths alone would suggest.

Leverage aversion and the orphaned low-risk asset

Following Fischer Black, Asness explains why low-risk stocks — low beta, low volatility, low leverage — outperform, backwards to theory. The textbook efficient frontier says everyone should hold the same best portfolio of risky assets and then lever it up or add cash to taste, rather than concentrating into a few high-return names and losing diversification. But many investors cannot or will not use leverage (borrowing to amplify a position); mutual funds are often barred by charter. So they buy concentration instead, bidding up risky assets and leaving low-risk ones ‘orphaned’ and a little too cheap. The mispricing, in his account, is less pure debt-phobia than a systematic misjudging of concentration risk against leverage risk.

Bubbles and the discipline of the word

On bubbles Asness is a self-described sceptic who nonetheless wrote ‘Bubble Logic’. Pressed to name today’s most likely bubble, he picks the diversified stock-and-bond portfolio — Shiller’s cyclically adjusted P/E puts US equities more expensive than in roughly 90 per cent of a century-plus of history, and real bond yields are similarly stretched, so the 50/50 blend is about the most expensive ever. Yet he refuses the label. A bubble, by his standard, is a price for which he ‘can’t come up with assumptions that would lead any rational investor to want to own this’ — true of late-1999 stocks, not of 2015 bonds, where Japan supplies a workout scenario. The honest conclusion is duller and more useful: not a crash in waiting, but a market priced to return about 2.5 per cent over inflation rather than the historical 5, which quietly wrecks the retirement maths of every saver still penciling in the old number.

Practical counsel and high-frequency trading

For small investors the chief error is overtrading — being, in his phrase, ‘momentum investors at a value time horizon’, bailing out of a strategy after five bad years exactly when discipline matters. He defends high-frequency trading as having made markets fairer and cheaper for the little guy, since nobody bothers to front-run a small order (‘you want to rob banks, not people’), while gently noting that cheaper trading helps only if it does not tempt people to trade more. On hedge funds he is bracingly candid: buy one of every fund and you would do better in an index, because as a group they do not hedge enough — running about 0.8 correlated with the S&P 500 — and charge too much for the privilege.

Comics, Cirque du Soleil, and the colour of the man

The conversation repeatedly wanders off the trading floor, and the page would misrepresent it to hide that. Asness is a self-confessed Marvel partisan who prefers ‘realistic’ superpowers (500 miles per hour, not the speed of light), argues Batman beats Spider-Man because ‘he cheats violently’, and names the Cirque du Soleil performers — not Gretzky, not the 1987 crash’s 20-standard-deviation move — as the most astonishing outliers he has witnessed, having once made the mistake of sharing a Las Vegas hotel gym with them. The colour is not incidental: the same taste for clean, testable propositions and for sniffing out where a story quietly cheats runs through both his comics arguments and his finance.

Related

• Cliff Asness — speaker; co-founder and CIO of AQR Capital Management, quantitative and factor investor
• Tyler Cowen — host
• What Makes a Great Investor — theme; temperament, discipline, and the psychology of sticking with a strategy through droughts
• Compounding — concept; why small annual excess returns matter over decades
• Howard Marks — investor sceptical of perfect efficiency, attentive to risk and market cycles
• Ray Dalio — practitioner of diversification and systematic, rules-based investing
• Matt Levine on Finance, Markets, and Financial Writing — related episode on markets, trading structure, and financial plumbing
• John Cochrane on Economic Puzzles and Habits of Mind — concurrent-cycle episode; asset pricing, risk premia, and efficient markets
• Bill Ackman on Activist Investing, Concentrated Bets, and Financial Battles — concurrent-cycle episode; the concentrated, fundamental counterpoint to systematic diversification