This Is Why We Can’t Have Nice Things

Here’s the most powerful animal on our farm — a deer tick, well embedded, gorging itself on human blood.

Of course, it’s not the tick itself that is so powerful, but the Lyme disease it can transmit, caused by a spirochete bacterium named Borrellia burgdorferi, pictured below. As you probably know, Lyme is a terrible disease, difficult to diagnose and difficult to treat once established. Once transferred via tick saliva, the wormlike burgdorferi bacteria quickly spread throughout the human host, particularly into joints, the heart and the brain. From there, the bacteria cause symptoms including intense arthritic pain, palsy and paralysis, loss of memory and extreme fatigue. Our immune systems typically fail to create the necessary antibodies to fight the infection, due to both the antibody-suppressing qualities of tick saliva and the antibody-hiding qualities of burgdorferi. And in a process still not entirely clear but suspected to be connected to an autoimmune failure spurred by burgdorferi, these crippling symptoms can persist for years, even after all of the bacteria have been killed through aggressive therapy. This is a potent parasite, and if you’ve lived for any length of time in Connecticut, you surely know at least one family that has been hit hard by a tick-borne disease.

But Lyme disease isn’t the reason that the deer tick is the most powerful animal on our farm.

No, it’s not Lyme disease. It’s Lyme disease!.

Huh?

Lyme disease is a physiological ailment caused by bacteria and injected into your body by ticks.

Lyme disease! is a mental ailment caused by words and injected into your mind by humans.

Lyme disease! is the IDEA of Lyme disease. It’s the mental construction of a world where Lyme disease and the bloodthirsty deer ticks and the grotesque burgdorferi bacteria are EVERYWHERE, an omnipresent threat to you and your children. Lyme disease! is an infectious meme, in the true and powerful meaning of the word, not the ha-ha cartoonized meaning that we see every day on social media. Memes are self-sustaining ideas that live in the human brain. They are as alive as any bacteria or virus, and they infect every aspect of our social lives.

What do I mean? I mean that families infected with the meme of Lyme disease! don’t allow their children to hike or play in our woods. I mean that the meme-infected next door neighbors have spent hundreds of thousands of dollars to wall off — literally wall off, including river barriers — their 20+ acres from all animal life that can’t fly. I mean that we have been sued — literally sued — by meme-infected parents who thought their child might have “caught rabies” (not Lyme, but close enough) from petting our dog some hours after the dog found a dead raccoon. No, I’m not making this up. And just wait. I can promise you that I’ll get “well, actually” emails from meme-infected readers of this note.

What we have in the wilds of Fairfield County, Connecticut (and I suspect in every exurb in the country), is a large population of wealthy people who for a variety of reasons want to live closer to nature, but are scared to death of nature! … the memes that infect our brains about the risky parts of nature, memes like Lyme disease! or rabies! or coyotes!. It’s not that these aren’t actual dangers. Lyme disease is a real thing and a real risk. So is rabies. So are coyotes. But our social lives aren’t governed by the actual risks of the real-life things. They’re governed by the memes. They’re governed by the metagames.

I wrote about this from the perspective of the real-life thing in Too Clever By Half, where I described why the coyotes in our woods always lose the metagame. They win every direct interaction with us tame humans because they’re smarter and braver than we are. But they lose in the larger metagame because the townfolk have access to armed animal control officers who are required — begrudgingly and remorsefully — to kill the real-life coyotes when the coyote! meme infects enough civilians. The lesson for all coyotes, four-legged and two-legged alike, is pretty simple. Don’t trigger the townfolk. Yes, you’re smarter and braver than they are. You can win the immediate game. But you will always lose the metagame if you’re too visible in your “winning”. Always.

There’s a larger perspective here, too, and a larger lesson. It’s the perspective of all of us. The meme-infected. Like me. Like you.

In my eight years on the farm, where I spend a lot of time clearing brush and cavorting around tick-rich environments, I’ve been treated for Lyme disease twice. Both times I had an attached tick, so I pulled it off and went to the doctor. Both times the doctor didn’t even bother testing me for Lyme, but just started me on antibiotics, because you can knock Lyme out if you treat it early enough. Maybe I had Lyme and maybe I didn’t. We’ll never know.

In the immortal words of Remo Gaggi when he and his fellow mob bosses of Casino decided to whack a loyal lieutenant who had the misfortune to have a slight opportunity to rat them out, “Why take a chance?”

Look, I get it. There is zero upside for the doctor to make a measured calculation of the actual risk of Lyme disease. There’s zero upside because the doctor knows that as bad and prevalent as Lyme disease might be, Lyme disease! is even worse and more prevalent. THAT’S the disease that the doctor was treating when she prescribed the antibiotics — Lyme disease!, not Lyme disease. Because she knew that even if the correct and rational treatment for Lyme disease was to do nothing or carry out some more tests, the absolutely correct and rational treatment for Lyme disease! was an immediate course of broad-spectrum antibiotics. It’s a no-brainer. There’s no doctor in the world who can stay in business for long if she doesn’t recognize the memes that infect her patients, who doesn’t nod understandingly and overprescribe when a mother wrings her hands over her child’s “exposure” to this dread disease! or that dread disease!, regardless of the disease truth. This is the metagame of modern health care.

Ditto with financial advisors.

You’re not going to stay in business for long if you don’t recognize the memes that infect your clients, memes like the fundamentals are sound! and we’re cautiously optimistic! and stocks for the long haul! and value! and bet on America!, all of which are most effectively treated with a profound over-allocation to U.S. equities under any and all circumstances. It’s not that these aren’t true and real things. They are absolutely true and real, just like Lyme disease is absolutely a true and real thing. But the true and real thing isn’t what drives our behavior. It’s the meme! that does that. There is zero upside for a financial advisor to make a measured calculation of the actual portfolio risk of a client’s under-exposure to U.S. large-cap stocks, because the actual portfolio risk isn’t the driving risk! that the client is infected with. So all financial advisors overprescribe U.S. large-cap stocks for their clients. We all know it’s true. We don’t like it, just like no doctor likes overprescribing antibiotics, but we do it anyway.

