Why Understanding Does Not Grant Control
“The greatest enemy of knowledge is not ignorance, it is the illusion of knowledge.” - Daniel J. Boorstin
You know the rules. You understand emergence, feedback, adaptation. You grasp the dynamics of complex adaptive systems, the way small interactions cascade into large-scale patterns, the way structure arises without a planner, the way the system computes at a level that exceeds any individual participant. You have read the literature, studied the examples, internalised the principles. You understand the game.
Now predict tomorrow’s close.
You cannot. And this is not because your understanding is incomplete. It is because the system itself is structured in a way that makes prediction impossible, no matter how complete your understanding becomes. You can know everything this series has taught and still not know what the market will do next. Not because you have failed to learn, but because the knowledge you have acquired is a fundamentally different kind of thing from the prediction you are seeking.
This is the asymmetry of knowing. Understanding grants orientation, not foresight. It changes how you see, how you build, what you survive. But it does not change the fundamental unpredictability of the system you inhabit. And coming to terms with this asymmetry is perhaps the most important intellectual and emotional challenge the embedded agent faces.
The Chess Problem and the Weather Problem
There are two fundamentally different kinds of prediction problems, and most people conflate them.
Chess is a prediction problem of the first kind. The rules are known. The board is visible. The number of pieces is finite. In principle, every possible game of chess can be enumerated. In practice, the number of possible games is astronomical, but this is a computational limitation, not a structural one. Given enough processing power, you could determine the optimal move in any position. The game is complicated, not complex. The difficulty is in the calculation, not in the nature of the system.
Weather is a prediction problem of the second kind. The rules are known. The physics of fluid dynamics, thermodynamics, and atmospheric science are well understood. We have extraordinary observational data: satellites, weather stations, ocean buoys, atmospheric soundings. We have supercomputers dedicated to running atmospheric models. And yet weather prediction degrades rapidly beyond a few days, and long-range weather forecasting remains fundamentally unreliable.
This is not because we do not understand weather. We understand it deeply. It is because the atmosphere is a system in which tiny differences in initial conditions can produce dramatically different outcomes. The sensitivity is not a flaw in our models. It is a property of the system. No improvement in measurement precision, no increase in computing power, no refinement of our physics will solve this problem, because the problem is not that we lack information or intelligence. The problem is that the system’s evolution is sensitive to details we can never measure with sufficient precision, because there is always a finer scale we have not captured.
Markets are weather, not chess. The rules of market dynamics are knowable. The outcomes are not. And the gap between knowable rules and unknowable outcomes is not a gap that more knowledge can close.
Computational Irreducibility
Stephen Wolfram gave this problem a precise name: computational irreducibility. A system is computationally irreducible when there is no shortcut to determining its future state. The only way to know what the system will do is to let it run. No model, no formula, no amount of processing power can substitute for the system evolving through time, step by step, interaction by interaction.
A simple example: consider a cellular automaton, a grid of cells that turn on or off based on simple rules about their neighbours. Some cellular automata are computationally reducible: you can predict the state of the grid at step one thousand without simulating all nine hundred and ninety-nine preceding steps. The pattern is regular, predictable, shortcuttable.
But other cellular automata, despite having rules just as simple, are computationally irreducible. The only way to know what the grid looks like at step one thousand is to run it for one thousand steps. There is no formula. There is no compression. The computation cannot be simplified. The system is its own fastest simulator.
Complex adaptive systems like markets are computationally irreducible. The interactions between millions of adaptive agents, each responding to the others, each adjusting their behaviour based on what the system is doing, produce outcomes that cannot be determined in advance. Not because the system is random (it is not), but because the computation that produces the outcome is irreducible. The market tomorrow is the result of every interaction between now and then, and no shortcut exists.
This is a hard limit. It is not a practical limitation that technology will eventually overcome. It is a mathematical property of a certain class of systems, and markets belong to that class. The embedded agent who grasps this stops waiting for the model that will finally crack the code, because no such model is possible. Not in principle. Not ever.
The Forecaster and the Embedded Agent
There are, broadly, two kinds of market participants. The distinction is not about strategy or asset class or timeframe. It is about the relationship between the participant and their knowledge.
