Every serious discussion of trend following eventually arrives at the same subject: drawdown. It is the most viscerally uncomfortable feature of the approach, the period when an account sits below its previous peak and the question of whether to stay in the strategy or abandon it becomes genuinely difficult. Most investors make the wrong call at exactly the wrong moment. Understanding what drawdown actually is, and what it is not, is one of the most useful things a trend following trader can do before a drawdown arrives rather than during it.
Download the full drawdown report (PDF)
What Drawdown Actually Measures
Winton Capital, one of the most analytically rigorous firms in the systematic trading space, produced a paper examining drawdown as a statistical measure of risk. The paper, “The Pros and Cons of Drawdown as a Statistical Measure of Risk,” was written by David Harding, Georgia Nakou, and Ali Nejjar. Its central argument is that drawdown is both more intuitive and more misleading than most investors realize, and that the conventional use of maximum drawdown as a quality indicator is deeply flawed.
The paper’s first point is foundational:
“As a description of an aspect of historical performance, drawdown has one key positive attribute: it refers to a physical reality, and as such it is less abstract than concepts such as volatility. It represents the amount by which you are less well off than you were; or, put differently, it measures the magnitude of the loss an investor could have incurred by investing with the manager in the past. Managers are obliged to wear their worst historical drawdown like a scarlet letter for the rest of their lives. However, this number is less straightforwardly indicative of manager quality as is often assumed. The seeming solidity of the drawdown statistic dissipates under closer examination, due to a host of limitations which are rarely explored sufficiently when assessing its significance as a guide to the future performance of an investment.”
The phrase “scarlet letter” is precise. A manager who suffered a large drawdown ten years ago and has compounded strongly since is penalized by the raw maximum drawdown figure every time it is cited, even if that drawdown was entirely consistent with the strategy’s parameters and was fully recovered. The number persists regardless of context. Most investors treat it as a verdict when it is really just a data point, and a poorly contextualized one at that.
The Measurement Interval Problem
One of the most practically important insights in the Winton paper concerns how the frequency of measurement distorts the drawdown figure:
“The first is that, all other things being equal, drawdowns will be greater the greater the frequency of the measurement interval. The maximum drawdown will be greater on a daily time series than on a weekly one, and weekly will be greater than monthly. Investments that are marked to market daily, such as managed futures, may thus appear at a disadvantage to less frequently valued investments (e.g. hedge funds).”
This is a structural distortion in how managed futures and trend following strategies are evaluated relative to other alternatives. A private equity fund or a real estate portfolio is not marked to market daily. Its drawdowns, if they could be measured with the same frequency, would look far more severe. The daily transparency of managed futures is a feature for investors who value liquidity and honesty about current valuations. But when that same transparency is used to generate a maximum drawdown figure, it systematically makes the category look worse than investment vehicles that simply report less frequently.
The Track Record Length Problem
The second structural distortion the paper identifies is equally counterintuitive:
“The second is that the maximum drawdown will be greater for a longer time series, so that managers with longer track records will tend to have deeper maximum drawdowns. This effect would have perverse consequences if the raw maximum drawdown were used to measure quality across the board, as, in general, managers that have survived longer have given evidence of professional competence through overcoming such adversities.”
In other words, the managers with the most experience and the longest records of survival are penalized the most by the raw maximum drawdown statistic. A manager with a twenty-year track record has simply had more time for a severe drawdown to occur. A manager with a two-year track record may look better on this metric purely because there has been less time for adverse conditions to materialize. Using raw maximum drawdown to compare these two managers is not analysis. It is noise dressed up as rigor.
Making Drawdown a Useful Statistic
The Winton paper does not argue that drawdown should be ignored. It argues that it must be interpreted correctly to mean anything:
“In order to make drawdown a more informative statistic, we must correct for track record length, measurement interval and volatility; we must take account of the error as well as making sure that we understand the nature of the return generating process (i.e. that it is reasonably parametric). Though some analysts correct for some of these factors, the conventional cursory use of drawdown as a statistic fails most or all of these tests, making it worse than useless.”
“Worse than useless” is strong language from a quantitative firm. The point is that a statistic which appears to convey information but actually misleads the analyst is more dangerous than no statistic at all. An investor who understands they have no information will seek more. An investor who believes they have accurate information when they do not will act on it confidently in the wrong direction.
Drawdown as a Feature, Not a Flaw
The practical implication for anyone trading or evaluating trend following systems is this: drawdowns are not evidence that the system is broken. They are a structural feature of a strategy that cuts losses, holds winners, and does not attempt to predict market direction. A system that produces no drawdowns almost certainly does not produce meaningful returns either, because the only way to eliminate drawdown entirely is to eliminate exposure, and eliminating exposure eliminates profit.
The correct question when evaluating a drawdown is not “how large was it?” but rather “was it consistent with what this strategy is expected to produce, and did the manager follow the rules throughout?” A drawdown that was larger than the strategy’s historical norms, or one during which the manager deviated from their process, tells you something worth knowing. A drawdown that fell within normal parameters, even if it was uncomfortable, tells you only that markets went through a difficult period for trend following, which they have done repeatedly throughout the strategy’s history and will do again. The traders who understood this in advance stayed in. The ones who did not, left at the bottom and missed the recovery.
For context on how trend following systems are structured to manage risk during these periods, see the rules and risk pages. For the broader picture of what the strategy has produced across full market cycles, the Turtle story remains the most documented starting point.
Frequently Asked Questions
What is a drawdown in trading?
A drawdown is the decline in an account’s value from its most recent peak to its subsequent trough, before a new peak is reached. It is expressed as a percentage of the peak value. A 20% drawdown means the account fell 20% from its highest point before recovering. Maximum drawdown refers to the largest such decline over a given period or track record.
How long do drawdowns typically last in trend following?
There is no fixed answer, as drawdown duration depends on market conditions, the specific system, and the portfolio’s diversification. Trend following drawdowns can last months or, in difficult environments, years. The strategy performs best during sustained directional moves in markets. Extended periods where markets chop sideways without trending produce the longest and most testing drawdown periods. Historical track records show that recoveries have followed, but timing cannot be predicted.
Should a large drawdown cause me to exit a trend following system?
The answer depends entirely on whether the drawdown is within the range the system was designed to produce. If a system was built and tested with the expectation of periodic 20-30% drawdowns, experiencing one is not a reason to exit. Exiting at the bottom of a drawdown locks in the loss and removes the investor from the recovery. The harder and more important question is whether the drawdown is consistent with the strategy’s parameters, and whether the manager followed their rules throughout.
Why does measurement frequency affect drawdown figures?
Because prices move between measurement points, and the more frequently you measure, the more intra-period volatility you capture. A daily series captures every dip that a weekly or monthly series smooths over. As the Winton paper explains, this makes daily-marked investments like managed futures appear to have larger drawdowns than less frequently valued alternatives, even if the underlying economic risk is comparable or lower.
Trend Following Systems
Want to learn more and start trading trend following systems? Start here.