As Hyman Roth said to Michael Corleone, this is the business we have chosen. This is the metagame of modern investment management.

But like I say, it’s bigger than that.

Five years ago, I started Epsilon Theory to talk about capital markets, and it will always be a core part of what makes me tick and what I choose to write about. But as important as it is to recognize and call out the memes that infect our markets, it’s even more important to recognize and call out the memes that infect our politics. And the human ticks who spread them. That effort starts today.

A preliminary observation before we get to the stuff that will annoy a lot of readers … everyone I’ve ever known, including me, comes at the question of memes and their influence on our decision-making from a very simple starting point — yes, they’re effective on other people, but not on me. I am smart enough and independent enough to be effectively immune from a meme infection.

No, you’re not.

If you get nothing else from Epsilon Theory, get this: we are ALL hard-wired — literally hard-wired through millions of years of neurological evolution — to respond positively to effective meme introduction. We are ALL programmed — literally programmed through tens of thousands of years of cultural evolution — to respond positively to effective meme introduction. It’s no exaggeration to say that our biological and cultural symbiosis with memes defines the modern human species. This is a feature, not a bug.

Eusocial animals (the “pure” form of what it means to be a social animal) swim in an ocean of constant intra-species communications. It’s why these species — the ant, the termite, the bee, and the human — are the most successful multicellular animal species on the planet. Eusocial animals have the ability to store, retrieve and broadcast information (yes, eusocial insects communally “remember” incredibly complex informational structures) in a way that non-eusocial animals simply can’t, and it allows the eusocial animal not only to survive its environment, but to master its environment. Any environment. Humans are essentially giant termites with opposable thumbs and fire, and that combination is particularly unstoppable. But it’s the termite-ness … it’s the swimming in an ocean of constant intra-species communication … that’s the most important of these qualities.

The downside, of course, is that we can no more resist the language of Hero! and Wizard! and Enemy! than an ant can resist the pheromones of its queen. These are the Old Stories and the New Stories alike. Memes are our greatest strength as a species. And our greatest weakness as individuals.

Memes are the stuff that Narratives are made of.

Fortunately, the human animal is a self-aware animal. For the most part. Kinda sorta. At least we have the ability to perceive our infection. Through a glass darkly, as the Old Stories would put it.

Self-awareness doesn’t mean some magical immunity to being influenced and played by the other players. On the contrary, if you think that you are immune to all this, well that’s prima facie evidence that you are not self-aware at all. That’s prima facie evidence that you are, in fact, the sucker in this big poker game of citizenship. No, self-awareness means a recognition that you ARE being influenced and played by the other players, so that you can use that knowledge of HOW you are being influenced and played to maintain YOUR personal liberty of mind and play YOUR best game.

We can’t change our nature as meme-susceptible human animals, but we can absolutely become better human animals, both instrumentally as game players and ultimately as citizens. We can absolutely NOT be suckers. We can absolutely NOT lose our liberty of mind — which is the only liberty that really matters — to the incessant meme-generation of the Nudging State and the Nudging Oligarchy.

So how do we avoid being the sucker within this largely invisible poker game of memes and narratives that we are immersed in from birth, a poker game that we are biologically and culturally evolved to play rather poorly?

First and most importantly, we can simply recognize that there is a logic and a process to meme introduction and contagion in the human animal. Here in Epsilon Theory I like to focus on one powerful contagion vector — the Common Knowledge Game — but there are many others. Like all of the invisible forces that drive our lives, once you start looking for embedded memes and the logic that drives them, you will see them EVERYWHERE.

Second, we can use the new tools of AI and Natural Language Processing (NLP) to visualize the meme introduction and contagion process. This is what I’ve called the Narrative Machine, and it’s as useful for understanding the behavioral drivers of politics as it is for understanding the behavioral drivers of markets. Why is visualization so important? Because it taps directly into the way our brains are hard-wired. Seeing is, in fact, believing, and by showing you visual evidence of political meme introduction and contagion, you will be far more likely to accept the worth of my broader argument. It’s why data visualization is such an important topic, and it’s why Ed Tufte is a personal hero of mine. [Optical Illusion / Optical Truth]

More generally, NLP can help visualize what I described as the “cartoonification” of political candidates and political issues. From The Icarus Moment:

Cartoons aren’t just created to mobilize positive sentiment and supportive social behaviors (although that’s pretty much all we see in capital markets, because it’s a positive-sum game, not zero-sum like politics). The negative cartoonification of Hillary Clinton was both the most vicious and the most effective gambit in the last 100 years of American politics. To be sure, The Clintons™ brought soooo much of this on themselves. If there’s ever been a political candidate more ripe to be transformed into a negative cartoon than Hillary Clinton, I am unaware of who that might be. But where Donald Trump embraces and actively creates his obvious cartoonishness, Hillary Clinton had her cartoon imposed on her unwillingly, to disastrous result. Today’s key to political and economic success is controlling your own cartoon. Yes, this is why Trump won.

So what does the Narrative Machine show us about meme construction and contagion in the last U.S. presidential election campaign?

Here’s an NLP analysis of 124,000 articles on Hillary Clinton published in non-paywalled top-tier U.S. media over the year prior to the presidential election — where linguistic similarities create clusters of articles with similar meaning, essentially a linguistic “gravity model” (for methodology background on all this, see The Narrative Machine).

Source: Quid, Inc. For illustrative purposes only. Past performance is no guarantee of future results. Quid, Inc. is not an affiliate of Salient. Software used under license.

It’s a dense narrative map because of the quantity of articles, but we can simplify the analysis by re-coloring the clusters by sentiment, and then isolating the negative attack memes.

