Click on the image above to listen to the podcast
Why Markets Can’t Stop Trending
In this episode of the Systematic Investor series, Niels and I dig into a question that sounds simple but unravels almost everything the textbooks teach: why do markets trend at all, and why can’t they seem to stop? I make the case that markets aren’t equilibrium machines edging toward fair value. They’re complex adaptive systems, closer to a flock of starlings than a clock, where structure emerges from the interactions between participants rather than from any single decision or piece of news.
From there we follow the thread through the rise of passive investing and what it’s quietly done to the balance of the market, the endogenous engine that drives most price movement, the inelasticity that turns a dollar of flow into five, and the empirical fingerprint that shows up across 68 markets and four decades. Niels keeps me honest throughout, pushing for the plain-English version at every turn. The conclusion we land on is a stark one: on the evidence, markets can’t stop trending, and the architecture that makes it so is hardening, not fading.
A flock, not a clock
Picture a murmuration of starlings over a winter field in the failing light. Thousands of birds moving as one, turning, splitting, recombining into shapes of extraordinary complexity. No bird leads. No bird holds the full picture. There is no choreographer and no plan. Each bird is simply running three local rules: stay close to your neighbours, match their speed, avoid collision. From those local rules, applied in parallel, a coherent structure emerges that lives in the interactions between the birds, not in any one of them.
That image is where I want to begin, because markets work the same way, and once you see it, you can’t unsee it.
For most of the last fifty years, finance has taught markets as the wrong kind of object. The dominant framework, the one embedded in every MBA syllabus, every institutional allocation model, every textbook, treats markets as if they were governed by the same laws as physical systems. Particles in motion. Forces pushing prices toward a stable equilibrium. Deviations from fair value dismissed as short-lived, unprofitable noise. It is elegant and mathematically tractable. And as a description of how real markets behave, it is the wrong picture entirely.
The equilibrium model treats a market like a clock. The parts of a clock don’t observe one another. The pendulum doesn’t change its behaviour based on what the gears are doing. The springs don’t panic when the hands move too fast. But market participants observe each other constantly. They imitate, they panic when others panic, they anchor to reference points others have just set, they build models of what everyone else will do, and acting on those models changes what everyone else does. The agents in a market are not particles. They are responsive. And that responsiveness is the entire story.
Treating a flock as if it were a clock is a category error. And the consequences of that error show up everywhere in how markets are modelled and explained.
The question that actually matters
A market is not a reflection of the underlying economy. It is not a representation of fundamental value, however you choose to measure it. It is constituted by the interactions of a population of agents, each holding a partial model, each acting on it, each changing the conditions the others observe. The economy, the banking system, the firm: these are loose, autonomous systems that correlate with the market and constrain it, but they are not antecedent to it. The market is its own thing.
So the most important question you can ask about a market is not what is its fair value, or what the central bank will do. It is this: what kinds of agents populate this system, and what are the properties of their interactions?
The property that matters most is whether an agent conditions its behaviour on price.
Some agents do, and they push back. The value investor who fades a stretched valuation. The counter-trend trader. The market maker fading extremes. The pension fund rebalancing out of an asset that has run too far. These are the balancing agents, convergent in the language I use in my own practice, and they are what stops reinforcing loops from running unchecked.
Other agents are conditioned on price in motion, and they amplify. The momentum algorithm that buys what is rising. The volatility-targeting fund that cuts exposure as volatility climbs. The stop-loss that fires at a threshold. These are the reinforcing, divergent agents, the fuel of trends.
And then there is a third category that has quietly become the dominant force of the last two decades.
What passive did to the ecology
A passive vehicle buys what the index tells it to buy, in the proportion the index dictates, at whatever price the market happens to offer. Price is not an input to its decision function. It does not buy more when something is cheap, or less when it is dear, or step in when a valuation gets crushed. It simply executes its inflows in proportion to the index’s current composition.
Passive is not price-sensitive. And because it is not price-sensitive, it does not participate in price discovery at all.
Let me be precise about the argument, because it is easy to mishear. At the level of the individual investor, owning a low-cost index fund is, for most people, a perfectly rational decision. I am not arguing against that. What I am making is a structural argument about what happens to the agent ecology when a growing share of capital becomes price-insensitive by design. Both things can be true at once: index funds can be a sensible choice for the individual, and the aggregate consequence of millions of those sensible choices can change the structural behaviour of the whole system in ways no individual choice ever anticipated.
Here is what follows. Passive has not eliminated the balancing agents. The value investor, the counter-trend trader, the market maker all still exist. What it has done is dilute their share of the ecology. The reinforcing side, meanwhile, hasn’t thinned at all; the forced flows around index inclusion and rebalancing have grown. So the population that pushes back against price has shrunk relative to the population that pushes price further along.
The structural consequence: reinforcing loops now run further before they meet resistance, because the resistance is structurally smaller. The system is more prone to extended directional moves, more prone to overshoot, more prone to cascade. Not because the reinforcing forces have grown stronger, but because the balancing forces have grown weaker.
The endogenous engine
There is a deep assumption buried in the equilibrium model: that price moves because news arrives. Information enters, rational agents process it, price adjusts to the new reality. Clean, linear, cause then effect. Every move has a tidy story, and that is the story the financial channels tell every day.
