We’re Doing It Wrong


Ed. Note: we are thrilled to announce the launch of Epsilon Theory Professional Service, where financial advisors and investment professionals can tap directly into our Narrative Machine market research and analysis. In this note we are highlighting the value proposition of the Narrative Machine research program. – Ben

Flammarion engraving (1888)

We’re all explorers seeking to pierce the veil, hoping against hope for a vision of the music of the spheres beyond this messy world. Looking for an Answer in the clockwork machine that we all believe incorporates and underpins markets.

I think we’re doing it wrong.

I think investment professionals, quant and non-quant alike, are misusing the massive computing power that each and every one of us has at our fingertips. Whether it’s the powerful computer that we call a smartphone, whether it’s the crazy powerful multi-threaded computer that we call a laptop, whether it’s the insanely powerful computing utility we call AWS or Azure or the like … we’re using machine computing processes as an extension of our human computing processes.

This is a classic anthropomorphic fallacy.

Meaning that we can’t imagine what it would mean to use computers in some other, non-human way. Meaning that we never even consider whether there’s a human way of perceiving the world, much less what a non-human approach might be.

Here, I’ll give you an example.

In every how-do-we-use-computers-in-investing conversation I’ve ever had with anyone in tech … in every how-do-we-use-computers-in-investing conversation I’ve ever had with anyone in finance … in every how-do-we-use-computers-in-investing conversation I’ve ever had with MYSELF … the conversation, either implicitly or explicitly, is ALWAYS about using computers to find some hidden formula that will make us lots of money.

Always. Without exception. Ever.

This is a slide that ET contributor Neville Crawley made a while back, and it slays in meetings. It resonates. It sings.

Oh yeah, I see why we want this artificial intelligence system (I mean, I don’t know why you’re calling it Big Compute, but whatever). 

It’s the next level. It’s the Giant Brain, replacing the Big Brain of all those computers that DE Shaw and Two Sigma and RenTech are using to figure out markets and mint money, which replaced the Little Brain of us humans scurrying around in the pits. AI is going to pierce through all the noise and find us the signal. It’s going to identify the pattern. It’s going to tell us the Answer.

Do you feel it? I feel it. It’s why I became a professional investor in the first place. To figure it out. To find those patterns and signals that would make me rich. To find the “tell” of markets.

But that’s not how AI works. That’s not how any of this works.

AI is not a giant brain, and there is no Answer to be found.

AI isn’t even a Difference Engine, as Charles Babbage called his programmable analytic device that we now call a computer. It’s more like a Connection Engine, able to “see” the similarities in a million-fold matrix all at once. It’s a non-human intelligence, more like an insect’s compound eye + nervous system than anything human-ish. And yes, there’s an oldie but goodie Epsilon Theory note for that.

As for the Answer …

The absence of an Answer – by which I mean the non-existence of a general closed-end solution or a predictive algorithm in any physical system of three or more interactive entities and certainly in any social system – is at the core of two canonical Epsilon Theory notes: The Three-Body Problem and Clear Eyes, Full Hearts, Can’t Lose. Honestly it’s the heart of the entire Things Fall Apart series of ET notes.

This message – that there is no predictive algorithm for social systems – bears repeating over and over, because the human brain is hard-wired to seek that algorithm. We literally cannot help ourselves. I believe it’s the root of every totalitarian impulse, large and small, that the human animal has ever experienced. That totalitarian impulse is most obvious – and most deadly – in our social system of politics, but it is no less present in our social system of markets.

We think of markets as a clockwork machine, as an intricate collection of gears upon gears. We believe that if only we examine the clockwork closely enough, we can identify some hidden gear or unbeknownst gear movement that will let us predict the clockwork’s movement and make a lot of money.

Our MODEL of markets is The Machine, and every Machine has a deterministic set of algorithms that create and drive it. Every Machine has an Answer.

This model – the market as machine – is an anthropomorphism.

There are lots of historical and anthropomorphic reasons why we think of social systems as machines. But they are all historical and anthropomorphic reasons. There’s nothing “natural” about it.  And yes, there’s an Epsilon Theory note on that, too.

