My conversation with Cliff Asness

Here is the full transcript, video, and podcast of the chat.  Cliff was great from beginning to end.  The first thirty minutes or so were an overview of “momentum” and “value” trading strategies, and to what extent they violate an efficient markets hypothesis.  Much of the rest covered:

…disagreeing with Eugene Fama, Marvel vs. DC, the inscrutability of risk, high frequency trading, the economics of Ayn Rand, bubble logic, and why never to share a gym with Cirque du Soleil.

Here is one excerpt:

COWEN: I think of you as doing a kind of metaphysics of human nature. On one side, there’s behavioral economics. They put people in the lab, one-off situations, untrained people. But here it’s repeated data, it’s over long periods of time, it’s out of sample. There’s real money on the line, and this still seems to work.

When you back out, what’s the actual vision of human nature? What’s the underlying human imperfection that allows it to be the case, that trading on momentum across say a 3 to 12 month time window, sorry, investing on momentum, will work? What’s with us as people? What’s the core human imperfection?

ASNESS: This is going to be embarrassing because we don’t have a problem of no explanation. We have a problem with too many explanations. Of course, we can observe the data. The explanations you have to fight over and argue over. I will give you the two most prominent explanations for the efficacy of momentum.

The first is called underreaction. Simple idea that comes from behavioral psychology, the phenomenon there called anchoring and adjustment. News comes out. Price moves but not all the way. People update their priors but not fully efficiently. Therefore, just observing the price move is not going to move the same amount again but there’s some statistical tendency to continue.

Take a wild guess what our second best, in my opinion, explanation for momentum’s efficacy is? It’s called overreaction. When your two best explanations are over- and underreaction, you have somewhat of an issue, I admit. Overreaction is much more of a positive feedback. It works over time because people in fact do chase prices. So if you do it somewhat systematically and before them you make some money.

One of the hard things you find out in many fields but I found out in empirical finance is those might be the right explanations but they’re not mutually exclusive.

And here is from the overrated/underrated part of the chat:

COWEN: …In science fiction, the author Robert Heinlein.

ASNESS: Early stuff, underrated. Later stuff, overrated.

COWEN: What’s your favorite?

ASNESS: That is a really — Methuselah’s Children.

COWEN: Ah, good pick.

ASNESS: I could have gone with the obvious. I’m a bit of a libertarian. I could have gone with, The Moon Is A Harsh Mistress. It’s his most famously libertarian book.

COWEN: But it doesn’t age so well.

ASNESS: No, no. I like Methuselah’s Children.

This was the funniest segment:

ASNESS: I live in Greenwich, Connecticut. In some parts of the world, if you said, “my daddy runs a hedge fund,” I’d say, “what’s a hedge fund?” In Greenwich, Connecticut, the kids say, “what kind of hedge fund is your daddy running? Is he event arbitrage? Trend following? What does dad do?”

Interesting throughout, as they are known to say…

Comments

If nothing else, he's write about Heinlein.

right, not write. Oh, how I wish this comments section had an edit tool.

It does - but only the owners use it.

Those monsters! At least they aren't editing the carbon emissions they produce.

'3 to 12 months'. Ah, there's the rub.

This sounds like "know when to get in, and when to get out."

Only for those who can't run a large number of positions. With 100 or 1000 stocks you just use an exit rule which works on average.

(I like this video, and yes behaviorial economics exists in the wild, and is not just a laboratory phenomena, or worse as some claim, artifact.)

If something works out on average over time then it must have an expected value of working out for each event. Getting in and getting out over any duration or number of events hould have the same requirements in principle: superior knowledge to the market. This is from the logical interpretation of a layman , so help me out if I'm wrong.

Sure, but you need a portfolio of sufficient size to reap expected value in practice.

IOW if I wanted to do this I'd look for a good mutual fund rather than attempt to roll my own. Tax advantage as well.

Following on from you and Brian, I really struggle to understand the 'overreaction' interpretation of momentum that Tyler and Cliff discuss.

