Before You Begin
The context, the language, and the question that structures everything that follows.
What This Series Is
This is a nine-episode investigation into how financial markets actually behave, tested against how they are supposed to behave. The investigation uses four decades of daily price data from sixty-eight futures contracts spanning eight asset classes: equities, bonds, currencies, energy, metals, grains, livestock, and softs.
The central claim is that markets are not random. The most quoted statistic in quantitative finance, the autocorrelation of daily returns, averages to zero across our dataset. This number underwrites the efficient market hypothesis, the random walk, and most of modern portfolio theory. But an average can conceal as much as it reveals. Beneath that zero lies a structured system driven by two opposing forces: positive feedback, which makes price moves persist, and negative feedback, which makes them oscillate. These forces produce a spectrum of market behaviour, from trending at one end to oscillating at the other, with a narrow transitional zone in the middle where the forces roughly cancel.
The mechanism behind this structure is singular. The work of Jean-Philippe Bouchaud and his collaborators established that the only force that moves price is the physical act of trading itself. Every other theory of price movement, whether based on fundamentals, sentiment, or momentum, is ultimately a theory about why people trade. The price itself moves only through transactions. This series extends that insight from microstructure to macro behaviour, asking what happens when you scale agent impact across all participants simultaneously. The answer is the spectrum we document across nine episodes.
This system has survived four decades of exponential growth in systematic trading capital without measurable decay. The series builds the evidence progressively. Each episode adds one layer. By Episode 9, the case is either standing or it is not. There is no hand-waving.
Where This Starts
This series is Phase 2 of a larger research programme. Phase 1 established that futures markets carry a statistical fingerprint that the random walk forbids: persistent feedback in daily returns that shows up in every market tested. Phase 1 proved the fingerprint exists. Phase 2 asks what it does.
You do not need to have read Phase 1. Everything you need is contained in the nine episodes that follow. But if you encounter a reference to Phase 1 in Episode 1, that is the context: earlier work proved markets have memory. This series investigates the nature of that memory, whether it is decaying, and what it means for anyone who trades or allocates capital.
What Is Known and What Is New
Three things in this series are established knowledge. Every quantitative researcher in the field knows them.
First, that daily returns carry weak but persistent autocorrelation. Lo and MacKinlay documented it in 1988. Thousands of papers have confirmed it since.
Second, that trend-following strategies deliver negative correlation with equities during bear markets. The managed futures industry calls this crisis alpha. It has been documented extensively.
Third, that the variance ratio test rejects the random walk. This has been known since the original Lo and MacKinlay paper and is not news to any practitioner.
What is new in this series begins in Episode 2 and culminates in Episode 9. The specific findings that we believe to be novel are: that markets are permanently coupled along a spectral axis from trending to oscillating behaviour, and that this coupling is permanent because it arises from diversification itself, which guarantees that agents apply impact across multiple instruments as part of a single investment process; that the transition between spectral states follows an earthquake-like pattern with rupture, aftershock, and settling phases; that cross-sectional dispersion is constant across all spectral states, proving the coupling never switches off; that strategy performance is structural alignment between strategy design (divergent or convergent) and the market’s spectral position; that crisis alpha is one expression of a permanent coupling mechanism, not a special property of crises, with the same mechanism producing additive returns during bull markets; and that the feedback operates asymmetrically between long and short positions due to the mechanistic structure of how agents are forced to act under stress, an asymmetry that the rise of passive investing has intensified.
We acknowledge that we may be unaware of prior work that touches on aspects of these findings. Science builds incrementally. If we have missed relevant precedents, we welcome corrections and will update the record accordingly.
How to Read the Figures
Each episode contains figures referenced by number. The figures are separate image files designed to be viewed alongside the text. Each figure caption describes what you are looking at and what it means. The series was designed so that a reader who examines only the figures and their captions would still grasp the core argument.
The Language
This series uses specific terms that may carry different connotations in everyday usage. Here is what they mean in this context.
Positive feedback does not mean good feedback. It means a process that amplifies. In a market, positive feedback means that a price move in one direction increases the probability of further moves in the same direction. The result is a trend: prices that persist in one direction, whether up or down. Think of a microphone placed too close to a speaker. The sound feeds back into itself and grows.
Negative feedback does not mean bad feedback. It means a process that dampens. In a market, negative feedback means that a price move in one direction increases the probability of a move in the opposite direction. The result is oscillation: prices that chop back and forth within a range, covering ground but not making progress in any direction. Think of a thermostat. When the temperature rises above the set point, the system pushes it back down. When it falls below, the system pushes it back up. The result is a range, not a trend.
The spectrum is the continuous range between these two forces. At one end, positive feedback dominates. At the other, negative feedback dominates. In the middle lies a narrow transitional zone where the forces roughly balance. This zone is not a third state. It is a crossover between the other two, a point of shifting dominance rather than an absence of activity. A market’s spectral position tells you which force is currently winning.
Mean-reversion in this series does not mean a sharp move in the opposite direction, the way the term is sometimes used colloquially. It means oscillation. A mean-reverting market is one where prices chop back and forth, with each move partially cancelled by the next. It is high energy with no sustained direction. The distinction matters because the mental image of a “snap back” is misleading. Mean-reversion is not a countertrend. Countertrend trading is directional in nature, requiring a trend to exist in order to trade against it. Mean reversion is range-bound. They belong to different families.
