Wealth is not found by the brilliant. It is built by the disciplined. Not through the right predictions, but through the right process
This series began with a simple demonstration. Two strategies with the same arithmetic average return produced different terminal wealth because one suffered deeper drawdowns than the other. The difference was not in the average. It was in the path. Fourteen episodes later, that demonstration has become a framework, the framework has been tested against dozens of independent implementations across 26 years, extended backward to 39 years with independent data, the costs have been quantified, and the portfolio has been constructed.
This final episode synthesises the argument the data has been building toward, across all four acts, from first principles to final construction. The argument is larger than trend following. It is about the nature of wealth itself.
Act I established the mathematical foundation: the arithmetic mean overstates the geometric outcome, drawdowns are exponentially destructive, and the path determines the terminal value. Act II identified the mechanism: markets produce trends structurally, the cut transforms those trends into a positively skewed return stream, and crisis alpha is the most valuable geometric property. Act III deployed the evidence: the analyzed programs converge on the same geometric fingerprint, maximum diversification across uncorrelated markets is itself a geometric strategy that ensures intersection with structurally inevitable tail events, survivorship bias and start-date dependence are addressed with data, the extended record confirms the fingerprint across 39 years and six crises, and the patience cost is quantified without flinching. Act IV converted the framework into practice: the geometric toolkit identifies future compounders out of sample through a disciplined walk-forward process, and the portfolio-construction guide establishes that right-tail preservation and structural allocation bands outperform the smoothing and calendar rebalancing that conventional practice favours. What remains is the synthesis: what does this evidence, taken together, reveal about how wealth is actually built?
The Complete Framework
The series has established six propositions and two corollaries. The propositions are foundational mathematical truths about compounding in a multiplicative world; the corollaries are their implementation consequences. Each was introduced as theory and confirmed with evidence.
Proposition 1: Wealth compounds geometrically, not arithmetically. The arithmetic mean overstates the compound return. The gap between them is the volatility drag, and it is created by adverse geometry: the deeper a path’s drawdowns, the lower the base from which it must recompound, and the wider that gap grows. Two strategies with identical arithmetic returns produce different terminal wealth when their paths differ, and the one that contains its drawdowns outcompounds the one that suffers them. The enemy is not variation but adverse variation; the upside roughness that produces positive skew is geometry to preserve, not to smooth away. (Episodes 1 and 2.)
Proposition 2: Deep drawdowns are exponentially destructive. A 50% loss requires a 100% gain to recover. The recovery cost is convex: it accelerates as drawdowns deepen, and every month spent recovering is a month not spent compounding. Drawdown avoidance is not a preference. It is a mathematical imperative. (Episodes 3 and 4.)
Proposition 3: Positive skew is geometrically superior. A distribution of many small losses and occasional large gains compounds more efficiently than one of many small gains and occasional large losses, even at identical arithmetic averages. The shape of the distribution, not just its centre, determines the compound outcome. (Episode 7.)
Proposition 4: Markets produce trends structurally. Financial markets are complex adaptive systems in which participants respond to price changes, creating feedback loops that generate persistent directional movement. Trends are not anomalies. They are structural features of systems with reflexive participants, information asymmetry, and regime transitions. (Episode 5.)
Proposition 5: The cut is the source of geometric wealth. The systematic truncation of losses produces the positive skew, the drawdown containment, and the crisis alpha that generate geometric value. The entry determines which trends are captured; the cut determines the geometry of the return stream. The cut is not a feature of the process. It is the process. (Episode 6.)
Proposition 6: Portfolio synergy converts individual underperformance into collective outperformance. A blend of equities and trend following produces more terminal wealth than either component alone, because the negative crisis correlation reduces the portfolio’s drawdowns, which reduces the geometric drag, which compounds forward through every subsequent month. The blend is greater than the sum. (Episode 9.)