If say Facebook's success in 2012 leads to an increase in investor confidence in 2013-2014 and corresponding eventual overreaction in share price, won't its future poor performance counteract this temporary overreaction? I suppose reweightings of the portfolio cancel this out but am not entirely happy with it.

I presume the rule is something like "own for 2 months stocks that have grown at greater than X% for 6 months." Automatic exit as momentum slows.

Gochujang (sorry MR won't let me directly reply to your comment)

Wouldn't this make the two hypotheses easily comparable?

If the liquidated 'used to have momentum but now have slowed' stocks continue to outperform then this would support underreaction, otherwise overreaction. I guess there's a large empirical discussion of this already?

Ah, from that perspective I am not sure. I think momentum traders are happy with something that works in testing, even though that is a gamble. A back tested algorithm may work great on 2005-2015 data, and then not in 2016.

It's complicated.

This is why the coming paradigm shift is going to come when we stop modeling representative agents with linear equations and start modeling representative classes with algorithms.

There's no download link on the Soundcloud page. :(

Fixed, thanks.
https://soundcloud.com/conversationswithtyler/cliff-asness

Note you can also subscribe via your favorite podcast app.

-Jeff

"When you back out, what’s the actual vision of human nature? What’s the underlying human imperfection that allows it to be the case, that trading on momentum across say a 3 to 12 month time window, sorry, investing on momentum, will work? What’s with us as people? What’s the core human imperfection?"

The drivers of price expectations change over time. Sometimes, visions of future marginal cost drive spot, as was the case until June for oil futures. Sometimes, spot drives futures, as is the case today. Sometimes it's inventory. There's not a single driver. Sometimes expectations are firmly held within a narrow band; sometimes within a broad band; sometimes we don't know what to expect. Sometimes we change models. Stuff happens, which is apparent to market insiders, but which non-specialized economists entirely fail to appreciate.

Here's my assessment of one particular BoE model, at CNBC. http://www.cnbc.com/2015/11/13/why-oil-could-rally-big-in-2016-commentary.html

Good interview, needed more about herding.

Price used to be the information, now news is the information.

The robots are winning.

Oh, you're just tempting me to participate, aren't you ?

The Moon is a Harsh Mistress ages just fine. I re-read it a few years ago and cried when Mike died.
Also, the plot line is surprisingly easy to interpret in either a left-wing or right-wing way. And you get all those fun notes about airlocking people for not paying for their air supply, line marriages, for-profit judges, and other alternative social structures.

I especially like the injection of Russian vernacular terms like 'stilyagi' .

I read it for the first time 2-3 years ago and didn't notice anything dated about it at all, really.

I haven't read all of Heinlein's books, and none of them recently, but I recall finding Stranger in a Strange Land to be very powerful, with most of his other works seeming juvenile.

I find the opposite: his juvenilia (in both senses) is his best work. But I disagree with Asness on TMIAHM. Aside from the big math error regarding the lunar catapult (and I don't think that's what he's referencing), it's great today. The idea of Mike and a central role for technology in a revolt is increasingly plausible.

It was Cowen who said it hadn't aged well. Not sure if Asness really agreed with that.

Yup, you're right. Thanks.

I think the awesomeness of Mike is that it portrays him as sort of emotionally childish, and lonely, which is maybe realistic for an AI who is the only one of his kind. Kind of like an autistic human - he can do mathematical calculations in a split second but can't understand humor. He's a surprisingly sympathetic character without having a body.
You sort of feel (and the characters sort of feel), that they exploited him in their need for help with the revolution. To him it was just a big joke. A game. But to them, he was their only way to win. It's a bit of a moral dilemma. You have that scene near the end where he asks for some time to analyze jokes , which is a big reminder that he's sort of like a kid with superpowers that the adults are manipulating for their own ends, as noble as the ends might be.

