Wealth is not created by the return you earn in any single year. It is created by the shape of the path your capital travels across all of them. This series exists because the mathematics of that path has been misunderstood, mistaught, and misapplied for decades, and the cost of that misunderstanding compounds silently inside every portfolio that ignores it.
The Compounding Problem
Every investor knows that compounding matters. It is the most cited concept in personal finance, invoked in every introductory textbook, every retirement calculator, every pitch deck. Albert Einstein may or may not have called it the eighth wonder of the world, but the attribution endures because the intuition feels right. A dollar invested at 8% doubles in nine years; left alone for forty years, it becomes twenty-two dollars. The arithmetic is elementary. The implications are not.
The reason compounding dominates long-term wealth creation is not merely that returns accumulate. It is that returns multiply. Each year’s gain is earned not on the original capital but on the original capital plus every prior gain. This is the multiplicative nature of investing, and it changes everything about how portfolios should be evaluated, constructed, and defended. In a multiplicative system, the sequence of returns matters as much as their average. The path matters as much as the destination. And the shape of that path, its geometry, determines how much of the theoretical return actually arrives in the investor’s account.
This distinction between theoretical and realised return is the central problem of investing. The finance industry overwhelmingly uses arithmetic averages, Gaussian assumptions, and the Sharpe ratio to evaluate performance, tools that assume a world of additive, normally distributed returns. But the world the investor actually inhabits is multiplicative, fat-tailed, and path-dependent. The gap between these two worlds is not a rounding error. It is the difference between the return that is advertised and the wealth that is actually created. A portfolio that loses 50% and then gains 50% has not broken even; it has lost 25%. The arithmetic average return is zero. The geometric reality is destruction. This is the compounding problem, and it sits at the heart of everything this series investigates.
The volatility drag, the mathematical tax that rough paths impose on compound growth, is the mechanism through which this problem operates. Two portfolios with identical arithmetic returns but different volatilities produce different terminal wealth, and the smoother path always wins. This is not an approximation. It is an identity: the geometric return is approximately the arithmetic return minus half the variance, and the consequences of that subtraction compound across every year of an investor’s life.
Why This Series Exists
We wrote The Geometry of Wealth because the investment industry has a measurement problem that hides in plain sight. The tools most widely used to evaluate portfolios were designed for a world that does not exist: a world of bell curves, stable correlations, and returns that arrive smoothly. In the real world, returns arrive in clusters, crashes are larger and more frequent than Gaussian models predict, and the months that matter most, the handful of extreme events that determine terminal wealth, are precisely the months that conventional tools handle worst.
Trend following, as a systematic investment process, is designed for the world that actually exists. It cuts losses to limit left-tail damage, lets winners run to capture right-tail opportunity, and operates across dozens of uncorrelated markets to diversify the path. These properties produce a return stream with a specific geometric signature: positive skewness, negative equity correlation, contained drawdowns, and crisis alpha. Each of these properties, individually, improves compound growth. Together, they create a return stream that is geometrically more efficient than conventional equity investing, even when its arithmetic return is lower. This is counterintuitive. It violates the assumption that higher returns always produce more wealth, and it cannot be understood through the Sharpe ratio, which penalises upside volatility as heavily as downside volatility and is therefore structurally biased against positively skewed return streams. Understanding why requires a different lens, the geometric lens, and that lens is what this series builds, one episode at a time, from first principles through to portfolio construction.
The series draws on 26 years of performance data from the NilssonHedge Trend Following database (January 2000 to January 2026). The analysis universe is the TTU TF Index, a set of 47 programs with track records of at least 15 years, with per-program exhibits drawn from the 38 programs that carry 20-year-plus records over the period. All returns are net of management and performance fees. Extended analysis uses individual manager records reaching back to the mid-1980s, spanning 39 years and six major equity crises. The data is real, the costs are included, and the limitations are stated openly.
The Architecture of the Series
The Geometry of Wealth is structured in four acts. Each builds on the last, and the cumulative argument is designed to be stronger than any individual episode. The reader who has followed it to the end should feel not that they have been persuaded, but that they have been shown something they can now verify for themselves.
Act I: The Broken Mathematics (Episodes 1 to 4)
The first act establishes the mathematical foundation by dismantling the tools the investment industry relies on. It demonstrates that arithmetic averages overstate compounded returns, that drawdowns impose an asymmetric cost on wealth that most investors dramatically underestimate, that the Sharpe ratio penalises exactly the return shapes that compound most efficiently, and that the Gaussian framework underpinning modern portfolio theory fails to account for the fat-tailed, crisis-prone markets where wealth is actually won or lost.