Corollary A: Maximum diversification is a geometric strategy, not a risk-management tool. Markets produce outliers as structural inevitabilities at every scale; the question is never whether an outlier will occur but which market and when. Each additional uncorrelated market is an independent draw from a fat-tailed distribution, more draws means a higher probability of capturing trunk-scale events, and each captured outlier permanently elevates the platform from which all future compounding proceeds. The cost of a missed outlier is not the outlier itself but the entire compounding chain that would have followed. A narrow portfolio needs luck; a maximally diversified one converts luck into process. (Episode 10.)
Corollary B: The right tail must be preserved, not smoothed. Aggressive volatility targeting and dynamic position sizing reduce exposure to the very conditions that produce outliers, because outliers are by definition high-volatility events. A system that cuts position size as volatility rises is selling the trunk while the trunk is growing. Classic trend following, which sets position size at entry and lets the trend run at full exposure, preserves the right tail that drives geometric wealth, and at the portfolio level structural allocation bands outperform calendar rebalancing that mechanically sells the outperforming component. Smoothing and compounding are different objectives. Only one builds wealth. (Episode 14.)
Six propositions and two corollaries, each confirmed by 26 years of evidence across the analyzed programs, and each further confirmed by extended data reaching back to 1987: two additional managers, six crises, and 39 years of compounding that includes the most hostile decade the process has ever faced. Together they describe a geometry of wealth that operates beneath the surface of markets, invisible to arithmetic evaluation but decisive in determining terminal outcomes.
The Final Chart
Figure 1 contains the entire argument: two panels, two time horizons, the same conclusion.
Figure 1: The complete evidence. Panel A: the extended record (Jan 1987 to Dec 2025), DUNN WMA ($130), EMC Classic ($112), and the S&P 500 ($63) across 39 years and six crises. Panel B: the NilssonHedge record (Jan 2000 to Jan 2026), Mulvaney ($75.50), Berkshire ($14.11), the 60/40 blend ($8.21), the S&P 500 ($7.48), and the TF Index ($6.67). Crisis periods shaded; log scale; all returns net of fees.
Panel A shows the extended record. EMC Classic turned $1 into $112 over 39 years and DUNN WMA turned $1 into $130, while the S&P 500 turned $1 into $63. Both trend followers outcompounded the most successful equity index in the world across nearly four decades, net of fees, through six crises including the 1987 crash, the LTCM crisis, and the global financial crisis. The process did not begin working in January 2000. It has been working since modern systematic trend following began.
Panel B shows the NilssonHedge record. Mulvaney’s $75.50 per dollar is the result of a systematic process, run by algorithms, with no fundamental analysis, no stock selection, and no macroeconomic forecasting. Berkshire’s $14.11 is the result of one of the greatest investment minds of the past century. The process produced more than five times the terminal wealth of the genius, not because it is smarter, but because it is geometrically more efficient. Mulvaney’s path included a 60.9% drawdown, Berkshire’s a 44.5% drawdown, the S&P’s a 50.9% drawdown, the TF Index’s a 21.0% drawdown, and the 60/40 blend’s a 24.8% drawdown. The terminal values are not accidents. They are consequences of the paths, and the paths are consequences of the geometry of each return stream: its skew, its drawdown profile, its crisis behaviour, and its volatility drag.
Notice what both panels reveal during the shaded crisis periods. The trend-following curves rise while the S&P 500 falls. In Panel A the pattern holds across six crises spanning 39 years, with DUNN WMA positive in all six. In Panel B the 60/40 blend dips less than the S&P and recovers faster, and its $8.21 exceeds the S&P’s $7.48 not because it earned more per year (its arithmetic return is lower), but because it lost less during the months that matter most.
The Costs Restated
Before the final word, the costs must be restated. A series that presents only the benefits is a sales pitch. This is an investigation.
The trend-following process produces a return stream positive in only about 55% of months. It spends roughly 83% of its history below its previous high-water mark. It underperforms the S&P 500 on a rolling three-year basis in about 69% of all windows, and its worst rolling five-year CAGR was negative. During the 2010 to 2019 decade it returned 3.3% annually while the S&P returned 13.6%; during the 2023 to 2026 rally it returned 0.8% while the S&P returned 22.8%. Roughly four in five of the trend-following programs in the database underperformed the S&P on raw CAGR, net of fees, and the TF Index itself underperformed the S&P on raw CAGR. The best individual implementations require institutional minimums that exclude most investors. The patience cost is not a footnote. It is the price.