Grok > TANSTAAFL

Both terms are iconic. However, Stranger in a Strange Land is very much a product of the 1960's, whereas I don't find that The Moon is a Harsh Mistress feels dated. There are certainly direct references to technology and politics that are probably dated but the underlying theme and flow of historical events still seems plausible.

Isn't Mike a trinary vacuum tube computer? That particular detail is wildly dated.

Still a fine portrayal of an AI and a great story.

Heinlein's early work, say 1939-1947, revolutionized American science fiction by clearly being consistently better than what had come before. On the other hand, he was still learning to be a professional writer.

His middle period of, say, 1948-1966 was hugely productive, especially the novels for boys. Kevin Drum suggests "Starman Jones" as Heinlein's peak for juvenile novels.

Heinlein was becoming less consistent toward the end, with more homers and strikeouts in the 1960s, but his ambitious 1966 novel The Moon Is a Harsh Mistress shows him at age 59 at the peak of his powers.

Then Heinlein's health, never good, was terrible for three years during which he couldn't write. When he came out of the fog finally, he found himself rich and famous due to hippies adopting his overblown 1961 book "Stranger in a Strange Land" as a manifesto. (As testament to his breadth, Heinlein's other two cult novels appealed to two different cults: Starship Troopers to militarists and The Moon to libertarians.)

Freed up from social restraints on his kinks, and with his literary powers in decline, Heinlein's later books were bestsellers, but undermined his reputation.

A respectable summary, except "Have Space-Suit, Will Travel" is the peak of the juveniles, with "Citizen of the Galaxy" a close second. And the quality drop-off after TMIAHM is even worse than you state; just about all of those books are terrible from start to finish.

That Have Space Suit Will Travel (1958) was the peak of Heinlein's boys novels was the opinion of a book of literary criticism about Heinlein I read back as a teenager.

I could definitely make the case for Spacesuit and Citizen of the Galaxy as the peaks, but it depends upon how much you like those two books' radical shifts in plot, setting, and tone. Citizen of the Galaxy, for example, has four wildly disparate sections. It starts off as riff on Kipling's "Kim" in an oriental slave kingdom, then suddenly shifts to a Finnish merchant culture with complicated marriage rules that anthropologist Margaret Mader (i.e., Margaret Mead) pops up explain, and then the hero is in the Space Marines (in a brief warm-up for Starship Troopers), and then it becomes a business story about a proxy battle for control of a corporation.

The whiplash plot of Citizen of the Galaxy is a head trip and a showcase for Heinlein's polymathic ability to make different worlds seem plausible. On the other hand, maybe Heinlein just had four separate stories rattling around in his head, none of which he could figure out how to flesh out into a full novel length? Or more charitably, Heinlein was at the peak of his powers around age 50 and had so many good ideas that he'd burn up multiple ones per his annual novel?

In contrast, his earlier Starman Jones is less flashy than the later books. But it has a humbler, more integrated plot where each development is prefigured by something earlier, leading up to Heinlein's best action climax, which is then topped by a boys' wish fulfillment ending that is surprising but inevitable within the logic of the story.

That said, there has to be a reason why Starman Jones tends to be overlooked. Perhaps it's because the book starts off more ho-hum than many of Heinlein's other novels -- a poor boy runs off to sea (or, in this case, space), and then builds to an unexpectedly impressive climax. (Bad relativity physics is a problem with Starman Jones, now that I think about it.)

A lot of Heinlein's other books are memorable for starting off with some sensational idea -- alien pods hijack human beings in The Puppet Masters, for instance, or the powered armor suits in Starship Troopers -- and then struggle to maintain momentum. Stranger in a Strange Land, for example, opens with as perfect a title as any in American literature and an exciting first chapter, but then rambles on too long. Most of his other books aren't as self-indulgent as Stranger, but in general they are more driven by a brilliant first chapter than last chapter.

