Why how much you risk matters more than what you risk it on
The Load
A bridge can bear a certain weight. Below that threshold, traffic flows safely. Above it, the structure fails. The failure is not gradual. The bridge does not sag proportionally as load increases. It holds, holds, holds, and then it does not.
The bridge’s capacity is fixed by its design: the materials, the geometry, the span. A well-designed bridge can carry heavy loads safely. A poorly designed one fails under modest stress. The difference is not in the traffic. It is in the architecture.
Position sizing is the bridge between your capital and the market’s forces. It determines how much load your system can bear before it breaks. Get it wrong and no amount of skill in analysis, timing, or strategy selection will save you. Get it right and you create the conditions for survival across regimes you cannot predict.
Everything in this series has been preparation for this. Understanding structure, regimes, feedback, and horizons matters only if you translate that understanding into appropriate exposure. Sizing is where knowledge meets survival.
The Ruin Boundary
There is a region of capital from which recovery is possible and a region from which it is not. The boundary between them is the ruin threshold.
This boundary is not abstract. It is geometric. A 50% drawdown requires a 100% gain to recover. A 75% drawdown requires a 300% gain. A 90% drawdown requires a 900% gain. The mathematics are unforgiving: losses and gains are not symmetric. The deeper you fall, the steeper the climb back.
At some point, the climb becomes impossible within any reasonable timeframe. You have crossed from the region of recovery into the region of ruin. Your capital still exists, but your ability to participate meaningfully in markets does not.
Position sizing determines how close you operate to this boundary. Size too large and a single adverse regime shift can push you across. Size too small and you survive but compound too slowly to justify the effort and risk. The geometry of survival requires finding the zone between ruin and irrelevance.
Why Sizing Dominates
Most attention in trading goes to entry: which asset, which direction, which timing. Sizing receives less focus. This is backwards.
Consider two traders with identical analytical skill. One sizes positions at 2% risk per trade. The other sizes at 10% risk per trade. After a sequence of losses that both will inevitably encounter, the conservative sizer remains in the region of recovery. The aggressive sizer has crossed into territory from which the climb back is impractical. The difference is not in what they traded. It is in whether their path remained survivable.
The aggressive sizer can be right more often, have better timing, and pick better assets, and still end up with less capital than the conservative sizer. Sizing is not a detail to optimise after strategy selection. It is the primary determinant of long-term outcomes.
This is why formulaic approaches to optimal sizing fail in practice. The Kelly criterion, often cited as the mathematical solution to position sizing, assumes stationarity: a known, stable edge applied repeatedly under consistent conditions. In markets understood as complex adaptive systems, this assumption is false. Your edge is not fixed. It shifts with regimes, adapts as other participants respond, and cannot be measured with the precision the formula requires.
Sizing to a theoretical optimum derived from stationary assumptions does not produce optimal results in non-stationary environments. It produces ruin. The formula tells you to size larger when your measured edge is larger. But in a CAS, a large measured edge often reflects a regime that is about to shift, not a permanent feature to exploit. Sizing up into regime fragility is precisely backwards.
The solution is not fractional Kelly. It is abandoning the frame entirely. Position sizing in a CAS is not an optimisation problem with a calculable answer. It is a survival problem requiring buffers against uncertainty that cannot be quantified.
Sizing Across Regimes
A position sized appropriately for one regime may be catastrophically wrong for another.
In the low-volatility attractor, moderate position sizes produce moderate returns with moderate drawdowns. The feedback mechanisms are stabilising. Liquidity is available. The system absorbs disturbances. Sizing calibrated to this environment feels comfortable and produces steady results.
When the market transitions to the high-volatility attractor, everything changes. Volatility doubles or triples. Correlations spike. Liquidity vanishes. Feedback mechanisms amplify rather than dampen. The position size that was moderate in the prior regime is now extreme. The same notional exposure that produced a 2% daily swing now produces a 6% daily swing. The drawdown that would have been 15% is now 45%.
This is not bad luck. It is regime mismatch. The sizing was calibrated to a regime that no longer exists.
The implication is uncomfortable: you must size for the regime you are not in. Positions that feel appropriately sized during calm are often too large for stress. Positions that feel conservative during stress are often appropriate for calm. The discomfort of undersizing during good times is the cost of survival during bad times.
The Normalisation Trap
ATR-based position sizing is a normalisation tool. It ensures that a position in a volatile instrument and a position in a quiet one contribute similar risk to the portfolio. Without it, your crude oil position would dominate your bond position simply due to differences in price movement. ATR solves this problem elegantly.
But normalisation is not the same as regime-awareness.
ATR is backward-looking. It tells you what volatility has been, not what it will become. Worse, it is pro-cyclical in a dangerous way. When volatility is suppressed, ATR is low, and the formula gives you larger notional positions. You size up precisely when regime shift risk is accumulating beneath the surface. When volatility finally expands, the damage arrives before the ATR adapts. You size down only after the hit.
This is not a flaw to be engineered away. It is a limitation to be acknowledged. ATR-based sizing does what it does: normalise risk contribution across instruments with different volatility characteristics. It does not protect you from regime transition. That protection must come from elsewhere.
This is why you trade small.