These costs are not weaknesses to be apologised for. They are the mechanism. The low win rate is the cut doing its job. The rolling underperformance is the price of crisis alpha. The time in drawdown is the process staying invested across dozens of markets while waiting for the trends that generate geometric wealth. An investor who wants the benefits without the costs is asking for a process that does not exist. There is no strategy that captures crisis alpha without underperforming during non-crisis periods, none that produces positive skew without a win rate below 50%, and none that contains drawdowns without spending most of its time below the high-water mark. The geometry of wealth is not a menu from which the investor selects favourite items. It is a unified system in which every benefit is paid for by a corresponding cost. Understanding this is the prerequisite for the discipline the process demands.
The Picks Illusion: Final Resolution
The investment industry is built on a story. The story says wealth is created by brilliant selection: the right stock, the right sector, the right trade at the right moment. It is compelling because it has heroes, whose biographies are studied and whose insights are venerated. The founding narrative is that wealth comes from knowing something others do not.
This series has presented an alternative, supported by 26 years of evidence across the analyzed programs and confirmed by an additional thirteen years of extended data from two independent managers. The alternative says wealth comes from the geometric properties of the return stream, regardless of what produced it. Selection is one input to the geometry; it is not the geometry itself. A process that optimises directly for geometric properties, that cuts losses, rides trends, and produces positive skew and crisis alpha, generates outcomes that rival or exceed the greatest selector who ever lived. The evidence is specific: seventeen of the analyzed trend-following programs beat Berkshire Hathaway on MAR, the measure of geometric efficiency, without selecting a single stock, reading a single annual report, or forming a single opinion about intrinsic value. The picks illusion is the belief that the picks were what mattered. The geometric evidence says the properties of the resulting return stream are what mattered.
The illusion persists because it is a better story. Genius selecting assets is dramatic; an algorithm following prices is invisible. Human narratives require protagonists, and the stock picker is a more compelling protagonist than the systematic process. But compelling narratives are not the same as accurate descriptions of how wealth is created in a multiplicative world. The illusion also persists because the industry’s infrastructure is built around it: fund marketing emphasises the manager’s biography, performance attribution decomposes returns into selection and allocation effects, and due-diligence questionnaires ask about the ability to identify mispriced securities. The geometric properties that actually determine compounding, skew, drawdown depth, crisis performance, and volatility drag, are relegated to supplementary risk statistics, if they appear at all. This series has argued that the supplementary statistics should be primary and the primary statistics supplementary. An investor who understands this inverts the evaluation: they start with the geometric output and work backward to the process that produced it, rather than starting with the narrative appeal and hoping the geometry follows.
Seventeen systematic processes, with no picks, no opinions, and no fundamental analysis, produced higher geometric efficiency than the greatest stock picker in history. Two of them have been doing it for 39 years. The picks are inputs. The geometry is the output. The output is what compounds.
Process Over Skill
There is only one Warren Buffett. His success cannot be replicated, systematised, or transferred; it is the product of a singular intellect applied with extraordinary discipline over an extraordinary career, and when he retires there will be no second Buffett. Trend following is a process. It can be learned, codified, systematised, and scaled, and it has been implemented independently by dozens of firms across multiple continents for decades. The geometric properties emerge across all these implementations not because the implementers are geniuses, but because the process itself produces the properties as a mathematical consequence of its architecture. The process is replicable and permanent.
This is not an argument that skill does not exist. Buffett’s record is real, and his $14.11 per dollar at a 10.68% CAGR and a MAR of 0.240 is exceptional by any standard. The argument is that a disciplined process, applied consistently to the true geometry of wealth, creates outcomes that rival or exceed the greatest individual skill the investment world has produced, and that the process has an advantage skill does not: it survives the practitioner. Buffett’s edge ends when he does. DUNN Capital has compounded through the same process since 1984, EMC Capital since 1985, Mulvaney since 1999. The process does not retire.