Episode 1: The Lie of the Average
The arithmetic average is the most dangerous number in finance. This episode shows how it systematically overstates the return an investor actually receives, introduces the geometric return as the true measure of compound growth, and shows that the gap between the two, the volatility drag, explains why two portfolios with identical average returns can produce wildly different terminal wealth. The S&P 500’s arithmetic return of 8.89% delivers a geometric return of just 8.02%, and that 0.87 percentage point gap, compounded over 26 years, is the cost of ignoring the path.
Losses and gains are not symmetric. A 50% loss requires a 100% gain to recover; a 75% drawdown requires a 300% gain. This episode maps the mathematics of drawdown recovery and its devastating implication: the deeper the drawdown, the longer the compounding interruption, and time lost to recovery is the most expensive cost in a multiplicative system. The S&P 500’s two major drawdowns consumed more than a decade of compounding time, and the portfolio that avoids even a portion of that damage compounds from a higher base for every subsequent year.
The Sharpe ratio is the most trusted and most misleading metric in portfolio evaluation. By treating upside and downside volatility as equally dangerous, it penalises return streams with positive skewness, exactly the shape that geometric compounding rewards. This episode introduces the MAR ratio (CAGR divided by maximum drawdown) as the metric that measures what matters: how much wealth is created per unit of geometric risk. Ranking the analyzed programs by Sharpe versus MAR reveals managers whose true compounding power is hidden by the conventional metric.
Episode 4: The Smooth World That Never Was
Modern portfolio theory assumes returns are normally distributed. They are not. This episode confronts the fat-tailed reality of financial markets, showing that extreme events occur far more frequently than Gaussian models predict, and that a small number of extreme months dominate terminal wealth while arriving far more reliably than any bell-curve model would suggest. A portfolio designed for the smooth world is undefended in the real one.
Act II: The Mechanism (Episodes 5 to 8)
The second act explains how trend following works as a geometric compounding engine, tracing the mechanism from its structural origins in complex adaptive systems, through the two operational components of cutting losses and letting winners run, to the crisis alpha that transforms portfolio-level compounding during the months that matter most.
Episode 5: Why Markets Trend (And Why They Always Will)
Markets trend because they are complex adaptive systems populated by agents with heterogeneous beliefs, behavioural biases, and institutional constraints. Drawing on complexity science and agent-based modelling, this episode argues that feedback among price-sensitive participants reproduces the statistical signature of real markets, fat tails, memory, and volatility clustering, and that the trend-following edge persists because it harvests a structural property of any market populated by price-sensitive agents, not a temporary inefficiency that capital can arbitrage to zero.
Episode 6: The Cut Is the Geometry
The systematic cut of losing positions is the single most important geometric operation in trend following. By truncating the left tail of the return distribution, the cut converts a symmetric return stream into one with positive skewness. This episode quantifies the geometric value of truncation: fewer deep drawdowns, faster recovery, less compounding time lost. The cut does not improve the arithmetic average. It improves the shape of the path, and in a multiplicative system, the shape of the path determines terminal wealth.
Episode 7: Letting Winners Run and Harvesting the Right Tail
The complement of the cut is the hold: letting winning positions run to capture the full extent of trending moves. This episode examines the right tail and shows that a disproportionate share of terminal wealth is generated by a small number of outsized winning months. The process does not predict which months will produce these gains; it simply remains positioned to capture them. The asymmetry between capped losses and uncapped gains is the geometric engine that produces positive skewness across the population of trend followers.
Episode 8: Crisis Alpha: Compounding When Others Are Destroyed
Crisis alpha is not insurance against equity declines. It is the same coupled system trending downward during crises that trends upward during bull markets, with trend following capturing the directional energy in both regimes. The geometric value extends beyond the crisis months themselves: by compounding from a higher base when equity portfolios are still recovering from deep drawdowns, the trend-following allocation earns returns on capital the equity investor no longer has. Across the three major crises since 2000, the TF Index averaged +38.6% while the S&P 500 averaged -38.1%.
Act III: The Evidence (Episodes 9 to 13)
The third act deploys 26 years of real performance data, all net of fees, to test the geometric thesis at the portfolio level, the individual-program level, and across time. It then confronts the costs honestly and builds the evaluation toolkit that replaces the Sharpe ratio.
Episode 9: The Portfolio: Where Geometry Becomes Architecture
This episode constructs the blended portfolio that defines the series’ central finding. A 60/40 allocation between the S&P 500 and the TF Index produced a terminal value of $8.21 from $1, exceeding the S&P’s $7.48 while reducing the maximum drawdown from 50.9% to 24.8%. The blend outperformed both components not by earning more per year but by losing less during crises. The geometric synergy between negatively correlated return streams creates a portfolio that compounds more efficiently than either component alone, and substituting top-quartile programs for the broad index amplifies the effect.