Here’s the sentiment map. 20% of the articles are negative, including lots of negative articles in non-attack memes like Primaries! and Supreme Court!, 45% are neutral, and 34% are positive. Hold that thought.

Source: Quid, Inc. For illustrative purposes only. Past performance is no guarantee of future results. Quid, Inc. is not an affiliate of Salient. Software used under license.

And here’s the re-spun narrative map after isolating the negative attack memes:

Source: Quid, Inc. For illustrative purposes only. Past performance is no guarantee of future results. Quid, Inc. is not an affiliate of Salient. Software used under license.

Beyond the frequency of articles associated with this or that meme (Emails! clearly dominating on that dimension, with 42% of all negative meme articles published), there are three critical dimensions in an interpretation of a narrative network: geometry, time dynamics, and affect. The map above gives us our geometry, and I’ve found a scatterplot (below) to be the best visual representation of time dynamics and affect. Between the two graphs a fascinating meme contagion pattern emerges.

Source: Quid, Inc. For illustrative purposes only. Past performance is no guarantee of future results. Quid, Inc. is not an affiliate of Salient. Software used under license.

First, geometry. There’s no real information in the north/east/south/west orientation of a narrative map, but there is significant information in distance, center/periphery orientation, and inter-cluster links, all of which can be understood with a simple gravity metaphor. The greater the distance between meme clusters, the less similarity in vocabulary and grammar employed in the individual articles that comprise the clusters (less gravitational attraction between the clusters). The more central the meme cluster to the overall network, the more coherence it provides to the overall narrative (a gravitational pull exerted in all directions). The more inter-cluster links (the long strands that connect one cluster to another), the more articles that explicitly have one foot in each camp, visualizing the gravitational tethers.

What we have in the Hillary meme network is a clear outlier in the Benghazi! cluster, as well as a clear super-cluster comprised of Wall Street!, Clinton Foundation! and Emails!, with Wall Street! and Clinton Foundation! being more central to the overall Hillary cartoon-ification, despite the far greater frequency of Emails! articles. The way to think about the peripheral nature of Benghazi! and Emails!, I think, is that these memes didn’t “take” in the same immediate and easy way that Wall Street! and Clinton Foundation! “took”. To use the deer tick metaphor, whatever ticks were trying to inject the Benghazi! meme never really got fully embedded in the body politic, while the Wall Street! and Clinton Foundation! ticks gorged easily to their little tick-hearts’ content. What’s really interesting, though, is the Emails! meme. Whatever the Emails! delivery ticks lacked in embeddedness, they more than made up for it in effort.

That’s my takeaway from the scatterplot representation of time dynamics and affect, where the green dots (sub-clusters of Emails! articles) are high in affect (the x-axis, representing the strength of “emotion” in article word choice, mostly negative, but some positive, too) and almost constant in duration (the y-axis, representing time). That second phenomenon — the degree to which there was an almost constant drumbeat of Emails! articles over the course of the campaign — is particularly rare and unusual.

Here’s what a typical meme infection looks like, as shown in a histogram for Clinton Foundation!. The meme percolates in the background for a while, explodes in an outbreak of virulence and Sunday talk show segments, and then dies back down again just as quickly.

Source: Quid, Inc. For illustrative purposes only. Past performance is no guarantee of future results. Quid, Inc. is not an affiliate of Salient. Software used under license.

Source: Quid, Inc. For illustrative purposes only. Past performance is no guarantee of future results. Quid, Inc. is not an affiliate of Salient. Software used under license.

Emails!, on the other hand, had multiple outbreaks and never died down. Sure, it got crowded out by other memes here and there, as the sum-to-100% histogram above shows, but I can’t tell you how unusual it is that a meme like Emails! persisted in such a virulent form for an entire year.

In the overall narrative network, not just the negative meme creation stuff, but the entire universe of media coverage, 6% of EVERYTHING written about Hillary Clinton for a YEAR was about Emails!.

This is nuts. It’s not an accident.

And please, I’m begging you, don’t send me a “Well, actually” note yelling at me about how Hillary Clinton’s handling of her email servers was a ridiculous, mendacious and probably illegal thing, that it was, in fact, a big deal.

I AGREE.

The Emails issue was a real and true thing, just like Lyme Disease is a real and true thing.

But you are the sucker at the poker table if you don’t recognize the incommensurability between the real and true Emails issue and the Emails! meme, if you don’t recognize how YOUR political behavior and YOUR liberty of mind was impacted by Emails! in a way that Emails could never achieve.

Mine certainly was. I was so righteously aggrieved by Emails!, thinking all along it was Emails. Emails! angered me for months. It made a difference to me. And then I did this analysis and saw how the meme was constructed and promoted. I saw how I had been played. If I knew then what I know now, would it have made a difference in my NeverTrump + NeverHillary position? No. But I’m not going to let it happen again. I’m going to do everything I can to protect my liberty of mind.

And in the spirit of in-for-a-penny-in-for-a-pound, here’s another sure-fire aggravating observation on the meme construction process around the most recent U.S. presidential election, this time from the Trump narrative map.

Source: Quid, Inc. For illustrative purposes only. Past performance is no guarantee of future results. Quid, Inc. is not an affiliate of Salient. Software used under license.

Above are all the different meme clusters associated with Trump for the year prior to the election, all from top-tier U.S. media, colored by sentiment. Lots of incendiary memes in there, right? But here’s the thing. First, the overall narrative network is comprised of 167,000 articles, about 35% more coverage than Clinton received. Second, of that coverage, only 15% of the articles are negative, with 50% neutral and 34% positive. Third, of the negative memes, none had a persistence pattern like Emails!. They all spiked and faded like Clinton Foundation!.

Trump got significantly more coverage than Clinton in major media outlets.

Trump got significantly more positive coverage than Clinton in major media outlets.

