The Verdict
The zero is the most important number in finance. Beneath it lies a system built from feedback, shaped by policy, and running hotter than the textbooks allow.
This series began with a number. The near-zero autocorrelation that appears when you compute the serial dependence of daily returns over a long sample. It is the number that launched a thousand assumptions. The efficient market hypothesis, the random walk, the capital asset pricing model: all of them rest on the premise that yesterday’s return tells you nothing about today’s.
Eight episodes later, the evidence is in.
The zero is real. But it is not what anyone thought it was.
The Three Zeros
Figure 9.1 The three zeros. Same structure, three different tests. Left: autocorrelation decomposes into positive and negative feedback. Centre: trend-equity correlation decomposes into bull and bear. Right: variance ratio decomposes into breakout and flat periods.
This series encountered three zeros. The first was the autocorrelation: a full-sample value of negative 0.001 across sixty-eight markets, concealing a system that oscillates between positive 0.035 in positive-feedback periods and negative 0.038 in negative-feedback periods. The second was the trend-equity correlation: a full-sample value of approximately negative 0.16 that decomposed into positive 0.07 during bull markets and negative 0.47 during bear markets. The third was the variance ratio: a full-sample value of 0.910 that decomposed into 1.130 during Donchian breakout periods and 0.282 during flat periods.
Three different statistics. Three different tests. Three different episodes. One spectral structure.
Each zero conceals the same architecture: two opposing forces that cancel when you average across time. Positive and negative feedback produce opposite autocorrelation. Bull and bear markets produce opposite trend-equity correlations. The two spectral states produce opposite variance signatures. The full-sample statistic, in every case, is the weighted average of the two states. It tells you the system’s centre of gravity. It tells you nothing about the system’s operation.
The zero is not the absence of signal. It is the exhaust of a system whose two forces cancel when you look from far enough away. Step closer, and the system is deafening.
The Verdict
Figure 9.2 The three states and the verdict. State 1: dead. State 2: dead. State 3: confirmed. The feedback structure is a permanent feature of how markets process information.
Episode 1 introduced a framework with three possible states. State 1: the structure does not exist, and the zero reflects genuine randomness. State 2: the structure exists but is decaying, converging toward efficiency. State 3: the structure exists and persists, a permanent feature of how markets process information.
State 1 died in Episode 4. Three independent tests, the rolling autocorrelation, the regime-conditional variance ratio, and the trend-equity conditional correlation, all revealed the same spectral structure. The random walk is dead.
State 2 died in Episode 5. The structure’s oscillation amplitude is statistically indistinguishable from where it was forty years ago. The universe mean rose from 0.0491 in the first decade to 0.0558 in the most recent. Six of eight asset classes show higher amplitude now than when the data begins. After four decades of exponential growth in systematic capital, the structure retains its full force. The adaptive markets hypothesis predicted convergence. The data shows none.
State 3 stands. The feedback structure is a persistent feature of financial markets. It is not an anomaly being arbitraged away. It is the mechanism by which prices discover value through the alternating cycle of overshoot and correction, positive feedback and negative feedback.
This is the verdict: markets are not random. They are not converging toward randomness. They are structured systems running a feedback cycle that has survived four decades of adaptive pressure without measurable decay. The zero is the centre of gravity for a system operating far from equilibrium.
The Architecture
Figure 9.3 Eight asset classes, eight spectral configurations. Duration ratio, intensity gap, positive-feedback fraction, and resulting full-sample VR. No two asset classes run the same way.
The feedback structure runs in every market, but it does not run the same way everywhere. Episode 6 revealed the fingerprint: each asset class has a distinct spectral configuration defined by its duration asymmetry, force intensity, positive-feedback fraction, and variance ratio gap.
Livestock run the longest positive-feedback episodes relative to their negative-feedback episodes: a duration ratio of 2.2x. Grains follow at 2.0x. Equities are symmetric at 1.0x. FX and energy show negative-feedback dominance at 0.5x and 0.6x respectively.