For example, Time for the Stars is kind of a rewrite of Starman Jones intended to fix the bad Einsteinian physics in the earlier book by building a novel around the Twin Paradox of relativity theory. It gets off to a great start with the idea that some tiny fraction of identical twins don't just have private twin languages, they communicate via telepathy and telepathy isn't constrained by the speed of light, but then Heinlein runs out of gas, and just reuses the action climax from Starman Jones.

"And the quality drop-off after TMIAHM is even worse than you state; just about all of those books are terrible from start to finish."

I thought Friday was excellent and it gets a high rating (3.81) on Good Reads.

If you want to read a college textbook on behavioral finance, read: Ackert and Deaves, Behavioral Finance: Psychology, Decision-Making and Markets. It is very good, summarizes the literature, has applied examples, and answers the questions that Cowen asked above.

Information economics and psychology. You need both a signal and a recipient of the signal who correctly interprets the signal.

By the way, the studies confirming behavioral economics hypothesis do not derive simply from tests with college students. The textbook above amply describes and lists all the finance journal articles which test behavioral econ principals in actual markets using market data.

I know it must come hard to those who genuflect to the invisible hand to believe that the hand itself is governed by a not so great cognitive engine.

Then it must be especially hard for those who desire a command economy to come to the realization that the committee will be staffed by orders of magnitudes less of those not so great cognitive engines. Plus, some or most of those not so great cognitive engines will be corrupt, and all of them poorly incentivised.

Congratulations on the first straw man of the day, with "a command economy."

Thomas, Perhaps you have a belief that upon seeing human error humans cannot undertake any effort to correct for mistakes. So, I guess you would have let the financial system melt in the belief of self-correcting mechanism. But, if experiences are any guide, even those who most vociferously opposed intervention passed TARP, as request by Bush and Paulsen.

Behavioral econ findings really do necessitate paternalism, and it's time we grow up and admit that. "But government agents suffer cognitive biases too!" Yea sure, but not as many, and at least they're aware of their own limitations. It's the rabble who can't control their passions.

Kahneman says no. No one is capable of appreciating their biases and acting to control them.

True Stan, but my experience is that with time "book learning" soaks in, and confidence levels adjust.

The transcript was probably the best thing I've read all month.

It sounds to me like both his 'under-reaction' and 'over-reaction' could derive from the same mechanism of lags in updating priors. At the beginning of a bubble cycle, people don't fully gauge how under-priced an asset class (under-reaction). By the time most peoples' priors have caught up, they're then too slow in figuring out when the thing is fairly valued and then overvalued (over-reaction). 'Chasing prices' is just a delay in adjusting priors in the opposite direction.

AQR is, by and large, a marketing firm. They have propagated this image of themselves as this "super smart quant shop", which they aren't. They run massive, conservative portfolios for large institutions. They are pretty transparent about the strategies their funds utilize and insist they are simple, well-known, and vetted by academics. Institutional investors go to them because it's a brand name and fees are lower than competitors'. No one has accused them of outperforming since the days they were an internal hedge fund at Goldman Sachs. They are basically Prudential, but with a quant veneer.

I enjoy reading Asness for his bluntness. He's a successful businessman, but his reputation as a great investor is undeserved, in my opinion.

They do sell factor models like they're some novel thing and not something everybody does. But you read above and you realize there are a lot of people who've never heard of autocorrelation.

Some of these techniques are old, but their reception was different when the EMH was Strong.

I thought Tyler's question about "how many inefficiencies remain" was a little sad, as if we can just drive out a few wrinkles and get back to a Strong world.

Sorry, this is a post Mandelbrot, post Kahneman, post Thaler, post Shiller world.

There were trend following hedge funds at least as far back as the 1970s. The EMH was never a serious component of investment management or trading.

That is a link to a gated article. I can see that it says "The efficient-markets hypothesis has underpinned many of the financial industry’s models for years."

But I would dispute this. Did they conduct a poll?

The narrative of "quant models failed" is banal and usually wrong or incomplete.

See also Fox's Myth of the Rational Market, or any review of that book.

You semi-accuse them of over-marketing and then, in same post, admit they're transparent (honest) about what they do.