Episode 10: The Compounding Machine: 26 Years of Geometric Evidence
This episode opens the database at the individual-program level. Mulvaney Capital turned $1 into $75.50 (18.03% CAGR), exceeding Berkshire Hathaway’s $14.11 by a factor of five. Roughly nine in ten of the analyzed programs produced positive skew, and seventeen of the thirty-eight beat Berkshire’s MAR of 0.240. The geometric decomposition identifies two archetypes: the concentrated right-tail exploiter, typified by Mulvaney, which generates extraordinary arithmetic returns through aggressive leverage and pyramiding across a focused market set; and the geometric optimiser, typified by Salus Alpha and AHL Alpha, which compounds through breadth, diversification, and minimal volatility drag. Both produce genuine long-term compounding, but they are not equally appropriate for all investors. The leverage fallacy is dismantled from several angles.
Episode 11: The Ensemble: Why Convergence Proves the Process
When roughly nine in ten of the analyzed programs show positive skewness, about four in five show negative equity correlation, and seventeen of the thirty-eight beat Berkshire’s MAR, the explanation shifts from individual brilliance to structural process. This episode tests the convergence hypothesis across the population and addresses three objections: survivorship bias, start-date dependence, and the capacity/decay argument. The extended record, using DUNN WMA and EMC Classic data from the mid-1980s, extends the analysis to 39 years and six crises, with crisis alpha delivered in every one.
Episode 12: The Patience Cost: What Geometric Wealth Demands
Every geometric benefit has a cost, and this episode quantifies it in unsparing detail. The TF Index is positive in only about 55% of months, spends roughly 83% of its history below its previous high-water mark, and underperforms the S&P 500 on a rolling three-year basis in about 69% of all windows. During the 2010 to 2019 decade it returned 3.3% annually while the S&P returned 13.6%, and about four in five of the programs underperformed the S&P on raw CAGR. But the episode goes further than quantifying the cost: it explains it. The patience cost is regime-conditional, not constant, and the 2020 to 2022 recovery confirmed the mechanism was intact. Investors who treat the QE decade as the expected baseline are calibrating from the worst case in four decades of data.
Episode 13: The Right Toolkit: Evaluating Managers Through the Geometric Lens
This episode replaces the Sharpe ratio with a seven-metric geometric toolkit: MAR, skewness, Sortino, crisis performance, drawdown duration, negative S&P correlation, and volatility drag. The toolkit is then applied beyond trend following to value investing, volatility selling, and carry, identifying the same structural pathology in different forms: negative skew, positive crisis correlation, and drawdown-to-CAGR ratios the Sharpe ratio cannot detect. Finally it is tested out of sample through a disciplined walk-forward process: each year, from the point-in-time universe of long-running programs with at least a 15-year audited record, the ten-manager ensemble with the highest MAR is selected and held for the following year, using only data available at the time. Choosing ten programs from a universe of fifty spans more than ten billion combinations, yet the foresight-free process converges close to the hindsight-optimal result, compounding at roughly 7.1% with a maximum drawdown near 11.5% and no look-ahead at any decision.
Act IV: The Construction (Episodes 14 to 15)
The final act translates the geometric framework into practical portfolio construction and distils the entire series into a set of propositions the reader can carry forward.
Episode 14: Building the Geometric Portfolio
This episode addresses the practical questions: how much to allocate, how to rebalance, and how to evaluate implementations. The geometric frontier identifies 40 to 60 percent TF as the optimal allocation range, raising CAGR and lowering drawdown simultaneously through the first 40 percent of allocation, with every allocation above roughly a quarter TF exceeding Berkshire’s MAR. The right-tail-preservation argument shows that classic implementations which let positions ride at full exposure through trends preserve the right tail that aggressive volatility targeting clips. The episode builds a monitoring framework on skewness, MAR, and crisis performance rather than Sharpe or rolling relative return, and argues that structural allocation bands outperform calendar rebalancing by giving the geometry room to express itself.
Episode 15: Process, Not Prophecy
The final episode assembles the entire argument into six propositions and two corollaries that constitute the geometric case. It revisits the key findings across all four acts, presents the 39-year extended record alongside the 26-year database evidence, restates the costs honestly, acknowledges the structural limitations that invite further research, and closes with the thesis that geometric wealth is not a prediction about the future. It is a description of the mathematical properties that have determined compound growth for as long as financial markets have existed.
The Geometric Case: Key Findings
Across fifteen episodes, the series builds a cumulative argument resting on five headline findings, each derived from real performance data, all net of fees.