Trump suffered from no infectious meme like Clinton suffered from Emails! in major media outlets.

I’m not saying whether all this is good or bad. I’m just saying that it IS. And what it isn’t.

This isn’t a Russia thing.

This isn’t a Facebook thing.

This is a mainstream media thing. A mainstream media thing comprised of people who, for the most part, would rather rip out one of their own fingernails with red-hot pincers than help Trump, but who, driven by the systemic pressures of their business and its utter reliance on Fiat News, did just that.

So what do we do about this?

Well … nothing. Or at least nothing to “fix” mainstream media directly. I say that because I don’t think it CAN be fixed, just like I don’t think mainstream political parties CAN be fixed. They can’t be fixed because both of these social institutions — media and political parties — are not broken from an  internal perspective of institutional profits and personal agency. On the contrary, they’re thriving.

Media and political parties are institutionalized ticks, and the tick business has never been better.

Look again at that Trump narrative map. Look at all the obvious negative attack memes — SNL, Late Night TV, Meryl Streep, JK Rowling, KKK, Megyn Kelly, Russia, Funny or Die, Judge Gonzalo — they’re not red! I mean, there’s some red in there, particularly for Megyn Kelly because it linked into the highly negative (and politically effective) Sexism meme, but for the most part the sentiment of the articles themselves is neutral to positive, even though they’re part of an obviously negative meme. How can this be? Sure, Fox and its ilk are going to be neutral to positive on all this, but they’re a small fraction of the universe here. Why is the “Failing New York Times” using neutral language to talk about Trump and the Ku Klux Klan? Why would they use language like “Trump’s ‘very fine people’ remark was taken by many as an endorsement of the KKK and other white supremacist groups”? There’s nothing inherently negative in those words. Why aren’t they hitting Trump harder?

Because metagame. Because the long-term evolutionary stable strategy for a tick species is not to maximize blood-sucking and egg-laying, but to balance resource gathering and reproductive success against the minimal requirements to keep the host species alive.

There’s that word. Balance. Like in “balanced” media coverage that of course is not balanced at all, but observes the forms of the free and fair press! meme that thoroughly infects all of us, not least the media participants themselves. Like in the balance of an equilibrium.

The current state of intense political fragmentation and conflict is a very stable evolutionary equilibrium for all of these professional meme-generation entities. Ratings are up. Subscribers are up. Engagement and participation are up. The host species is showing signs of exhaustion and stress, but nothing potentially fatal. If Trump did not exist, professional meme-generation entities would have to invent him.

So they did.

And once the miracle of Trump does exist, professional meme-generation entities must be careful not to kill him.

So they won’t.

Successful ticks have the same secret as successful coyotes — they play the metagame really well — and there is no more effective metagame player than giant corporate media.

They’ve been manipulating memes for a really long time. It works really well for them.

It just doesn’t work very well for us.

We are infested by ticks.

And yes, I understand that this is a horrifying photograph. I’ve used it because I want everyone to be equally horrified by the degree to which OUR ears are stoppered up by these monsters. Because as revolting as this picture may be, all of those ticks won’t kill the dog. They just destroy his hearing and ruin his brain.

Seeing is believing. Once you see the meme introduction and contagion process, you WILL take every step necessary to rid yourself of them. You WILL become more self-aware. You WILL achieve a greater liberty of mind, which is the only effective treatment for a meme infection.

And that’s what we can do. That’s what Epsilon Theory can do. Not try to be a “fact checker”, because that’s a fool’s gig in a world of Fiat News, where everything you hear is in service to this Narrative or that, a self-serving political or economic view served up with some veneer of “fact”. No, what we can do is measure what IS, without attaching any affect or opinion as to whether it is RIGHT. What we can do is visualize what has heretofore been HIDDEN, so that we can go beyond the immediate communication game and SEE the metagame.

Because you’re smart enough to make up your own damn mind.

PDF Download (Paid Subscription Required): http://www.epsilontheory.com/download/15704/


The Many Moods of Macro

I think the original version of this gag is from a Far Side comic in reference to Irish setters, although I’ve omitted it out of respect for Gary Larson’s wishes. Truth be told, I always felt that Old English Sheepdogs would have had a better case for “creature who looks more or less the same regardless of circumstance” than setters. I guess this is one of those things that is infinitely transferable to whatever kind of dog you had growing up.

Unless your childhood dog was a global macro portfolio manager, however, I suspect the rather monotonic flavor of their returns has puzzled you from time to time. For all its inputs, for all its data packaged together from far-flung corners of the globe, all synthesized into sensible and well-researched models, the typical macro fund’s positioning and success is heavily reliant on a small number of influential drivers and environments.

On the surface that’s not necessarily a bad thing, unless you’re paying a ton for it, which you probably are, even in 2018. After all, repeatability and persistent premia are not bugs, but features that we seek out from systematic investing. But for investors in systematic tactical strategies and global macro hedge funds, the expectation of persistent novel sources of return should be scrutinized. In a Three-Body Market, they should be doubted.

A horse having a wolf as a powerful and dangerous enemy lived in constant fear of his life. Being driven to desperation, it occurred to him to seek a strong ally. Whereupon he approached a man, and offered an alliance, pointing out that the wolf was likewise an enemy of the man. The man accepted the partnership at once and offered to kill the wolf immediately, if his new partner would only co-operate by placing his greater speed at the man’s disposal. The horse was willing, and allowed the man to place bridle and saddle upon him. The man mounted, hunted down the wolf, and killed him. The horse, joyful and relieved, thanked the man, and said: ‘Now that our enemy is dead, remove your bridle and saddle and restore my freedom.’ Whereupon the man laughed loudly and replied, ‘Never!’ and applied the spurs with a will.