These are not cosmetic differences. They are instructions. A divergent strategy that applies a single lookback uniformly to livestock and equities is ignoring a fundamental difference in the structure’s natural frequency. The fingerprint is the structure’s instruction manual, and most strategies ignore it.
The eight spectra run largely independently, with low cross-class regime correlation. Equities are essentially uncorrelated with every other class. When equities shift toward oscillation, commodities may be trending. This asynchronous timing creates a diversification benefit that is deeper than conventional return diversification: it is spectral-state diversification.
The Paradox and Its Resolution
Figure 9.4 The four forces. Regime is cyclical and dominant. Short-side impairment is structural but overstated by symmetric models. Horizon crowding is a redistribution. The escalator/elevator asymmetry requires independent calibration. The edge is not dying. It has relocated.
Episode 7 confronted the paradox at the heart of this series. The feedback structure persists. Simple trend-following returns do not. Over the full sample, trend delivers a MAR of 0.54 versus the S&P 500’s 0.08: nearly seven times as much return per unit of worst-case pain. But within this full-sample picture, decade-level performance has been declining.
The resolution required testing whether this decline was structural or regime-specific. The answer is both, but not in equal measure.
The dominant force is regime. The variance ratio itself varies by macro environment. During the 2020 to 2022 inflation, the twenty-day VR rose to 1.042. Every divergent strategy recovered to pre-QE quality during this period. The lost decade from 2009 to 2020 was primarily a VR drought. When the macro environment provided genuine trends, the edge returned.
The second force is structural: short-side impairment. Central bank backstops, passive index flows, and systematic dip-buying have severely impaired short-side standalone alpha. The long side dipped and recovered. The short side collapsed under symmetric models. However, as Episode 8 demonstrated, asymmetric calibration reduces the damage. Short-side portfolio utility, the crisis alpha, the negative bear-market correlation, the convex tail payoff, remains mechanistically intact.
The third force is the crowding gradient. The twenty-day Donchian is dead in every regime. The 300-day Donchian delivered a Sharpe of 0.74 post-2020. The edge has redistributed from shorter to longer horizons, and an ensemble that spans both provides diversification benefit that is increasing, not decreasing.
The QE era imposed all four forces simultaneously. Decade-level bucketing conflated the effects and produced the narrative of monotonic structural decline. The data tells a more honest story: one force is cyclical, one is structural but overstated by symmetric lenses, one is a redistribution, and one is a frequency mismatch requiring independent calibration. The edge is not dying. It has relocated.
What the Structure Means
If you manage money, allocate capital, or think about how markets work, here is what this series has established.
First, the random walk is not a reasonable model for the vast majority of traded markets. The zero that anchors modern finance is an artifact of averaging over a structured system, not evidence that the system is unstructured.
Second, the feedback structure is not an anomaly to be exploited. It is the process by which markets discover value. Positive feedback is the mechanism by which new information is incorporated: prices overshoot as heterogeneous agents update their beliefs at different speeds. Negative feedback is the correction: prices oscillate as the overshoot resolves. You cannot arbitrage away price discovery. You can only participate in it.
Third, the structure’s output is regime-dependent. When macro conditions provide genuine trends, the variance ratio rises and divergent strategies work. When conditions suppress trends, the VR compresses and divergent strategies struggle. The structure is always running. Its output varies with the fuel supply.
Fourth, the short side of trend-following may be permanently diminished. Central bank intervention and passive flows have structurally changed how markets behave during downtrends.
Fifth, the fingerprint provides an edge that does not decay with crowding. A strategy that calibrates its lookback to each asset class’s duration asymmetry, monitors the variance ratio for regime shifts, and diversifies across the eight independent spectral configurations will outperform a uniform approach. The fingerprint is the structure’s instruction manual. The book this series is drawn from provides the framework for reading it.
