How cutting losses protects the compounding engine with convex efficiency, and why the willingness to be wrong often is the price of being right geometrically
PREVIOUS: THE GEOMETRY OF WEALTH | Episode 5 of 15: Why Markets Trend,(And Why They Always Will)| NEXT: THE GEOMETRY OF WEALTH | Episode 7 of 15: Letting Winners Run and Harvesting the Right Tail
The TTU Trend Following Index loses money in 45% of all months. The S&P 500 loses money in 35%. The trend following process has more losing months than the stock market. It also has less than half the maximum drawdown and more than double the MAR ratio. This is not a contradiction. It is the geometric logic of cutting losses.
The conventional view of investment success is shaped by batting averages. Good investors are right more often than they are wrong. Good funds have more winning months than losing months. A strategy that loses money 45% of the time, nearly half of all periods, would be dismissed by most investors as unreliable. The psychic pain of being wrong nearly every other month is real, and most investors are unwilling to endure it.
But Episodes 1 through 4 showed that terminal wealth is determined not by the frequency of losses but by their depth. A process that loses 2% in 140 months and never loses more than 9% in any single month will compound more wealth than a process that loses 4% in only 109 months but occasionally loses 17%. The arithmetic frequency favours the second process. The geometric outcome favours the first. This entire series is built on the proposition that the geometric outcome is the one that determines how much money you actually have at the end. And the geometric outcome is determined by one thing above all else: the depth of the worst losses.
The cut is the mechanism that controls this depth. It is not a risk overlay. It is not a safety feature added to a strategy. It is the strategy. Everything else is downstream of the cut.
The Anatomy of the Cut
Trend following has a simple mechanical structure. When a price moves in a direction that confirms a developing trend, the process establishes or adds to a position. When the price reverses by a predefined amount, the process exits or reduces the position. The specific rules vary across managers: some use moving average crossovers, others use breakout channels, others use statistical measures of momentum. The timeframes vary from weeks to months. The instruments span equities, bonds, currencies, and commodities. But the deep structure is universal: enter in the direction of a trend, exit when the trend reverses.
The exit rule is the cut. And the cut has three geometric properties that connect directly to the compounding framework of the first four episodes.
First, the cut caps the left tail. When a position begins to lose money, the exit rule limits how far the loss can extend before the position is closed. The exact level varies by manager and by market, but the principle is universal: there is a predefined point at which the process says “enough” and takes the loss. This cap on individual trade losses is the mechanism that produces the truncated left tail observed in Episode 4. The S&P 500 had four months with losses exceeding 10%. The TF Index had zero. Not because trend following avoids bad markets, but because the cut removes the position before a bad market can inflict a catastrophic loss.
Second, the cut operates with convex efficiency. Recall from Episode 2 that the recovery curve is convex: each additional unit of loss costs exponentially more than the last. A loss capped at 5% requires a 5.3% recovery. A loss allowed to reach 20% requires a 25% recovery. A loss allowed to reach 50% requires a 100% recovery. By capping losses in the shallow end of the recovery curve, the cut avoids the exponential escalation that begins beyond 20% to 30%. The geometric benefit of the cut is not linear. It is convex. The deeper the loss that the cut prevents, the exponentially greater the compounding value preserved.
Third, the cut accepts frequent small losses as the price of avoiding infrequent large ones. This is the trade-off that defines the entire process and that most investors cannot tolerate. The trend follower will cut a losing position and pay a small, bounded loss, many times. False signals, whipsaws, range-bound markets: all of these produce small cuts. The cost is a low batting average, the 45% win rate that looks so unattractive on a fact sheet. The benefit is that no single loss, and no accumulation of losses in a crisis period, can push the portfolio into the catastrophic region of the recovery curve. The many small cuts are the insurance premium. The avoidance of a single deep drawdown is the payout. And as Episode 2 demonstrated, that payout is geometrically enormous.