— Isaac Asimov, Foundation (1951)

At their core, most macro models are central banking models and macro managers are carry investors. They willingly tied themselves to success in predicting bank actions, and in so doing had a wonderful stretch of good returns and low correlations with stocks. Now that predicting bank action will increasingly require short carry positioning, and now that betting on uncoordinated action has gotten tougher, they’re feeling the spurs. This is your choice, too: buck the rider or feel the spurs.


The impulse to find a “a man who can make a plan work”, from F.A. Hayek’s brilliant Road to Serfdom cartoons, is not just a political one, but infects the way we make decisions as investors. We make portfolio plans ourselves, with our committees or with our advisors, and they…rarely turn out exactly like we wanted.

We never have the best possible portfolio we could have had. There is no decision structure that won’t yield questions of the, “Well, why weren’t we 100% in U.S. growth stocks the last three years?” ilk from our clients. More often, we made some real mistakes. We misread the risk environment and weren’t fully invested for our clients. Knowing we shouldn’t, we gave up on a value strategy in a 7-year drawdown right before sentiment turned. We sold bonds ahead of what we thought were inevitable rate hikes and were wrong for five years.

We know we need to fix these kinds of errors.  Too often our solution is to find the team with a model that understands “how all this madness fits together” and can exploit it for us. That’s the allure of systematic macro and tactical asset allocation. It’s well-intended. It’s also a path paved with peril.


This is Part 1 of the multi-part Three-Body Alpha series, introduced in the recent Investing with Icarus note. The Series seeks to explore how the increasing transformation of fundamental and economic data into abstractions may influence strategies for investing — and how it should influence investors accessing them. 

After Ben wrote The Three-Body Problem, and then again with The Icarus Moment, I suspect I reacted like many other readers. I didn’t have to predict whether I thought asset prices were increasingly driven by abstractions or higher degree derivatives of economic and fundamental data, as Ben argued. I was observing it. But between those observations and the related belief that most alpha-oriented strategies have been forced into deeper levels of the Keynesian Newspaper Beauty Contest — that we are increasingly in the business of predicting what others are predicting others are predicting rather than the impact of changes in real economic facts — I have the same questions: How should this impact our strategies? Our portfolios? The questions we ask our advisers and fund managers?

These questions only matter if this whole state of affairs persists, of course. If there’s a hypothetical criticism you could level at the framework Ben and I are working from, it is that the Narrative-driven market isn’t really a thing, that it’s just a label we are throwing on a period of temporary loopiness created by central bank-driven hyperliquidity. A period that appears to be ending. If you think this is the case, then we’re the guys in burlap sacks on the street corner shouting, “This Time It’s Different” right before things go back to normal. I’m empathetic to the view. I mean, I think the view is wrong, but I’ve heard that people are comforted when you tell them you’re empathetic to their view.

I think it’s wrong because we aren’t just observing this in financial markets. We are observing polarization and quantization — rounding words and numbers to their nearest analog — in nearly every human social sphere. Our media-connected Panopticon converts every uttered word into a loaded ideological message, in which every action is a symbol in service to a Narrative. Yes, central bankers were the original missionaries in our little history, but CEOs, financial media, crypto-experts, senators, regulators, traders and other power brokers are all wise to the game now. So, if you want to tell me we will see a return to a market in which the transmission of economic data and fundamental characteristics of issuers manifests in asset prices over some meaningfully investable period of time, fine. But you’ll have to tell me why you think that’s going to happen in politics, culture and media, too.

The painful Catch-22 for the investment advice industry is that people expect times of uncertainty to be the opportunity for advisers to prove themselves. When I talk to financial advisers and RIAs in times of perceived dislocation in asset prices, they want to know whether they ought to transition some of their stock portfolio to hedge funds. They want to know if now is the time to allocate to long/short equity managers. They want to find someone who can steer exposure to take advantage of dispersion when the dislocation corrects. When geopolitical volatility doesn’t manifest in market risk, the conversation is similar. Investors want a macro strategist with a model that answers how it all fits together. Maybe it’s as simple as adding a tactical asset allocation overlay through one of the big turnkey platforms, or maybe it’s hiring a systematic global macro hedge fund. There’s finally dispersion again, and the beta rally is over — now go be tactical and find alpha!

Want to know why Ben’s notes Tell My Horse and Three-Body Problem each yielded more emails from fund managers, CIOs, pension executives and investment professionals than many notes combined? Because they don’t know exactly what to tell their clients who are looking for macro guidance. Because not only is this environment not turning out to be a goldilocks regime for tactical, cross-asset, alpha-seeking managers, it’s becoming an environment in which even the things that used to work aren’t working for them. At all.  Nowhere is that truer than in strategies sitting at the confluence of what in our framework we are calling Systematic strategies operating Economic models. Don’t believe me? Here’s the very long-cycle trend, seen through the lens of the HFRI Macro: Systematic Diversified Index.

Source: eVestment April 2018. An investor cannot invest directly in an index. For illustrative purposes only.

The systematic universe has a lot of trend-following funds. Many of those have performed quite poorly. But that isn’t all that’s happening here. Even the broader Global Macro category, represented here by the HFRI Macro (Total) Index, looks similar. We could similarly split the systematic category into those focused only on trend-following and those that are not, and it would tell the same story.

Source: eVestment April 2018. An investor cannot invest directly in an index. For illustrative purposes only.

What do we mean by “Econometric GTAA”?

The investment industry loves to obfuscate, and terminology can be a bear. Let’s cut through it.

In Hedge Fund Land, “Macro” — represented in the second chart above — is a term of art. Jargon. It refers to a universe of hedge funds, usually self-labeled, that pursue strategies that mostly allocate across and between different broadly defined assets. The term “Systematic Macro” simply refers to those which do so on a mostly systematic basis. By systematic, I mean that the trades are typically generated based on a system rather than determined by a human. That means different things for different funds, and many will have individual sleeves of the portfolio or elements of the portfolio’s construction that still come under human scrutiny. But in general, these funds trade based on generalized ideas and principles memorialized in code.