I’m sorry, Ray Dalio, but as a philosopher you’re a fantastic hedge fund manager. 

To be clear, market-as-machine is a perfectly useful anthropomorphic model for most investment purposes, just as Ptolemy’s Earth-centric universe was a perfectly useful anthropomorphic model for most navigational purposes. Seriously, if your goal is to sail your ship from Tyre to Ostia, then you can’t do better than celestial navigation per Ptolemy. If you want to go to the moon, on the other hand …

Anthropomorphic models break when a revolutionary invention allows us to SEE the world in a non-human way.

For the Ptolemaic earth-as-center-of-the-universe model, that revolutionary invention was the telescope and the ability to see sunspots and Jupiter’s moons and all sorts of astronomical objects and phenomena that were, literally, previously invisible to the HUMAN eye.

AI is the revolutionary invention that breaks the market-as-machine model. It allows us to SEE narrative and sentiment and all sorts of social objects and phenomena that were, literally, previously invisible to the HUMAN eye.

To be sure, this new invention that lets us see in non-human ways isn’t a sufficient condition to break an anthropomorphic model. These models become so embedded in our social institutions and our minds that, as with Ptolemaic science and the invention of the telescope, it can take a hundred years and a lot of violence for a better model to be widely accepted.

And that’s the problem with AI for most investors, quant and non-quant alike.

If you use computers in your investment research process – and I know you do – I will bet you umpteen zillion dollars that you have those computers looking at structured historical data in an effort to find some repeating pattern. I will bet you do this rigorously and intentionally if you’re a quant. I will bet that you do this all the same, but non-rigorously and haphazardly if you’re not a quant.

Whether you realize it or not, you are using the market-as-machine model. You are looking for the Answer. Go on, you can admit it. You’re among friends here. I’m like Big Lou in the insurance ads … I’m one of you. It is embedded in our minds and in our businesses. Mine, too.

But here’s the thing.

If you use AI as just another input to that market-as-machine investment research process, you will get puzzling “results” that don’t help you very much. It will be just like using a telescope to get better measurements of the retrograde motion of Mars as it orbits around the Earth in your Ptolemaic model.

You will be disappointed by AI.

I want to suggest a different way to think about markets, a non-anthropomorphic model that works WITH the revolutionary invention of AI and Big Compute.

The market is not a clockwork machine.

The market is a bonfire.

We all know the physics of fire. The underlying rules of combustion are as clear and as deterministic as any pendulum or gear movement. Fire is not magic. Fire is not somehow separate from science or rigorous human examination. We know how to start fires. We know how to grow and diminish fires. We know how to put fires out. In a technical sense, Ray, you can classify fire as a machine.

But you’d never think that you could possess an algorithm that predicts the shape and form of a bonfire.

You’d never think that if only you stared at the fire long enough, and god knows humans have been staring at fires for tens of thousands of years, that somehow you’d divine some formula for predicting the shape of this or that lick of flame or the timing of this or that log collapsing in a burst of sparks. 

No human can algorithmically PREDICT how a fire will burn. Neither can a computer. No matter how much computing power you throw at a bonfire, a general closed-end solution for a macro system like this simply does not exist.

But a really powerful computer can CALCULATE how a fire will burn. A really powerful computer can SIMULATE how a fire will burn. Not by looking for historical patterns in fire. Not by running econometric regressions. Not by figuring out the “secret formula” that “explains” a macro phenomenon like a bonfire. That’s the human way of seeing the world, and if you use your computing power to do more of that, you are wasting your time and your money. No, a really powerful computer can perceive the world differently. It can “see” every tiny piece of wood and every tiny volume of oxygen and every tiny erg of energy. It “knows” the rules for how wood and oxygen and heat interact. Most importantly – and most differently from humans – this really powerful computer can “see” all of these tiny pieces and “know” all of these tiny interactions at the same time. It can take a snapshot of ALL of this at time T and calculate what ALL of this looks like at time T+1, and then do that calculation again to figure out what ALL of this looks like at time T+2.

Want to guess who spends more money on Big Compute than everyone else in the world combined?

It’s the U.S. government, through the Dept. of Defense and the Dept. of Energy.