A persistent objection must be answered here, at the end. As systematic capital has grown, critics argue, the feedback structure must erode, each decade showing weaker trends and a shrinking edge. A separate study tested this directly across 68 futures markets over four decades, a period in which hundreds of billions in systematic capital flowed into the strategy. It found no decay: the autocorrelation oscillation amplitude in the most recent decade was statistically indistinguishable from the first, the feedback structure retaining the large majority of its original magnitude, with several asset classes showing higher amplitude now than in the 1980s, including the markets most heavily targeted by systematic capital. The most parsimonious reading is that the structure has not been arbitraged away because it cannot be. It is not an inefficiency exploitable to zero. It is the permanent consequence of a market populated by agents, human and algorithmic, whose behaviour is conditioned on price. As long as one agent responds to price, feedback exists, and every new systematic strategy that enters the market does not eliminate the mechanism. It becomes part of it.
The evidence does not suggest that every investor should abandon stock selection and allocate entirely to trend following. It suggests something more nuanced: that the geometric properties of the return stream matter more than the method used to produce them. An investor who selects stocks but produces negative skew and deep drawdowns will compound less wealth than one who follows a systematic process that produces positive skew and contained drawdowns. The method is the means; the geometry is the end. This applies to any strategy, value, momentum, carry, volatility selling, or macro, each of which can be evaluated through the same lens: what is the skew, the maximum drawdown, the MAR, the crisis behaviour, the volatility drag? The geometric framework is not an argument for trend following over other strategies. It is an argument for evaluating all strategies through the lens that actually determines terminal wealth.
One final lesson from the mathematics deserves emphasis. The leverage fallacy of Episode 10 showed that CAGR is path-dependent while maximum drawdown is not: doubling leverage roughly doubles the drawdown but does not double the CAGR, because the volatility tax scales with the square of volatility, so each increment of leverage delivers diminishing returns to compounding and proportional increases to drawdown. Geometric efficiency cannot be manufactured through leverage. It must be earned through the quality of the signal, the design of the process, and the discipline of the path. And the proof is in the application: when the geometric toolkit was used to select managers out of sample, through a disciplined walk-forward process choosing a ten-manager ensemble each year by MAR over a rolling fifteen-year window, using only data available at the time of each decision, the resulting portfolio compounded at roughly 7.1% with a maximum drawdown near 11.5% and a MAR in the region of 0.6, well above the S&P’s 0.157, with no look-ahead at any decision. The toolkit does not merely describe the past. It selects future compounders, because the properties it measures are structural, not cyclical.
The Final Ledger
One dollar invested in January 2000, followed through 313 months to January 2026, net of fees.
The Final Ledger. Jan 2000 to Jan 2026, net of fees. Every terminal value has a path. The geometry of wealth is the study of which paths lead to which outcomes, and why.
The Path column is the series in miniature. Mulvaney’s $75.50 required surviving a 60.9% drawdown. Berkshire’s $14.11 required one unreplicable mind. The 60/40 blend’s $8.21 required the discipline to hold trend following through a decade of underperformance. The S&P’s $7.48 required surviving a 50.9% drawdown that cut the compounding base in half. None of these outcomes is free; all are the product of a path. The extended record adds two more rows to the investor’s mental ledger: DUNN WMA, $1 into $130 over 39 years through a 60.3% drawdown, and EMC Classic, $1 into $112 over 39 years through a 45.2% drawdown. Both required the same discipline across a longer horizon that includes a decade of deliberate trend suppression. Both outcompounded the S&P 500. The geometry is the same. The horizon is wider.
The Final Word
Wealth is not found. It is built.
It is not built by the right prediction at the right moment. It is built by the right process, applied with discipline, across decades, through the barren years and the abundant ones. The process does not require genius. It requires understanding: that returns multiply, that drawdowns destroy, that the path matters more than the average, and that a return stream’s geometry, its skew, its drawdown profile, its crisis behaviour, determines its terminal value.