The trend follower does not avoid losses. The trend follower curates losses. Small, bounded, frequent, endurable. Every cut is a loss. But every cut is also a wall between the portfolio and the convex destruction that lies beyond it.
Loss Capping in the Data
The theory is precise. The evidence is overwhelming.
Read this table carefully. The TF Index has more losing months than the S&P 500: 140 versus 109. But its average loss is 32% smaller (−2.64% versus −3.90%). Its worst single month is roughly half as severe (−8.55% versus −16.79%). It produced half as many months worse than −5%. And it produced zero months worse than −10%, while the S&P 500 produced four and Berkshire produced seven.
The TF Index does not avoid losing. It avoids losing badly. It accepts the frequency of small losses as the cost of eliminating catastrophic ones. The batting average is lower. The depth of each loss is smaller. And because depth, not frequency, is what drives the convex recovery curve, the geometric outcome is superior.
The worst month matters enormously. The S&P 500’s worst month, October 2008 at −16.79%, destroyed nearly one-sixth of the index’s value in 31 days. The TF Index’s worst month, March 2003 at −8.55%, was half as severe. But the geometric difference is not half. It is convex. A 16.8% loss requires a 20.2% recovery. An 8.6% loss requires a 9.4% recovery. The ratio of loss depths is roughly 2:1. The ratio of recovery burdens is 2.15:1. Convexity means the geometric advantage of loss capping is always larger than the arithmetic difference in loss sizes would suggest.
When the Market Falls, the Cut Responds
The cut’s value is most visible during exactly the months that matter most for compounding: the months when the S&P 500 experiences its deepest losses.
In October 2008, the month that destroyed nearly 17% of the S&P 500’s value, the TF Index gained 10.9%. In September 2001, the month after the September 11 attacks, the TF Index gained 9.1% while the S&P fell 8.1%. In June 2002, deep in the dot-com wreckage, the TF Index gained 10.8% while the S&P fell 7.1%.
These are not coincidences. They are the mechanical output of the cut operating during crises. When equity markets crash, the trend following process has typically already cut its long equity positions and may have established short positions that profit from the continued decline. The cut removed the process from the left tail of the equity distribution before the worst of the damage occurred. It is not prediction. The process did not forecast that October 2008 would be catastrophic. It responded to the price action of September 2008, and of August 2008, and of the months before that, each month’s deterioration triggering a further reduction or reversal of positioning. By the time the worst month arrived, the process was positioned to profit from the very move that was destroying buy-and-hold portfolios.
Across the S&P 500’s worst decile of months, the 31 months that inflicted the deepest losses, the TF Index was positive in 20 of them and averaged +2.92%. The process is not merely surviving the worst months. It is compounding during them. And as Episode 2 showed, compounding during the months that destroy other portfolios is the single most valuable geometric property an investment process can possess.
The Counterfactual: What the Cut Is Worth
The geometric value of loss capping can be quantified through a simple counterfactual.
The TF Index compounded $100,000 to $667,058 over 26 years. Now consider a modified version of the same return stream in which every monthly loss is doubled: a −2% month becomes −4%, a −5% month becomes −10%, and so on. Every positive month is left unchanged. The only modification is the removal of the loss cap, allowing the downside to expand. The result: $12,943.
Read that again. The identical positive months, the same upside capture, the same winning trades, the same trend profits. Only the losses were deepened. And $667,058 became $12,943. The loss-capping mechanism is worth 98% of the terminal wealth. Without it, the same return stream becomes geometrically worthless.
This is not a theoretical curiosity. It is the single most important empirical finding in this series. It quantifies the proposition that has run through every episode: in a multiplicative world, the depth of losses is the dominant determinant of terminal wealth. Not the frequency of wins. Not the size of the best month. Not the average return. The depth of losses. And the cut is the mechanism that controls the depth of losses.
Doubling the losses on the TF Index, while keeping every gain identical, destroys 98% of terminal wealth. The cut is not a feature of the strategy. The cut is the strategy. Everything else is commentary.