Weak.

False dichotomy. They are transparent about the strategies and oversell the 'science'.

Not false, your point and defense of it are ridiculous.

He may or may not be a great investor but by your description he's honest.

You seem bitter and thus say silly and inconsistent things.

Well, what do you know? Looks like I may have been right again with my "Conceivably":

http://media.mhfi.com/documents/2015-Finlit_paper_17_F3_SINGLES.pdf

http://marginalrevolution.com/marginalrevolution/2015/11/how-honesty-varies-across-countries-politically-incorrect-paper-of-the-day.html#comment-158784739

The first link is to a paper about financial literacy, and the second goes to a comment by "Harding" that says "Not Gentile Russians." apparently in response to a claim that "Brighten Beach [sic] is all Russians."

What in the world do they have to do with each other? Is this your impostor / alternate personality again?

TC is an amazing interviewer.

ASNESS: I live in Greenwich, Connecticut. In some parts of the world, if you said, “my daddy runs a hedge fund,” I’d say, “what’s a hedge fund?” In Greenwich, Connecticut, the kids say, “what kind of hedge fund is your daddy running? Is he event arbitrage? Trend following? What does dad do?”

Right, it's not a very clear, concise, and universal term. Usurer or moneylender is better and can be understood everywhere, even by kids.

Hedge funds don't generally lend money. Even when they buy debt it usually isn't at origination.

Do Jesus freaks object to equity too?

They're effectively engaged in moneylending. They don't create and operate industry, they lend money to industry and hope to earn interest on that money. Various legalistic arrangements such as equity and secondary markets for debt just obfuscate this fundamental activity of moneylending. Islamic law does the same thing; it bans usury technically, but effectively allows it through the equity arrangements it sanctions.

This is why usurer or moneylender is the better term. It is clear and can be understood even by kids, while "hedge fund" is obfuscatory, even for many adults.

Please explain how a market neutral equities hedge fund is money lending. Or volatility arbitrage. Or FX. Or Index arb.

Perhaps your argument works for long-only or long-biased funds. But then not every money lender is a usurer, which you seem to imply.

Usury means moneylending. The use of usury to just mean exorbitant interest rates is part of the obfuscatory language I mentioned above.

Yeah, tons of kids know what a 'usurer' is.

Also, grow up. Capitalism is not evil.

I did say "usurer or moneylender". Kids can easily grasp what a moneylender is.

Where did I say that capitalism is evil?

What exactly is wrong with lending money?

Where did I say that there's anything wrong with lending money?

Is there a reliable third party source for evaluating Asness's long term performance across all his funds net of fees? I'd be very interested to see this.

"ASNESS: Early stuff, underrated. Later stuff, overrated."

I'm not sure his later stuff is overrated. I was under the impression that the majority of readers consider Heinlein's later work to be weaker. Certainly the plot lines lean towards the fantasy side of Science fiction in the later period.

This seems more accurate than what's said in the interview.

I think some of the confusion stems around what qualifies as later works. I'd say the 1980+ works were probably reasonably rated (not as good as earlier works but good moments not his best). The works between Stranger in a Strange Land and Time Enough for Love are probably overrated mostly because they're considered his best work and most iconic work. His earlier work which contains a lot of his juvenile targeted work is good but not considered as good as the middle period, probably because a lot of it was targeted at a younger audience. It's all an opinion and heavily weighted by preference though.

"The works between Stranger in a Strange Land and Time Enough for Love are probably overrated mostly because they’re considered his best work and most iconic work. "

Hmmm, interesting thought. But he won Hugo's for "Stranger in a Strange Land" and for "The Moon Is a Harsh Mistress" and was nominated for a Hugo for "Glory Road" and "Time Enough for Love" during that period. I'm not convinced that's really a case of being overrated. I'm not sure how you can be considered overrated with that track record. At least not on any kind of objective scale.

There is a mistake in the transcript, "verses" should be "versus".