The forecaster believes that understanding should yield prediction. Understand the economy well enough, and GDP growth becomes foreseeable; understand company fundamentals, and earnings come into focus. Understand market dynamics deeply enough, the reasoning goes, and price itself should become legible. The forecaster’s relationship to knowledge is instrumental: knowledge is a tool for seeing the future, and if the future remains opaque, the solution is more knowledge, better models, deeper analysis.
The embedded agent has a different relationship to knowledge. The embedded agent understands that knowing the rules does not tell you the outcome. Understanding emergence will not let you predict what emerges. Grasp the feedback dynamics all you like; they still will not tell you when the feedback amplifies and when it dampens. And recognising reflexivity reveals nothing about which reflexive loop dominates next. Knowledge changes your relationship to the system without giving you power over it.
This is not anti-intellectual. It is the opposite. It takes the deepest possible understanding of complex systems and follows it to its honest conclusion: these systems are not predictable, and understanding why they are not predictable is itself a form of understanding that matters enormously. The embedded agent is not less informed than the forecaster. The embedded agent is more honestly informed.
The forecaster’s frustration is that the world refuses to yield its secrets. The embedded agent’s insight is that the world’s refusal to yield is itself the secret. The system is structured so that knowing the rules does not determine the outcome. This is not a failure of the knower. It is a feature of the known.
What Understanding Actually Grants
If understanding does not grant prediction, what does it grant? Why bother to understand complex adaptive systems at all, if the understanding cannot tell you what happens next?
Because understanding grants something different from prediction, and what it grants is arguably more valuable.
First, understanding grants orientation. The sailor who understands ocean currents, wind patterns, and storm dynamics cannot predict the weather three weeks from now. But the sailor can read the sky, interpret the swell, recognise the signs of a building storm. The sailor is oriented within the system. Orientation does not eliminate surprise, but it reduces the frequency of being caught completely off guard. The oriented agent is rarely blindsided, even if they are frequently wrong about the specifics.
Second, understanding grants structural alignment. If you know that markets are complex adaptive systems, you build strategies aligned with how such systems behave. You build for fat tails rather than normal distributions, for regime changes rather than stationarity. You assume reflexive feedback, not linear cause and effect. You do not predict what the system will do. You build a vessel suited to the kind of sea you are sailing on.
Third, and most important, understanding gives you a survival architecture. The most important consequence of understanding complexity is not that you trade better in normal conditions. It is that you survive the conditions that destroy those who did not understand. The trader who understands computational irreducibility sizes for surprise. The trader who understands reflexivity builds in circuit breakers. And the one who has absorbed what emergent computation means never bets the portfolio on outsmarting the system. Understanding does not prevent drawdowns. It prevents fatal drawdowns.
Fourth, understanding grants the capacity to be wrong well. This is perhaps the most underappreciated gift. The forecaster who is wrong experiences failure: the model did not work, the prediction was incorrect, the analysis missed something. The embedded agent who is wrong experiences information: the system did something my model did not anticipate, which tells me something about the system’s current state. Being wrong, for the embedded agent, is a data point, not a defeat. It is feedback from a system that is computationally irreducible, and feedback is the only way such a system communicates.
The Navigator and the Stars
Before GPS, before charts, before instruments, Polynesian navigators crossed thousands of miles of open Pacific using only the stars, the swells, the colour of the water, and the flight patterns of birds. They could not predict the ocean. They did not know what the next storm would bring or when the current would shift. But they had something that no prediction could provide: a deep, embodied, continuously updated orientation to the system they were moving through.
They read the ocean the way the embedded agent reads the market. Not with formulas. Not with forecasts. With an intimate, experiential knowledge of how the system tends to behave, what its patterns look like, how the signs relate to the conditions. This knowledge did not eliminate uncertainty. The ocean remained unpredictable. But the navigators’ understanding of the ocean’s dynamics gave them the ability to respond to whatever the ocean did, to adjust course, to find safe harbour, to survive passages that would have killed anyone who attempted them with prediction alone.
The analogy is not romantic. It is structural. The Polynesian navigator and the embedded agent face the same problem: navigating an irreducible system with skill, experience, and structural understanding, knowing that prediction is impossible and that survival depends not on foresight but on the quality of the response.
Understanding the ocean did not let the navigator predict the storm. But understanding the ocean let the navigator survive the storm. And survival, in a system you cannot predict, is the only meaningful form of success.