The blended portfolio outperforms both components. A 60/40 allocation between the S&P 500 and the TF Index produced $8.21 from $1, exceeding the S&P’s $7.48 and the TF Index’s $6.67, while reducing the maximum drawdown from 50.9% to 24.8%. The blend earned less per year on average but kept more of what it earned. This is geometric synergy: two imperfect return streams combining to produce a path smoother than either alone.
Trend following delivers crisis alpha with extraordinary consistency. During every major equity drawdown since January 2000, the TF Index produced positive returns while the S&P 500 fell. The extended record, spanning 39 years and six crises from the 1987 crash through the 2022 bear market, shows the same pattern: six for six. The mechanism is structural, not coincidental, and it has operated for as long as systematic trend following has existed in its modern form.
The geometric properties converge across the population. Roughly nine in ten of the analyzed programs show positive skewness, about four in five show negative equity correlation, and seventeen of the thirty-eight beat Berkshire Hathaway’s MAR. These are not properties of a few exceptional managers. They are structural features of the trend-following process itself, observable across different firms, different markets, and different time periods.
The patience cost is real and unsparing. About four in five of the programs underperformed the S&P 500 on raw CAGR. The TF Index spent a full decade, 2010 to 2019, trailing badly. The process is positive in only about 55% of months and spends roughly 83% of its history in drawdown. The geometric benefits and the patience costs are inseparable: the crisis alpha exists because the process accepts years of relative underperformance as the price of structural protection.
The geometric toolkit works out of sample, and across strategies it was not designed to evaluate. Applied to value investing, volatility selling, and carry, the same seven metrics identify the structural weaknesses the Sharpe ratio cannot detect: negative skew, positive crisis correlation, and drawdown burdens disproportionate to the premium earned. Tested forward, the walk-forward allocator, selecting each year the highest-MAR ten-manager ensemble from the eligible long-record universe using only data available at the time, converges close to the hindsight-optimal result without any look-ahead. The geometric fingerprint visible in historical data persists in out-of-sample performance, which is the evidence that the process selects managers who compound wealth in the future, not merely managers who compounded it in the past.
What This Series Is Not
The Geometry of Wealth is not a prediction that trend following will outperform equities. It is a demonstration that trend following produces a return stream with specific geometric properties, that those properties improve compound growth when combined with equity exposure, and that the evidence for this is consistent across 26 years of population-level data and 39 years of individual-program records.
The series acknowledges its limitations openly. It relies primarily on a single database. Its complexity-theory arguments, while grounded in established science, are ultimately difficult to falsify as applied to financial markets. The fee analysis, while net of reported fees, does not capture the full cost of implementation for a new allocator. Capacity is addressed empirically rather than resolved theoretically: the available cross-market evidence shows no decay in the harvested structure despite large growth in systematic capital, but fund-level capacity constraints from slippage and market impact remain real at sufficient scale. And the January 2000 start date, though extended to the mid-1980s through individual-program records, remains a constraint on the primary analysis. Each limitation invites further research. None invalidates the geometric findings.
The reader who has finished this series will not know whether trend following will perform well next year. They will know why it has performed the way it has for nearly four decades, what mathematical properties drive that performance, what it costs, and how to evaluate it. The geometry of wealth is not a forecast. It is a framework.
The investor who understands the geometry of wealth does not predict the future. They engineer a path that compounds through it.
Data and Sources
All performance data is drawn from the NilssonHedge Trend Following performance file (January 2000 to January 2026), net of management and performance fees. The analysis universe is the 47-program TTU TF Index, defined by a minimum 15-year track record, with per-program exhibits drawn from the 38 programs carrying 20-year-plus records over the period. Extended analysis uses the DUNN WMA program (from November 1984) and the EMC Classic program (from January 1985), with S&P 500 Total Return from January 1987, over a 39-year common window. Crisis periods are the 1987 crash, the 1990 recession, LTCM/Russia 1998, the dot-com bust (2000 to 2002), the global financial crisis (2007 to 2009), and the 2022 bear market. MAR ratio is CAGR divided by the absolute maximum drawdown; the geometric return is approximately the arithmetic return minus half the variance. The walk-forward allocator selects, each year, the highest-MAR ten-manager ensemble from the point-in-time universe of programs with at least a 15-year audited record, equal-weights it, and holds for the following year, using no hindsight data. Mechanism and structure-persistence material drawn from external research is identified as such in the relevant episodes and should be cited to its original sources at publication. No simulated or hypothetical data appears anywhere in the series.
The series title shares its name with Brian Portnoy’s book on the relationship between money and meaning. This series addresses a different question, the mathematics of compound growth, but the shared language of geometry and wealth reflects a common conviction that how we think about returns matters as much as the returns themselves.
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.