Who haven’t seen the famous meme-movies about the markets? “The Big Shot”, “The Margin Call” and, of course, “The Wolf of Wall Street” (in which, remarkably, there were no mathematical models involved – just plain old bullshitting).

And, when we saw that guy, the second place on a math Olympiad, we probably felt a pang of pain and regret. That could have been me. Math, it turns out, is not that complicated after all.

The much more difficult thing is to understand where some mathematical formalisms are indeed applicable, and where they inevitably will yield bullshit (the “garbage in – garbage out” meme).

Have you seen these nice textbooks like *“Brownian Motion and Stochastic Calculus”* and *“Methods of Mathematical Finance”*? You should.

There is a small problem, however. The actual markets are empirically are nothing like a Brownian Motion, far from it. It is irrational and emotional people being trading shit according to the current set of internet memes. Aside from that there are malicious actors of all sorts, ranging from “Institutional Investors” and “Marker Makers” trading against retail, to plain and simple FTX-like arrangements, where an *exchange’s insiders information (the data)* is accessible to an affiliated trading desks.

There is nothing “Brownian” out there, nothing “Rational” too. No “Perfect Knowledge” and any other theoretical bullshit, just *algos against degens* and lots of other shilled scams.

These *observable and measurable* facts together throw all that nice and fancy mathematics out of a window. While the math itself is correct, and captures some generalizations about the Brownian motion and what not, it is *not meaningfully applicable* to the actual markets, which are constantly evolving and changing as we speak or model them (just like that).

Just as application of mathematical probability to a wrong types of processes (not discrete, fully observable ones like cards or dices) will always yield bullshit (by definition), so would an application of stochastic calculus to a wrong types of processes.

This is the same principles as with *simulations*. First, simulating what is not fully understood and *not /properly captured* (some major factors are missed or unaccounted for, new major factors emerge, other diminish), will always simulate something totally disconnected from the actual reality.

Second, the *disconnectedness* itself is another fundamental principle, which has been captured in the notion that every new coin toss is totally independent from the previous ones.

The markets, by the way, are nothing like coin tosses and any attempts to re-frame (oversimplify) them as such will, in principle, yield bullshit.

This brings us to the most fundamental principle, which could be stated informally as “Reality First!”. It has been captured many times by the ancient notions of *Causality*, which has been known in the East as *Karma* and captured famously captured as *Modus Ponens* rule by the ancient Greeks.

Everything is caused by a sum-total of everything else (all the relevant factors) in that particular locality. This is *The Universal Law*. There is no randomness (which is a pure abstract (generalized) mathematical concept). The complexity of causality does not cast it into randomness. The complexity cannot be seeped under the rug. At least not in the context of the actual markets.

So, aren’t all the mathematical models bullshit? Well, these which are based on over-simplified *naive theoretical assumptions* *are*.

Isn’t this a bit of a *too strong statement?* Not at all. We have so much socially constructed and organized pseudo-scientific sectarian bullshit, which, like all other sectarian bullshits, are based on having a higher social status, that recognizing an another *instance* of the same common social dynamics is nothing special at all.

One more time – the Universal Law of Causality cannot be side-stepped or ignored. What *caused* every single market movements is always “Out There”, and could be captured if we collect enough data. Sometimes it is just a single fucking tweet by Musk, other times it is a *mass-hysteria* about some company which produces internet routers or graphics cards capable of doing very fast matrix multiplications.

The fact that these both bubbles look very similar *does not imply that these are the same phenomena*, but it shows that the social aspects are the major ones, and that there, again, is nothing “random” or “Brownian” in there.

Which models actually work? Well, constantly updated ones, which try to measure and analyze the current social dynamics in realtime and which maintains dynamic, constantly updated “weights” for *some* identified social and *derived* factors.

This is similar to what a Deep Neural Network could *capture* (given enough *measured* data), but the network will allways be outdated, because, by definition, it *captured the past* (just like a photographic snapshot), which is no longer relevant, since the underlying reality has been changed, evolved right while it has been trained.

The solution is to constantly feed the new data (that hopefully correctly captured the actual changes), but this will fail too, because what we capture is an imperfect information and mostly verbalized irrelevant abstract bullshit.

This is actually, another principle (yes, yes, I know) – only relatively *stable* environments can be captured by any kind of Neural Networks. The stable environments remain stable most of the time, and when not, it is called a “black swan” or a “catastrophe”.

This is a serious topic and it goes as deep as “how biology is even possible”. The answer is that the environment is stable enough to sustain its “stable intermediate forms”. So stable, that a slight fluctuation in a temperature could wipe out most of live forms (by boiling out most of their proteins). This is what stable means – no such major fluctuations.

So, is the “quant” a meme? Yes, if what they do is application of an abstract calculus based on academic theories. Just like wrong simulations, it will allways yield a disconnected from reality abstract bullshit, like a speech of a parrot bird (which is literally what modern LLMs produce).

There is nothing new in this conclusion. *“The Man Who Solved the Market”* (a classic meme book) basically shows the same results – all abstract theoretical models have failed miserably, and only some constantly re-trained neural nets are able to *be slightly better than a coin toss* in a long run (because some *patterns* do actually reccur and re-emerge due to the actual underlying social dynamics, which are similar but never exactly the same).

And no, no Karpathy or any other meme guy could just apply ML and solve the markets. The reason is in the infallible *Stable Environment Principle* outlined above. Not everything can be captured by a Deep Network – it can only capture a relation, and that relation has to actually be Out There in the first place.

If we capture what is no longer accurate we cannot *meaningfully* apply it, it is that simple. Yes, we could produce something which looks and feels very convincing, like speech of a bird, or a sophisticated pseudo-scientific text, but it is just a sophisticated and convincing bullshit.

And, yes, Deep Learning is the best we can do. It, obviously, won’t “predict the future outcomes” (it is incapable of doing do in principle, which is what this post is about) but it will identify and show some reccuring patterns in what we have just observed (measured) at the “Right Edge” (of a chart).

Think of less naive “indicators” or of a system which *observes* what all the other market participants see (they see basically the very same charts) and identify the same emerging patterns *they* may act upon in a consistent way. Just thing of the FOMS madness a few years ago, where every single degen became a trader based on what Jerome is going say.

And these “quants” are indeed a meme.