Twenty years of microstructure research, from the likes of Bouchaud, Gabaix and others, keeps converging on the same uncomfortable finding: the vast majority of price movement cannot be attributed to identifiable news. Prices move on days with no news. They move in ways bearing no proportional relationship to the news that does arrive. Some of the largest moves in market history happened with no catalyst of equivalent magnitude.
If news isn’t the primary driver, what is? What I call the endogenous engine, the system processing itself. A small breakout lifts price. The lift draws in momentum systems that buy what is rising. They alter volatility. The volatility change triggers volatility-targeting and rebalancing. The rebalancing shifts positioning across portfolios, which forces further price adjustment. Each step happens because of the price action, not because of any external information. The original spark, news or just an order imbalance, becomes increasingly irrelevant as the cascade develops. The trigger is the match; the system is the fuel. The fire’s shape and extent depend on the dryness of the wood, not the size of the match.
This is reflexivity, in the sense Soros had right decades ago. Price-sensitive participants don’t simply observe markets and react; their reactions reshape the market, which reshapes the next observation, which reshapes the next reaction. The loop runs continuously. Passive participants, importantly, are present but not in the loop. They hold their shares without driving the recursive cycle.
When a dollar moves five
There is one more amplifier worth planting. Research on market inelasticity has produced a striking finding. When the bulk of shares sit with price-insensitive holders, the effective pool available to absorb active buying or selling is structurally thin. An active buyer or seller must find the other side of the trade within a much smaller fraction of the float, the fraction still responding to price.
The counterintuitive result: each unit of active flow has to move price further to find a willing counterparty. A dollar of net buying doesn’t lift market capitalisation by a dollar; it lifts it by a multiple. The best work in this space puts the multiplier on the order of five, with the literature ranging from roughly three at the conservative end to closer to seven at the aggressive. The exact number matters less than the structural fact: the multiplier is large, far larger than classical finance assumed, and it is a direct consequence of the agent ecology. Fewer price-sensitive players in the active mix means more violent price action per unit of flow, not less. The inelasticity ensures that the trends the engine produces are larger than the flows that created them.
The fingerprint
None of this is theory. A colleague and I built a deliberately minimal agent-based model of a market with just two kinds of agent: Chartists, the divergent traders conditioned on recent price action, and fundamentalists, the convergent traders who push back when price strays from value. One knob: the proportion of Chartists, dialled from zero to one hundred per cent.
With zero Chartists, the simulated market behaved exactly as classical finance insists real markets should: independent returns, Gaussian tails, no memory, no volatility clustering. The Hurst exponent, a measure of memory in a time series, sat at 0.54, statistically indistinguishable from a random walk. Excess kurtosis, a measure of fat tails, was negative. Five-sigma events: zero. A world that has never actually existed.
Then we turned the knob. The behaviour didn’t drift gradually; it changed through a phase transition, like water becoming steam. And the threshold was not a fifty-fifty balance. At roughly twenty-five per cent Chartists, everything changed at once: kurtosis erupted, the Hurst exponent leapt above 0.85, memory appeared, fat tails appeared, volatility clustering appeared. Every statistical signature of a real market, switched on at a single critical boundary.
Above that threshold, the simulated market became statistically indistinguishable from sixty-eight real futures markets across eight asset classes, six continents and forty-one years of history. The same fingerprint, produced by feedback alone, with no news, no central banks, no geopolitics, no fundamentals. Feedback isn’t a minor mechanism at the edge of market behaviour. It is the central mechanism.
And twenty-five per cent is not a high bar. Once you set passive aside as price-insensitive, virtually every remaining active participant is price-sensitive to some degree: stop-losses, margin calls, volatility-targeting, options hedging, performance-chasing are all feedback. Real markets sit far above the threshold, and they have for decades. Across rolling five-year windows over forty years, the Hurst exponent never once fell to the 0.5 baseline. It ranged between 0.62 and 0.82, surviving Black Monday, the move to electronic trading, the global financial crisis, the rise of high-frequency trading and the COVID crash. There is no era of available market data in which markets behaved the way the equilibrium model said they should. That world simply doesn’t exist in the historical record.
One final test. We asked whether those sixty-eight markets were behaving as sixty-eight independent systems or as one coupled engine. If independent, the fraction sharing the majority feedback sign should have averaged around 0.55. It averaged 0.6 persistently across the whole sample, and in the most coordinated regimes ran as high as 0.82. The reason isn’t mysterious: the same investors run the same models across the same markets. A buy signal in one market often fires a buy signal in another. The world’s futures markets are not sixty-eight independent feedback engines; they are one permanently coupled feedback engine with different surface manifestations across asset classes.
What this means for the craft
If feedback-driven behaviour has been present for at least forty years, never weakened, appearing in every market on every continent across every asset class, then what trend following has been doing for those forty years is not exploiting a temporary inefficiency or catching a window that might close. It is harvesting a structural feature of how markets actually work. The directional moves trend followers capture are the natural output of the endogenous engine running through the feedback architecture. They are the murmuration in motion.
And given what passive is doing, diluting the balancing agents while leaving the reinforcing ones intact, the architecture is not weakening. It is hardening. The float is thinner, the multiplier on flow has grown, and the system is producing the same fingerprint with more violence per unit of flow than it did twenty years ago.
So when Niels put it to me plainly, I gave him the plain answer. Markets can’t stop trending. And the evidence points to it getting stronger, not weaker, from here.
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.