Autocorrelation measures whether today’s price move is related to yesterday’s. Positive autocorrelation means today’s move tends to continue yesterday’s direction (trending). Negative autocorrelation means today’s move tends to reverse yesterday’s direction (oscillating). Zero autocorrelation means no relationship (random). This is the primary measurement tool throughout the series.
The variance ratio measures whether multi-day price swings are bigger or smaller than single-day swings would predict. If prices are random, a five-day swing should be predictable from the size of daily swings, in the same way that flipping a coin fifty times produces a predictable range of outcomes from flipping it ten times. A ratio above one means trends are building. A ratio below one means moves are cancelling out.
Divergent strategies are those that push prices further from where they started: trend-following, momentum, breakout trading. They profit when moves persist. Countertrend strategies also belong to this family because they are directional, trading against the prevailing move.
Convergent strategies are those that push prices back toward where they started: value investing, mean-reversion trading, contrarian positioning. They profit when moves reverse toward a central value.
MAR ratio is the performance metric used throughout this series. It divides the annualised return of a strategy by its worst peak-to-trough drawdown. Unlike the more commonly used Sharpe ratio, which penalises upside volatility equally with downside, the MAR ratio measures what practitioners actually care about: how much return did you earn for the worst pain you had to endure? For a strategy like trend-following, which generates many small losses and occasional large gains, the Sharpe ratio systematically understates quality. The MAR ratio does not. A trend strategy with a MAR of 0.53 earns 53 cents of annual return for every dollar of maximum drawdown. The S&P 500 over the same period delivers a MAR of 0.07.
The Three States
The series tests three possible states for the structure we measure.
State 1: the structure does not exist. The near-zero autocorrelation reflects genuine randomness. The rolling windows, the regime shifts, the conditional correlations are all artefacts of measurement. The efficient market hypothesis is correct. This is the null hypothesis.
State 2: the structure exists but is decaying. Markets once carried exploitable patterns, but four decades of algorithmic trading, quantitative finance, and massive capital deployment have gradually eroded them. The system is converging toward efficiency. The zero is not here yet, but it is coming.
State 3: the structure exists and persists. It is not an anomaly being arbitraged away. It is a permanent feature of how markets process information through the alternating cycle of positive and negative feedback.
By Episode 5, the first two states are dead. What remains is the third. The rest of the series explores what it means.
The Arc
The nine episodes follow a three-act structure. Each episode asks one question and builds on the answer from the last. Together they form a single, cumulative case.
Act I: The Revelation
Episode 1 begins with the most trusted number in finance and shows that it is a lie. The near-zero full-sample autocorrelation conceals a machine. Rolling windows reveal structured oscillation between trending and mean-reverting behaviour in every market. The zero is not the absence of signal. It is the exhaust of a system whose two forces cancel when you average across time. If the average conceals this much structure in a single market, what happens when you look at all sixty-eight simultaneously?
Episode 2 answers that question, and the answer changes everything. Markets are not running sixty-eight independent engines. They are permanently coupled along a single spectrum. The heatmap that reveals this coupling is the most important figure in the series. It raises a question that Episode 2 cannot answer alone: if the coupling is real, what does it produce?
Episode 3 delivers the consequence. The permanently coupled spectrum is the mechanism that creates crisis alpha. The correlation between trend-following and equities is not constant. It flips sign and deepens with the severity of the drawdown. The spectral framework explains the flip. But the framework also predicts something the industry has never articulated: that the same mechanism should produce additive returns during bull markets. It does.
Act II: The Proof
Episode 4 changes the standard of evidence. Revelation is not proof. The variance ratio provides independent confirmation of the spectrum through a completely different test. We decompose it by spectral state, something nobody has done before. The result eliminates State 1 with finality. The random walk is dead. But a deeper question remains: is the structure decaying?
Episode 5 kills State 2. After four decades of exponential growth in systematic trading capital, the oscillation amplitude is statistically unchanged. Six of eight asset classes show higher amplitude in the most recent decade than the first. The adaptive markets hypothesis predicted convergence. The data shows none. The structure persists. But if the structure persists, why have simple trend-following returns declined?
Episode 6 reveals that the two forces are not symmetric. Positive feedback is gentle and sustained. Negative feedback is intense and brief. The ratio between them varies by a factor of six across asset classes. This fingerprint is the key to understanding why the same engine produces different results in different markets.
Episode 7 confronts the paradox directly. The engine persists. The simplest harvest does not. Four forces explain the decline, and the dominant force is not what the industry assumes. The resolution changes how you think about the future of trend-following. But one of those forces, the impairment of the short side, raises a question that Episode 7 cannot resolve.
Episode 8 investigates whether the short side is genuinely dead or invisible to a symmetric lens. Bull trends are slow escalators. Bear trends are fast elevators. A symmetric lookback is structurally misaligned with the faster downside dynamics. The resolution has practical consequences for every trend-following portfolio.
Act III: The Verdict
Episode 9 delivers the verdict. Three zeros. Three tests. One spectral structure. The case is closed. What remains is the framework for building on it.
This research series is drawn from 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.