The Arc
Figure 9.5 The full arc. Act I revealed. Act II proved. Act III delivered the verdict.
The series traced an arc from surface to depth.
Act I revealed the feedback structure. Episode 1 showed that the near-zero autocorrelation conceals structured oscillation between positive and negative feedback. Episode 2 mapped that oscillation across sixty-eight markets, introduced the spectral coupling framework, and showed that the coupling is permanent. Episode 3 demonstrated that the coupling mechanism produces the crisis alpha that defines trend-following’s value to equity portfolios.
Act II proved it. Episode 4 confirmed the structure through the variance ratio, a completely independent test, and eliminated State 1. Episode 5 showed that the structure has not decayed despite four decades of adaptive pressure, eliminating State 2. Episode 6 revealed that each asset class runs a distinct spectral configuration, the fingerprint.
Act III resolved the paradox. Episode 7 confronted the declining strategy returns and resolved them into four separable forces. Episode 8 demonstrated that bull and bear markets operate at different frequencies, requiring asymmetric calibration. Episode 9 is this episode. The verdict.
The zero is the most important number in finance. It is not what anyone thought it was. It is not evidence of randomness. It is not the absence of signal. It is the equilibrium point of a far-from-equilibrium system, the centre of gravity for a structure that never stops running. Beneath every zero in this series lies the same feedback system, measured from a different angle. The system is everywhere. It is not decaying. And understanding its architecture is the difference between trading it and being traded by it.
Endnotes
- Three zeros synthesis: (a) Full-sample ACF(1) across 68 contracts: mean = −0.001. Regime-conditional: positive-feedback mean ACF = +0.035, negative-feedback mean ACF = −0.038. Source: Episode 1, rolling 504-day windows. (b) Trend-equity correlation: full-sample −0.16 (200-day MA strategy vs S&P 500). Bull-market conditional: +0.07. Bear-market conditional: −0.47. Source: Episode 3. (c) Variance ratio at 3-month horizon: full-sample 0.910. Donchian breakout VR: 1.130. Flat VR: 0.282. Gap: 0.847. Source: Episode 4, Donchian(20,mid) classification.
- State 1 elimination: three independent tests across Episodes 1, 3, and 4. (a) Rolling ACF: structured oscillation in all 68 contracts. (b) Conditional correlation: trend-equity correlation shifts from +0.07 to −0.47 across market regimes. (c) Regime-conditional VR: breakout VR 1.130 > 1.0, flat VR 0.282 < 1.0.
- State 2 elimination: Episode 5 compared ACF oscillation amplitude across four decades. Universe means: D1 = 0.0491 (n=55), D2 = 0.0537 (n=67), D3 = 0.0532 (n=68), D4 = 0.0558 (n=68). Paired t-test (D1 vs D4, n=55): p = 0.27. No significant decline. 34 of 55 paired contracts (62%) show higher amplitude in the most recent decade.
- Four-force decomposition from Episodes 7–8. (a) Regime: universe-mean VR(20) ranges from 0.837 to 1.042. Annual VR(20) correlates with MA 200d annual Sharpe at r = 0.38. (b) Short-side impairment: MA 200d long-side SR: 1.49/0.59/0.67. Short-side: 0.76/−0.08/0.05. Return contribution shifted from 69/31 to 93/7. (c) Crowding gradient: DON 300d post-2020 SR = 0.74, DON 20d = 0.17. Cross-horizon correlation declining. (d) Escalator/elevator: asymmetric grid search confirms shorter short lookbacks dominate.
- Full-sample MAR ratios: Trend (200-day MA, 68 contracts, vol-scaled): annualised return 3.3%, max drawdown 6.0%, MAR 0.54. S&P 500 buy-and-hold: annualised return 4.5%, max drawdown 54.2%, MAR 0.08.
- Same dataset as Episodes 1 through 8. 68 CSI ratio-adjusted continuous futures, 8 asset classes, September 1984 to January 2026.
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.