Know why they’ve spent BILLIONS of dollars on the world’s most advanced supercomputers?

To calculate fire.

Not just any old fire, of course, but nuclear fire. This is why the most advanced computers in the world today have been built – to simulate the explosion of nuclear weapons. To see the future by calculating the future, not by analyzing the past for predictive algorithms.

It’s a hard concept to wrap your head around, this distinction between calculating the future and predicting the future, but it’s the key to thinking about your investment research process in a non-anthropomorphic way. It’s the key to successfully and profitably incorporating the revolutionary invention of AI and Big Compute into your investment research process.

Now let’s be really clear … we’re a loooong way from performing the market equivalent of simulating H-Bomb explosions with the Narrative Machine. No, we’re more at the stage of taking a rudimentary telescope and aiming it at the sky. We have neither the ability to “see” market participants at a molecular level nor the ability to “know” the physical interaction rules of these participants at anywhere near the same precision or “resolution” that the DoD can see or know nuclear reactions.

But when we show you a Narrative map of Inflation like this, that’s the path we’re on.

We’re taking ALL of the thousands of financial media articles published over some period of time that mention “inflation”, and comparing every word and every phrase in every article to every other word and every other phrase in every other article. It’s a million-fold matrix that “sees” these publications and their inchoate arguments all at the same time and measures their connectedness and similarities all at the same time, then visualizes that connectedness in dimensions that make sense to a human eye and mind – color, distance, size, position, etc.

This is narrative-space, and it’s something that investors have always felt or believed existed, but we’ve never been able to SEE. Until now. 

We can’t give you a secret formula for predicting markets from looking at narrative-space.

But we can tell you what IS in narrative-space.

What good is that?

  • We think we know some of the “rules” for calculating what’s NEXT in market participant behaviors from what IS in narrative-space. This is the Common Knowledge Game, and it’s a forward-looking, actor-based way of evaluating the path of markets.
  • More importantly, we think that YOU already know many of the “rules” for calculating what’s next in whatever corner of the market is important to you. We think that experienced discretionary investors, traders, allocators and advisors have an enormous amount of internalized knowledge about the relationship between narrative-space and market participants in their arena of expertise. We think that a visualization of narrative-space can weaponize your internalized knowledge.

Tapping directly into the Narrative Machine is not for everyone.

If you’re looking for a new variable for your regression analysis, you don’t want this. If you’re looking for a new data feed that you can analyze and arb, you don’t want this. If you’re running a purely systematic or passive investment strategy, you don’t want this.

But if any aspect of your investing or your portfolio allocation is still human … if any aspect of your investing or your portfolio allocation is still discretionary … we think you’ll find the Narrative Machine research project worth a look.

Not because we can give you an Answer.

But because we can advance your Process.


Leave a Reply

Please Login to comment
newest oldest
Notify of
Mike S
Mike S

“Want to guess who spends more money on Big Compute than everyone else in the world combined? It’s the U.S. government, through the Dept. of Defense and the Dept. of Energy.” Tapping directly into the Narrative Machine is not for everyone…It will be an interesting journey, but narrative is imbedded in the world of digital exhaust, if you want to see where DoD tactics are going in the digital world and how that might cross pollinate, then look below! Narrative is a way, but cutting through the noise will only get louder and more complex. DARPA leaders are calling this “mosaic warfare.” Modern-day weapons to pieces in a jigsaw puzzle. Each piece of the puzzle is exquisite. It can only fit one way into the picture, and if one loses a piece, then the picture is incomplete. A tile in a mosaic is one small part of a bigger picture. “If you lose one tile, not a big deal,” he said. In this metaphor, a tile equals an individual weapon. Part of the concept is “combining weapons we already have today in new and surprising ways,” Grayson said. Key will be manned-unmanned teaming, disaggregating capabilities, and allowing commanders to seamlessly call on effects from sea, land or air depending on the situation and no matter which of the armed services is providing the capability. An example in the air domain might include a series of drones to accompany a typical battle formation of four fighter aircraft. One of the robotic… Read more »