Depending on who you are talking to, you will also hear strategists, fund managers or consultants talk about “GTAA”, or Global Tactical Asset Allocation strategies. When a fund manager calls a strategy that allocates across assets GTAA instead of Macro, he usually means that he is willing to sell it to you for a lower price, often means that his trades will be confined to a defined, larger set of simple long and short expressions on broad asset classes, and sometimes means that he will have a general bias toward being long financial markets exposure. This is part of the universe I’m writing about here.

There’s a third category of strategies which are not precisely a sub-set of Systematic Macro, but at least occupy a big, overlapping part of the Venn diagram: Managed Futures and CTAs. This, again, is where terminology gets confusing. Managed Futures and CTAs are technically a structural category, by which I mean that they aren’t so much defined by what they do as by what they are. Because futures contracts were originally an instrument devised to trade commodities more efficiently, this industry and its structures formed around strategies for trading futures contracts on those commodities — which makes sense, since CTA stands for “Commodity Trading Advisor.” Over time, as extraordinarily liquid futures contracts became available in equities and interest rate markets, the CTA structures were able to accommodate strategies that looked almost exactly like what we’d see in a global macro hedge fund. But, as I noted in my quip above, this part of the universe tends to trade more often based on price trends, rather than what’s going on in the economy or in companies and other issuers.

So, we have three heavily overlapping Venn diagrams — Systematic Macro, GTAA and Managed Futures. But this piece is about strategies I’m labeling as Econometric GTAA. What do I mean? I mean strategies which trade long and short across a broad range of markets based on computer models driven by (and sometimes predicting the trajectory of and rates of change in) inflation, interest rates, asset flows, economic growth, corporate margins and earnings more broadly, tax rates, trade policy, balance of payments, and trade deficits, etc. These strategies will find their way into your portfolios in many ways. If you buy a Global Macro hedge fund, you will probably get a lot of this. If you hire a Managed Futures fund, you may get some of this, although as I mentioned, it is more likely to be driven by trend-following. If you buy a Multi-Strategy hedge fund, you will probably get a lot of this. But this isn’t just hedge funds. If you buy a “rotation” or “tactical” strategy from an ETF strategist or Tactical Asset Management Plan (TAMP), you will probably get a lot of this. If you hire a financial adviser from an institution with a home office that recommends asset allocation models, you are probably getting some muted flavors of this. So what do these strategies look like?

What Econometric GTAA Models Look Like

When you hire a “tactical” advisor or portfolio manager, while there is a huge amount of surface-level diversity, what you’re usually getting is some subset of the below:

Illustration of Typical Tactical Asset Allocation Framework

Source: Epsilon Theory April 2018. For illustrative purposes only.

The basic framework here takes in some combination of what I’m calling Econometric Data, Market Data and Sentiment. The mix may differ dramatically. In the case of Managed Futures strategies, many ETF strategists and “Tactical Allocation” funds, and others that call themselves Systematic Macro, models may skew entirely toward Market Data. They may even emphasize price movements above just about every other kind of input. This piece isn’t about those funds, and it isn’t about those strategies. It’s a big topic that deserves its own note, because most of the off-the-shelf model portfolios-in-a-box, ETF-based strategies, sector-rotation strategies and tactical allocation funds rely almost entirely on models driven by Market Data alone, usually simple valuation and momentum models.

But what we’re focused on is that top half — the Econometric Data. The approach that managers and strategists will take to incorporate these data will differ. In some cases, strategies will impute a direct transmission engine between econometric data and (implied) expected asset prices, and their desired position. For example, a strategy may be something as simple as rank-ordering countries by their short-term interest rates, buying the ones with higher rates and shorting the ones with lower rates. There are all sorts of implicit views this expresses on investor asset pricing behavior and risk, but the explicit mechanism for establishing positions connects relative interest rates to relative asset price returns over some period of time. This is what the illustration refers to as Implicit Price Behavior models — “Certain values of variable X will more often than not result in changes in the prices of asset price Y.”

In other cases, one or more (usually more) bits of data will be incorporated with economic logic into an interim model. That interim model will typically represent a more explicit simulation of the behavior of another actor or actor(s). For example, rather than estimate a simple relationship between, say, changes in consensus inflation expectations and whether inflation-linked bonds will outperform nominal bonds, many Econometric GTAA strategies will take in GDP growth, earnings growth, balance of payments, wage growth, producer price momentum, corporate margin and money supply data to predict the pressures on central banks to make changes in interest rate policy. That output would then influence views and positioning on a range of assets.

As with the bulk of tactical asset allocation strategies, ETF models, etc. mentioned earlier, the influence of these interim models or even many of the direct transmission models of Econometric Data to positions is often presented in context of valuation of the underlying assets, and momentum of the price of the underlying asset and/or the model’s signal itself. In other words, a fund may predict the pressure on a central bank to act, but it may be the momentum or change in that variable which produces a tradable signal. Alternatively, a trade may be conditioned on some valuation or momentum state, or even by the state of another interim model (i.e. “We trade when we have confirmation between our geopolitical framework and recent price action.”).

The types of positions these models establish will differ as well. For most strategies — especially those that fancy the GTAA moniker — the views tend to be long/short, and usually asset class neutral. For example, a model might pair a 5% long or overweight position in US stocks with a 5% short or underweight position in, say, German stocks. For others, the views might compare assets with cash. In other words, the models decide whether to have market exposure at all. This question of “directionality” is a big one. It’s one that tends to exaggerate differentness among practitioners of these strategies and pigeon-hole the emphasis of more risk-focused managers into a smaller number of relative value trades between assets.

But What is It, Really?

So with all these inputs, with all this diversity, we have our pick of a lot of interesting multi-asset and macro strategists with a lot of interesting different models, right?