A second counterfactual reinforces the point. Take the TF Index’s actual return stream and replace only its 10 worst months with the S&P 500’s 10 worst months. Every other month is untouched. Terminal wealth drops from $667,058 to $470,726, a loss of $196,332, or 29.4% of terminal wealth. Ten months. Out of 313. Changed from moderate losses to severe losses. And nearly a third of terminal wealth vanishes. The convexity of loss impact is not a gradual effect. It is a cliff. Allowing even a handful of months to escape the cut’s control is geometrically catastrophic.
The Shape of Drawdowns
Loss capping at the monthly level translates into drawdown containment at the portfolio level. The S&P 500 experienced 32 distinct drawdown episodes over 26 years. The TF Index experienced 29. But the shape of these drawdowns is qualitatively different in ways that determine long-term geometric wealth.
Chart 11: Drawdown profiles over 26 years. The S&P 500 is dominated by two deep craters (dot-com and GFC) that pushed below −40%. The TF Index experiences frequent shallow drawdowns but never breaches −21%. The depth distribution, not the frequency, determines geometric outcomes.
The S&P 500 had five drawdowns deeper than 10%, three deeper than 20%, and two deeper than 40%. The TF Index had ten drawdowns deeper than 10%, but only one deeper than 20%, and none deeper than 30%. The TF Index has more drawdowns, but they are all contained in the shallow region of the recovery curve, the region below 20% to 25% where recovery burdens are approximately linear and manageable.
This is the cut operating at the portfolio level. Individual trade cuts cap monthly losses. Monthly loss caps prevent the accumulation of damage that produces deep multi-month drawdowns. Deep multi-month drawdowns are the events that push the portfolio into the convex region of the recovery curve. By preventing the accumulation, the cut keeps the portfolio in the geometric safe zone: the region where recovery is a matter of months, not years, and where the compounding engine is never seriously interrupted.
The S&P 500’s two deepest drawdowns, at 44.7% and 50.9%, required 74 and 53 months to recover. The TF Index’s deepest drawdown, at 21.0%, was recovered far more quickly because the recovery burden was only 26.6%, not 104%. The same compounding engine that drives both return streams, the 7% to 8% CAGR that both processes deliver, recovers a 21% drawdown roughly three times faster than a 51% drawdown. The cut does not make the compounding engine faster. It prevents the compounding engine from being broken.
The Psychological Cost of the Right Geometry
If the cut is so geometrically powerful, why does the investment industry not universally adopt it? The answer is that the cut exacts a psychological toll that most investors cannot bear.
Losing money in 45% of months is painful. Cutting a position that subsequently recovers is painful. Watching a market reverse the day after a stop-loss triggers is painful. The catalogue of small losses that the trend follower accumulates, the whipsaws, the false signals, the range-bound markets that grind a portfolio sideways, feels like failure. Month after month, the process appears to be doing something wrong. It is being wrong nearly half the time.
But “wrong” is an arithmetic judgement. The frequency of losses is an arithmetic quantity. The depth of losses is a geometric quantity. And the geometric quantity is the one that compounds. The trend follower is arithmetically wrong often and geometrically right always: every cut, even the cuts that look like mistakes in hindsight, is a wall between the portfolio and the possibility of a catastrophic drawdown. The cuts that look wrong, the ones where the market immediately reversed, were still geometrically correct. They limited the worst-case loss. They preserved the compounding base. They kept the portfolio in the shallow end of the recovery curve. The fact that the worst case did not materialise on that particular trade does not diminish the geometric value of the protection.
This is the deepest tension in the investment industry. The behaviour that feels like disciplined investing, holding positions through volatility, averaging down on losers, maintaining conviction against the market, is the behaviour that exposes the compounding engine to the deepest possible drawdowns. The behaviour that feels like weakness, cutting losses frequently, admitting error quickly, accepting small defeats, is the behaviour that protects the compounding engine most effectively. The gut instinct of most investors is precisely inverted relative to the geometric reality.