Thanks for this note, Ben, it’s really illuminating. I think the “Doing it wrong” problem you identify in finance is very similar to a “Doing it wrong” problem I see in the broader AI field today. Instead of your dichotomy of “predict vs. calculate/simulate” I use “fit vs. model”, but I think they are more or less the same thing. A lot of mainstream AI today is about taking a set of possible inputs, each one associated with a desired output, and coming up with a function that does the best possible job doing the desired association. A grad student looking to get published will take a bunch of pictures of dogs, each one labeled “dog”, a bunch of trucks labeled “truck”, etc. and try to end up with a neural network that takes a new picture of a dog and calls it “dog”. Somebody trying to find an edge in a market might take a set of training examples, each one consisting of the price history of the week leading up to day T, each one associated with the return that occurred on day T+1. This person wants to be able, at some time in the future, to take the returns of the last week and predict what will happen tomorrow. Either way, they’re FITting a relationship between inputs and outputs, which is a very different thing from MODELing the real-world system that happened to produce that relationship in the first place. The line can be blurry and there… Read more »

Mark Kahn
Mark Kahn

I learned early on, when I started, back in the ’80s, that there is no Answer to the markets – certainly not one anyone could find – but I could “feel” something in the market – when sentiment was shifting or a market “story” was dead, for example – and combined with a rigorous application of almost-trite trading and investing rules (that many spout but few sincerely and consistently follow) have made a reasonably successful career out of trading and investing, despite never getting anywhere near an Answer. Over all these years, never believing there was an Answer out there helped keep me humble and saved me a lot of time, but as Ben pointed out in “American Bandstand,” it can be rational to act as if you believe a bit (even if you don’t) because it is part of the common knowledge. Thoughtful arm-length believing – and letting others think you believe – can help your trading and investing and is, IMO, the only way, at times, to keep your career and credibility afloat (sad, but true). I kind of recognized the common knowledge game, but never fully understood it – but “felt” it enough to use it in my investing. Ditto, “inflection points,” “when the story fails,” and so many other smart ET ideas and approaches that I didn’t even know I was leveraging in my trading and investing. It wasn’t until 2013, when I read Ben’s ET manifesto, that I realized I’d been nibbling at the edges… Read more »


Completely agree. This blog has been a revelation. I used to work as an equities analyst and then left because I realized that approaching markets the standard way (think CFA syllabus) was a mug’s game. You’d only be as good the rest of the market if you do what the rest of the market does. I went into foreign policy work thinking I’d learn something about geopolitics and, in turn, global macro, but in the last few weeks I’ve probably learned as much about the latter as I had in years working in diplomacy. You guys might meanwhile like this: https://www.the-american-interest.com/2017/07/03/flattened-disintermediation-goes-global/

Jane VanFossen
Jane VanFossen

Michael, thanks for the link. It’s a good one.


The Latest From Epsilon Theory


This commentary is being provided to you as general information only and should not be taken as investment advice. The opinions expressed in these materials represent the personal views of the author(s). It is not investment research or a research recommendation, as it does not constitute substantive research or analysis. Any action that you take as a result of information contained in this document is ultimately your responsibility. Epsilon Theory will not accept liability for any loss or damage, including without limitation to any loss of profit, which may arise directly or indirectly from use of or reliance on such information. Consult your investment advisor before making any investment decisions. It must be noted, that no one can accurately predict the future of the market with certainty or guarantee future investment performance. Past performance is not a guarantee of future results.

Statements in this communication are forward-looking statements. The forward-looking statements and other views expressed herein are as of the date of this publication. Actual future results or occurrences may differ significantly from those anticipated in any forward-looking statements, and there is no guarantee that any predictions will come to pass. The views expressed herein are subject to change at any time, due to numerous market and other factors. Epsilon Theory disclaims any obligation to update publicly or revise any forward-looking statements or views expressed herein. This information is neither an offer to sell nor a solicitation of any offer to buy any securities. This commentary has been prepared without regard to the individual financial circumstances and objectives of persons who receive it. Epsilon Theory recommends that investors independently evaluate particular investments and strategies, and encourages investors to seek the advice of a financial advisor. The appropriateness of a particular investment or strategy will depend on an investor’s individual circumstances and objectives.