Meh, not so much.

We don’t claim to have some secret sauce for analyzing drivers of fund performance. Most approaches are pretty well-trod ground at this point, although I was tempted to measure facial width-to-height ratios just for fun. But no, we’re simple — boring multi-factor regressions against some basic style and market factors. Fortunately, as is often the case, simplicity tells most of the story. Of the 74 funds in the HFRI Systematic Macro universe, 60 have positive, statistically significant betas to interest rates. Around 40 have betas higher than 1.0. These are betas in a multi-factor context that includes a range of market and traditional style factors. To be fair, many of these funds are implemented through futures or with market neutral positioning that allows them to earn cash returns, but this is a small portion of this effect. In the end, you are roughly three times more likely to run into a Systematic Macro fund with returns that look like a levered bet on interest rate-sensitive instruments than one that neutralized (read: hedged) the aggregate systematic influence of rates on its risk profile over time.

Source: Epsilon Theory, HFR, eVestment April 2018. An investor cannot invest directly in an index. For illustrative purposes only.

I don’t think this is an artifact or false positive from the data. Anecdotally, as I’ll argue, I think that many systematic macro funds really do execute strategies that are structurally biased toward being long bonds. But they also have a related bias. They like to own things where you collect a payment from someone else to own it. That someone else may be an issuer, a government or a hedger. Across macro funds, GTAA strategies and model portfolios, the only strategy that is as common as trend-following and a bond bias — systematic ones in particular — is buying higher yielding assets and selling lower yielding assets. We call these “carry trades”. Below are the significant betas (above and beyond the relationship to bonds) of each of the 74 funds to our measure of multi-asset carry trading. Most are positive, only a couple are negative, and the rest are largely positive but not significant after accounting for the existing carry component in their interest rate exposure.

Source: Epsilon Theory, HFR, eVestment April 2018. An investor cannot invest directly in an index. For illustrative purposes only.

Systematic Macro and Econometric GTAA funds have other systematic exposures as well, including a general bias toward short exposure in commodities, and a long bias toward equities, and these aren’t just present in the trend funds that have had those exposures because they have worked. They are common throughout.

If you talked about this with your tactical guy/gal or macro strategist, I know what he’s going to tell you, because it’s what they tell me, too. “There may be some of this, but we’re not directional. We may be long, we may be short, and we have a lot of other trades and signals.” I think you’ll find that this is sometimes completely true. For example, Bridgewater’s Pure Alpha strategy is consistently neutral to most market factors, although if you looked deeper into carry trades, you’d find a pretty persistent positioning in favor of higher yielding currencies. But generally speaking, your manager is telling you a half truth — literally. Across this universe, using static exposure to the most basic of market factors (stocks, bonds, commodities and currencies), you can explain around half of the variability in returns.  That’s not the problem, except that you’re paying them 1.5-and-20 for something you can and should get much cheaper. The problem is that after you take out the 50% you can explain with market factors, the hand-waving, black box, smoke-and-mirrors half they’re trying to sell you as their edge is a tire fire.

I mean, gods, look at this mess. Over the last five years, you would have gotten a positive Sharpe Ratio on whatever these guys did that wasn’t static long exposure to financial markets from only 23 of the 74 funds. Believe your model-driven macro guy when he tells you he isn’t only directionally long rates, -carry and trend-following. But be skeptical when he tells you that you ought to have a positive return expectation on whatever the other stuff he’s doing is.

Source: Epsilon Theory, HFR, eVestment April 2018. An investor cannot invest directly in an index. For illustrative purposes only.

The Many Moods of Macro

So how should we feel about the non-tire fire half of these returns — by which I mean the rates and carry half?

As I’ve alluded to above, the first mood of macro has always been betting on the behavior of central banks. There are a variety of reasons for that, but the first is that Systematic Macro, like other hedge fund strategies — and systematic ones in particular — is drawn to trades, strategies and markets with lower natural volatility. That means, generally, that positioning driven by the models will emphasize either directional exposure to lower volatility asset classes like bonds, or relative value trades (i.e. going long one asset and short a similar one). Importantly, it also means that the models will favor what they perceive as loosely correlated trades or positions. If you have better-than-random confidence in what central banks are likely to do, you have a full range of options to implement those views with characteristics that are attractive to a manager seeking to sell itself as limiting downside with significant uncorrelated upside. The other reason that funds were so fond of strategies reliant on their central bank behavioral models, of course, is that the models themselves were historically pretty effective.

But there’s another, more important reason for the attachment to central bank-driven econometric models. While most trades with significant upside potential and convexity require you to pay a premium — or be “short carry” — during the bond bull run of the last 30 years, a macro manager with a decent prediction model for the behaviors of central bank actions could put on trades with convex characteristics that paid him a positive carry over almost all of this three-decade span. For example, a prediction of a future rate cut would not only get the benefit of the signal but would also receive the term premium in an upward sloping rates curve, a premium that can be significant even in the front end of the curve.  Not only that, because central banks in different countries pursued frequently divergent policies with explicitly different inputs and aims, that manager could put on multiple such trades. And what’s more, these positions were uncorrelated to most portfolios’ primary sources of risk — we were living in a golden age!

This preference for long carry positioning is itself, I think, the second constant mood of macro. While it has historically manifested in part as a willingness to carry a directionally long bias in bonds, it also manifests in a preference for anything that pays you more to own it than something else. In Systematic Macro funds, you will see this in models which — from a variety of econometric inputs — ended up with a consistent preference for higher carry currencies, especially certain emerging markets currencies. The underlying model mechanics might differ. For example, the “emerging markets balance sheet quality” thinkpiece was a staple of the mid-2000s. The story went that the higher yields you earned for owning — I don’t know, Turkish lira — were compensation for perceived risk, but that the econometric support for fiscal stability and quality meant that the real risk of permanent capital loss was far lower. There are a hundred models with rationale like this that all lead to a pronounced bias toward carry.