The disposition effect, one of the most robust findings in behavioural finance, formalises this inversion: investors tend to sell winners too early and hold losers too long. They take profits prematurely (cutting the right tail) and refuse to take losses (extending the left tail). This behaviour is the exact opposite of what geometric compounding rewards. The trend following cut reverses the disposition effect systematically: it holds winners (extending the right tail) and cuts losers (truncating the left tail). It does, mechanically, what human psychology resists doing naturally.
The disposition effect causes investors to cut their winners and hold their losers. Trend following reverses this: it holds winners and cuts losers. The cut does systematically what human intuition refuses to do. This is not a coincidence. It is the reason the edge persists.
The Cut and the Picks Illusion
We can now state the relationship between cutting losses and the picks illusion with precision.
The picks illusion, introduced in Episode 1, is the belief that investment success is primarily determined by the quality of the assets selected. Pick the right stocks. Find the right sectors. Identify the winning themes. The emphasis is on the entry: what to buy, when to buy it, what will go up.
The cut redirects attention from the entry to the exit. It asserts that the exit rule, not the entry decision, is the dominant determinant of geometric wealth. A trend follower can enter on the wrong side of a trade, as happens in 45% of months, and still compound wealth over decades because the exit rule limits the damage of each wrong entry. Conversely, a stock picker can be right on every entry, selecting companies that eventually appreciate enormously, and still fail to compound wealth if they hold through the 50% drawdowns that occur along the way. The entry tells you what to own. The exit tells you how much you keep.
This is not an argument against intelligent selection. It is an argument about the hierarchy of geometric importance. In a multiplicative world, loss control is structurally more important than asset selection. The cut is geometrically senior to the pick. A portfolio with mediocre picks and excellent loss control will compound more wealth over 26 years than a portfolio with excellent picks and no loss control. The data has been showing this throughout the series. The TF Index does not select the best assets. It applies the best exit rules. And the exit rules, the cuts, are worth 98% of the terminal wealth.
The Running Ledger
Our $100,000 continues. This episode adds the loss profile that reveals the source of each benchmark’s geometric properties.
The TF Index loses more often and loses less badly. It has the worst frequency profile and the best depth profile. The depth profile is what compounds. Zero months below −10%. A worst month half as severe as the S&P 500’s. The cut is not a risk management overlay. It is the architecture from which the entire geometric outcome is built.
The Bridge
The cut protects the compounding engine by truncating the left tail. But protection alone does not create wealth. Something must drive the compounding forward, must create the positive returns that accumulate across decades. This is the second operation of the trend following process: letting winners run. If the cut is the shield, the trend ride is the engine.
Episode 7 will show how the process harvests the right tail of the distribution, the outsized winning months that drive long-term geometric wealth. It will show that the patience to hold winning positions through noise and volatility is the mirror image of the discipline to cut losers, and that together, these two operations produce the positive skew, the convex payoff profile, and the geometric superiority that the data has been revealing since Episode 1.
PREVIOUS: THE GEOMETRY OF WEALTH | Episode 5 of 15: Why Markets Trend,(And Why They Always Will)| NEXT: THE GEOMETRY OF WEALTH | Episode 7 of 15: Letting Winners Run and Harvesting the Right Tail
Data and Sources
All performance data from the NilssonHedge Trend Following Performance Database (January 2000 to January 2026). All returns are net of management and performance fees. Monthly loss profiles computed from 313 monthly observations. Worst decile defined as the 31 months with the lowest S&P 500 Total Return. Counterfactual analyses: “doubled losses” multiplies every negative monthly TF Index return by 2 while leaving positive returns unchanged; “10 worst months substituted” replaces the TF Index’s 10 worst months with the S&P 500’s 10 worst months, matched by rank. Drawdown episodes counted from peak to recovery; a new episode begins when a new all-time high is set. The disposition effect reference: Shefrin and Statman (1985), “The Disposition to Sell Winners Too Early and Ride Losers Too Long: Theory and Evidence.”
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.