The evidence of dozens of independent implementations, operating across 26 years that included four crises and two of the longest bull markets in history, confirms that the trend-following process produces the geometric properties that compound wealth efficiently. The extended evidence of two further implementations, across 39 years and six crises, confirms that these properties are not artefacts of a particular starting date or era. They are properties of the process itself. It achieves this with a lower win rate, more time in drawdown, and longer stretches of underperformance than equities. The costs are structural; the benefits are structural; both are consequences of the same process.
The investor who allocates to this process must accept the costs: the years of watching buy-and-hold outperform, the rolling metrics that show underperformance two times in three, the mediocre months and the low win rate and the time below the high-water mark. These are not optional. They are the price. In return, the investor receives a set of geometric properties that, over every full market cycle in the dataset, have transformed portfolio efficiency: positive skew, drawdown containment, crisis alpha, and the negative equity correlation that produces a blend greater than the sum. These properties do not depend on any individual manager’s brilliance. They are properties of the process itself, and the process is available, in various forms, to any investor willing to understand the geometry and pay the patience cost.
The series has not argued that trend following is the only path to geometric wealth. It has argued that the geometry is what matters, and that trend following is the clearest, most evidence-rich example of a process engineered for geometric efficiency. Other processes may produce similar properties; the framework for evaluating them is now in the reader’s hands: six propositions, two corollaries, a seven-metric toolkit, and evidence spanning 26 to 39 years that establishes what geometric efficiency looks like in practice. The reader who has followed this series from Episode 1 to Episode 15 now possesses what most market participants do not: a mathematical framework for understanding why some paths produce more wealth than others, independent of what generated those paths. That understanding does not guarantee wealth, and it does not eliminate the patience cost or the difficulty of holding a divergent strategy through years of underperformance. But it provides the only foundation on which the necessary discipline can be built, which is the understanding that the cost is real, the geometry is real, and the two are inseparable.
Wealth is not found by the brilliant. It is built by the disciplined. Not through the right predictions, but through the right process. The blueprint is a process, not a prophet. And the geometry is the wealth.
This concludes The Geometry of Wealth.
Data and Sources
All performance data is drawn from the NilssonHedge Trend Following performance file. The primary window is January 2000 to January 2026 (313 monthly observations); the extended record runs January 1987 to December 2025 (468 months). All returns are net of management and performance fees. The wider manager universe referenced across the series is the 47-program TTU TF Index, defined by a minimum 15-year track record, with per-program exhibits drawn from the 38 programs with 20-year-plus records, consistent with Episodes 10, 11, 13, and 14. The S&P 500 Total Return, Berkshire Hathaway, and TTU TF Index series are from the same file, and the 60/40 S&P/TF blend uses monthly rebalancing.
The extended record uses the DUNN Capital Management WMA program (from November 1984) and the EMC Capital Advisors Classic program (from January 1985), measured over the 39-year common window with the S&P data (January 1987 to December 2025): DUNN $1 to $130.25 at 13.30% with a 60.3% maximum drawdown, EMC $1 to $112.37 at 12.87% with a 45.2% maximum drawdown, and the S&P 500 $1 to $63.49 at 11.23%, all from the NilssonHedge master file, consistent with Episode 11. The 26-year final ledger figures (Mulvaney $75.50, Berkshire $14.11, the 60/40 blend $8.21, the S&P $7.48, the TF Index $6.67) and Figure 1 were verified directly against the source monthly returns.
The walk-forward allocator result referenced from Episode 13 (a ten-manager ensemble selected annually by MAR over a rolling 15-year window, drawn from the point-in-time eligible universe of long-running programs, approximately 7.1% CAGR at an 11.5% maximum drawdown, with no look-ahead) is documented in that episode. The six propositions and two corollaries reference the findings of the specified episodes. The leverage analysis applies leverage multiples directly to the monthly return series. The structure-persistence results (a 68-market study of decade-on-decade autocorrelation amplitude) derive from a separate research study and are not reproducible from the performance database; they should be cited to that study directly when this episode is published. No simulated or hypothetical data appears anywhere in this series.
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.