The transcript is accurate. I misspelled it in my mind while saying it.

Despite this, transcript has been corrected. Thanks.

-Jeff

I call bullshit on Asness. If momentum trading works, why is so hard to quantify? Consider for a moment how many math PHDs have tried and failed to find a successful momentum trading strategy.

My theory for why momentum trading works is that it's a good, simple marketing strategy to use on gullible investors.

Tons of academic and practitioner papers testing the same very basic simple momentum strategies in tons of places and finding it works rather ubiquitously.

But saying the same wrong thing, rudely, in multiple separate posts really adds to your credibility.

Asness is a salesman talking past the sale. Nevermind that his product doesn't work.

Test any momentum strategy you can think of and the expected return will be zero minus the vig. Try it! There's data everywhere.

Momentum trading makes some people rich for the same reason the lottery does: despite horrible odds, the Law of Large Numbers won't catch you when a market supermove happens. That doesn't make it a wise strategy.

I enjoyed the video.

I'm glad Tyler briefly had Cliff comment on the size effect (what works in big stocks tends to work better in small stocks) and to also briefly comment on whether momentum still works (since early 2000s it seems momentum effect has been diminished).

On the question of "are there other undiscovered factors?" I guess I wonder if there is undiscovered data that could lead to new exploitable anomalies? For example, in the p&c insurance the development of a new data source, credit scores, revolutionized pricing/risk determination (and are extremely powerful predictors). Similarly, telematics - directly monitoring how you drive via technology - is transforming the industry now.

In investments, I've seen studies indicating that short interest in stocks is a powerful predictor of excess returns. (shorts are right and earn excess return). Same effect with earnings revisions. I'd expect further anomalies to come from areas where data is difficult to come by and analyze.

Those aren't really anomalies in the traditional sense. If I spend money to create new information, it wouldn't challenge anyone's efficient market beliefs if that information was tradable until it's priced in.

The factors being discussed are weird because they are easily observable and yet seem to provide returns (brushing some issues under the rug). What's odd is that they don't seem like "risks" that you should be compensated for.

The Moon is a Harsh Mistress is not dated at all. In fact, it's becoming more prescient by the year:

First, the discovery of hollow lunar 'lava tubes' and entrance skylights raise the probability that one day we will be inhabiting the moon underground in 'warrens' just as Heinlein predicted.

Second, Mike was not a designed AI, but a complex system that 'woke up' after the number of nodes and interconnections added to him reached a threshold. This is increasingly likely to be the path towards the first true AI. We have no idea how to design a conscious intelligence, but complex adaptive systems display 'intelligence' without any design whatsoever.

Third, the idea of Russians, Chinese and Europeans living together in a mixed society is not nearly as far-fetched as it might have seemed during the peak of the cold war when Heinlein wrote the book.

Where Heinlein consistently went wrong in his speculation was that he thoroughly bought into the Malthusian argument that the Earth would soon be grossly over-populated and run out of food, and that expansion into space would be largely about solving the food/resource problem. Thus you have a moon colony shipping grain to Earth, characters in other books emigrating to Ganymede to avoid food shortages, population controls in books like "Time for the Stars", mass Chinese emigration due to overpopulation in "Tunnel in the Sky", and so on. It's easily his largest failed prediction.

The Moon is a Harsh Mistress is one of the best books ever written depicting cultural adaptation to a Libertarian legal framework. Heinlein came up with unique solutions for social safety nets, legal systems, the kind of culture that might arise, etc.

So whenever new information is received, the market will either respond accurately, or it will overreact, or it will underreact. Solid insight. BTW, I've heard that uncomfortable temperatures can be caused by either too much heat or not enough of it, can anyone confirm whether or not this is accurate?

Well, there are those who would argue that uncomfortable temperatures don't exist - that the temperature is always comfortable and if we think it feels too hot or too cold we're mistaken.

Truly when someone doesn't be aware of then its up to other
users that they will help, so here it takes place.

Comments for this post are closed