The distribution of trend following returns makes this imperative, not optional. Most trades lose. The strategy’s edge does not come from a high win rate. It comes from the asymmetry between small, frequent losses and rare, large gains. The outliers justify the method. Everything else is the cost of capturing them.
But outliers arrive on their own schedule, not yours. You cannot predict when the trend that pays for two years of losses will begin. You only know that if you are not present when it arrives, the entire structure collapses. The losing periods are not obstacles to be endured before the strategy works. They are the strategy working. You are paying, trade by trade, for the right to be positioned when the distribution finally delivers.
Trading small is what keeps you at the table through the paying period. Size too large and a string of losses pushes you toward the ruin boundary before the outlier arrives. You sized for the gains you hoped for rather than the losses you would encounter first. The Outlier Hunter does not optimise for the winning trades. The Outlier Hunter optimises for surviving long enough to capture them.
The buffer against regime uncertainty does not live in the position sizing formula. It lives in total portfolio exposure. You run the system at a fraction of what calm conditions would permit, not because you lack confidence in your normalisation, but because normalisation solves a different problem than survival.
The Horizon Constraint
Sizing must match your operating horizon.
A position that is appropriately sized for a one-year holding period may be inappropriately sized for a one-day risk window. Volatility scales differently across horizons. Correlations change. The drawdown you can tolerate over a year may arrive concentrated in a week.
Short-horizon traders face constraints that long-horizon investors do not. They must survive daily and weekly fluctuations that average out over longer periods. Their sizing must be more conservative relative to expected returns because they are exposed to the full roughness of short-term price action.
Long-horizon investors face different constraints. They must survive the psychological pressure of extended drawdowns that short-horizon traders never experience. A drawdown that lasts two years tests patience in ways that a drawdown lasting two weeks does not, even if the magnitude is similar.
Sizing is not a single number. It is a function of the horizon over which you must survive. As discussed earlier in this series, structure itself is scale-dependent. Sizing inherits that dependence. Mismatch between sizing and horizon is as dangerous as mismatch between sizing and regime.
Correlation and Concentration
Individual position sizes do not determine portfolio risk. Correlation determines portfolio risk.
Five positions of 5% each do not create 25% exposure if the positions are uncorrelated. They create something closer to 11% effective exposure, diversified across independent bets. But the same five positions create nearly 25% effective exposure if correlations spike to near unity during stress.
This is the diversification illusion. Portfolios that appear well-diversified under normal correlations become concentrated under stress correlations. The sizing that seemed conservative when correlations were low becomes aggressive when correlations are high.
Sizing for survival means sizing for stress correlations, not normal correlations. It means treating diversification as a conditional benefit that may disappear precisely when you need it most. It means carrying less gross exposure than naive diversification calculations would suggest.
The Uncertainty Buffer
All of the above assumes you know your edge, your volatility, your correlations, and your regime. You do not.
Edge estimates are uncertain. Historical volatility underestimates future volatility during regime transitions. Correlation estimates are unstable. Regime identification is imperfect. Every input to a sizing calculation carries error, and the errors compound.
This uncertainty requires a buffer. Size smaller than your models suggest. Leave room for the model to be wrong. The buffer is not wasted capacity. It is insurance against the limits of your own knowledge.
Traders who size to theoretical optimum assuming their estimates are correct will eventually encounter a situation where their estimates were wrong. The buffer is what allows them to survive that encounter and continue operating. Those without the buffer do not get a second chance to update their estimates.
Sizing as Architecture
Position sizing is not a parameter to optimise. It is architecture to design.
It encodes your beliefs about regime uncertainty, horizon constraints, correlation instability, and the limits of your own knowledge. It determines whether your system survives the events that stress-test all systems.
The right size is not the one that maximises returns over a backtest. It is the one that allows you to continue operating through the regimes the backtest did not include. It is the one that keeps you in the region of recovery when others cross into the region of ruin.
This requires accepting lower returns during favourable periods. It requires watching others with larger positions compound faster during calm. It requires the discipline to maintain conservative sizing when everything suggests you could safely take more risk.
The discipline is easier to maintain if you understand what it purchases: the right to continue participating. The traders who sized aggressively during calm are not in the market during the recovery. They crossed the ruin boundary during the stress. The conservative sizer is still present, still compounding, still in the game.
The Bridge and the Load
The bridge does not know what traffic will cross it tomorrow. It cannot predict the convoy of heavy trucks that arrives during a storm. It can only be designed to bear loads across a range of conditions, including conditions it has not yet experienced.
Your sizing is your bridge. It must bear the load of regimes you have not yet encountered, drawdowns you have not yet experienced, correlations you have not yet observed. It must be designed not for the traffic you expect, but for the traffic that is possible.
Size for survival. Everything else follows from staying in the game.
The question is not how much you can risk.
The question is how much you can risk and still be here tomorrow.
This is the eleventh article in a series exploring the deep structure of markets. Next: “Inhabiting Uncertainty: The Practice of Not Knowing”
Want the theoretical foundation for why trend following works?
The Fractals of Finance: Determinism, Adaptation and the Geometry of Markets bridges complexity science with practical trading implementation. With a foreword by Jerry Parker, original Turtle Trader.
Available now on Amazon in paperback, hardcover, and Kindle.