More Data Will Not Help
The contemporary response to unpredictability is data. More data. Better data. Faster data. The belief is pervasive: if the system is opaque, the solution is more information; if the model fails, more inputs. And when the prediction is wrong, the answer is always a bigger dataset.
In computationally reducible systems, this belief is justified. More data about a mechanical system genuinely improves prediction. More observations of planetary orbits genuinely improve astronomical forecasts. When the system is regular and the uncertainty comes from measurement error, more measurement is the right answer.
In computationally irreducible systems, more data does not resolve the fundamental problem. It can improve short-term forecasts marginally. It can reveal statistical regularities that persist for a while. But it cannot overcome the structural limit imposed by irreducibility, because the system’s future state depends on interactions that have not happened yet, and no amount of historical data captures interactions that have not occurred.
This is why the era of big data has not solved the prediction problem in finance. We have more data than at any point in history: tick-by-tick prices, order book snapshots, satellite imagery, sentiment analysis, alternative data of every conceivable kind. And markets remain stubbornly unpredictable over meaningful horizons. Not because the data is insufficient, but because the system is irreducible. The data describes the past. The future will be computed by a process that includes the data-driven actions of everyone who analysed the past, and those actions will change the outcome in ways the data could not anticipate.
The embedded agent who understands this uses data differently. Not as a window into the future, but as a lens for understanding the present. Not as a prediction engine, but as an orientation tool. The data tells you where you are, what the system has been doing, what patterns have been active. It does not tell you where you are going, because no one and nothing can tell you that.
The Liberating Asymmetry
There is a moment, when the asymmetry of knowing truly lands, that feels like loss. If understanding cannot predict, then what is the point of understanding? If the embedded agent knows the rules and still cannot know the outcome, then is the knowledge hollow?
The answer is the opposite. The knowledge is fuller than prediction ever could be.
Prediction is brittle. A correct prediction succeeds once and must be repeated indefinitely. A wrong prediction fails once and must be recovered from. The forecaster lives on a treadmill: each correct call buys only the obligation to make the next one. The value of prediction is exhausted in its consumption.
Understanding is durable. It does not expire when the next trade is placed. It does not need to be right about specifics to be right about structure. The embedded agent who understands complex adaptive systems carries that understanding across every regime, every crisis, every surprise. It is the foundation on which every adaptation is built, every response calibrated, every recovery structured.
Understanding is also compounding. Each new experience, including each surprise, deepens the agent’s grasp of how the system operates. Being wrong is not a setback but an addition to the experiential database. The embedded agent who has survived a crash, a liquidity crisis, a regime transition, a drawdown that the model did not predict, emerges with a richer understanding than they entered with. The knowledge grows precisely through the experiences that prediction would have tried to avoid.
This is the liberating asymmetry. Understanding does not yield prediction, and that is not a limitation. It is the correct relationship to a system that cannot be predicted. The embedded agent who accepts this asymmetry is freed from the impossible obligation of foresight and granted the sustainable advantage of orientation.
You can know everything about how the weather works. You still cannot tell me if it will rain next Thursday. But you can build a roof. You can carry an umbrella. You can read the sky with an educated eye and adjust your plans with informed flexibility.
That is what understanding grants. Not the power to see the future. The capacity to build for a future you cannot see. And in a computationally irreducible world, that capacity is worth more than any prediction could ever be.
Richard Brennan writes on systematic trading, complex adaptive markets, and the philosophical foundations of trend following at atstradingsolutions.com. His books include The Fractals of Finance and Complex Adaptive Markets. The forthcoming Carved by Impossibility completes the trilogy.
Want the theoretical foundation for why trend following works?
The Fractals of Finance: Determinism, Adaptation and the Geometry of Markets
The book explores the full architecture of feedback, fat tails, and fractal structure in financial markets, and what it means for how we trade, invest, and understand risk.
Available now on Amazon in paperback, hardcover, and Kindle.
Want a practical field manual for trading trends and capturing outliers?
The Aussie Turtles Trend Following Guide: A Field Manual for Hunting Outliers adapts the timeless principles of the original Turtle traders into a systematic, rules-based approach for modern markets. Co-authored with Adam Havryliv.
Available now on Amazon in paperback, hardcover, and Kindle.