The third ubiquitous mood of macro is pro-trend, and in particular, medium-to-long-term trend following. I’ve noted that much of the Systematic Macro universe overlaps with Managed Futures and CTAs that have made trend-following their bread and butter, but even among Econometric GTAA strategies, it is quite common to use trends and momentum to drive positions or to condition other models. It is, perhaps, even more common to measure trends in the econometric variables as strategies or factors of their own. Systematic models will also be driven toward pro-momentum stances by their risk management techniques. Because positive returning assets are generally lower volatility assets, things that have done well will tend to score well when the portfolios are being built, even if there is no explicit model saying, “Buy stuff that’s going up!” Beyond that, because many of these funds added implicit or explicit stop-loss logic in 2009, even among funds I would qualitatively (and subjectively) describe as following Econometric GTAA strategies, you can frequently explain much of the variation in returns with price momentum across asset classes of various horizons.

All of this is, incidentally, also why so many of your discretionary macro managers have spent the last two decades trotting out their Eurodollar guys to meet with you. They’re the guys who got paid to bet on central bank actions with diversifying, pro-momentum, carry-paying trades.

Systematic Macro in Three-Body Markets

Now, I don’t have much to say about whether I think that carry trades and momentum trades will work — or at least I don’t have much to say in this note. Suffice it to say that there are good behavioral and empirical reasons to think that they are persistent premia that compensate investors for bearing a certain type of risk, and also good reasons to think that may manifest somewhat differently in markets driven by Narrative abstractions. These are topics for another note.

But I do think it is clear that Econometric GTAA strategies have struggled mightily to adapt to a Narrative-driven market with everything else they are doing. When I have met with strategists and macro funds over the last few years, their language has been static. They walk me through the economic nonsense of negative interest rates and the upside asymmetry created by the zero barrier. They describe how relative growth rates, debt levels, deficits and trade balances cannot possibly support the relative yields of Treasuries and Bunds. They present compelling cases for curve trades between European markets that should converge given influences on ECB asset purchases, or for relative outperformance of this equity market over that because of the extension of corporate margins beyond some historical threshold for some historically long period of time. In other words, the Explicit Behavior models from the illustrated earlier in this note are running through the old motions on what central banks, asset owners and governments are going to do, and none of them are working.

It’s not as if managers and strategists haven’t tried to adapt to a world in which these trade drivers are subsumed into central bank communications strategy and the utilitization of markets by political figures. But when the models driving your trading are built to understand the cost and transmission mechanisms of capital, and your asset prices are driven by how investors think other investors are responding to a stronger adverb in front of a maybe hawkish adjective, it’s got to be more than just adapting and updating your models. It’s recognizing that over an expanded time horizon, asset prices are being driven by a wholly separate set of variables.

What worries me more than anything, however, is that the ability of GTAA and macro strategies to access the moods that may still work (mostly models that end up looking like carry trades) may be limited by their construction and their design in the emerging environment. As macro and GTAA strategists adapt to rising interest rates and inflation, I think you’re going to see some of their models under strain. They will be under strain because their central bank behavior models will be shouting “short rates”, but their DNA, their risk management framework, and maybe even explicit pro-carry models — will be screaming “God, that short looks expensive. Are you sure?” This dependency — implied already in the assessment of sensitivity to bond markets — is real. Since 2000, the average Global Macro hedge fund has generated roughly 2.4x its average return in months where the discount rate was reduced, with almost no advantage during periods with a rising discount rate. I think we may be entering an environment in which the only things that have really worked for these funds get lost in the wash. In their place, I think investors can expect to get weaker, lower Sharpe strategies based on Market Data-driven positioning. A lot of momentum and value, and not much novel insight — in many cases, a very expensive balanced market portfolio with a value overlay.

What do I do if I’m an allocator?

  • I don’t put all my trust in a macro strategist or tactical manager’s econometric models to ‘figure this all out’.
  • I probably own fewer of these funds and have less of my portfolio invested in them in the aggregate.
  • I ask for a manager’s rates and FX attribution. If it’s positive for most of the 2000s and starts to sour and turn over in, say, 2013-2014 and hasn’t recovered, I take the plastic binding rings out and file the pitchbook away in that special Iron Mountain filing cabinet in the print room. You know, the one with the lock on it that the guy comes for every couple weeks?
  • I challenge my macro PMs to explain how their models might approach rates and FX trading differently when rates are rising and if central bank policy remains largely coordinated. If they say, “We’re not directional and have always been agnostic on long or short positions, so it wouldn’t be any different,” I do my most exaggerated eye roll, put my chin on my fist, bat my eyelashes and give ‘em my best ca. 1991 Glamour-Shots-at-the-mall smile until they tell the truth.
  • I actively seek out managers who are incorporating and increasing the role of sentiment analysis, investor asset flows, market structure and Narrative, and reducing the role of econometric models, whether explicitly or through a systematic process for rotating capital between models (or turning off models whose alpha has evaporated).
  • I actively seek out managers who have actually figured out multiple robust models for trading commodities effectively other than trend-following models.
  • I continue to invest with managers accessing Systematic Macro’s traditional moods, but I’m only willing to pay what those replicable strategies are worth.
  • I look for managers who act boldly but hold their views loosely — in a word, humility.

The last goes for you and me, too. We must be humble about our ability to make good predictions about asset prices and returns. In a Three-Body Market, we should be even more humble than usual. But blindly handing over the reins of our asset allocation decisions to impressive people who claim to have developed the one model that will unravel this market’s mysteries is not an act of humility. It’s the very same act of expedience that caused so many of these managers to saddle themselves and their strategies to central banks, and it is the reason they’re feeling the spurs today. Buck it — or feel the spurs yourself.

PDF Download (Paid Subscription Required): http://www.epsilontheory.com/download/15710/