Trading strategies work on paper more often than not. They also tend to work in genuinely calm conditions. When the weather turns, they collapse. How do you avoid deploying a fragile system into a hostile market? With a stress test.
A stress test is a form of backtest in which the data deliberately represents difficult conditions. The purpose is to establish whether the trading system holds up under those conditions rather than under average ones. This article covers the three stress tests worth running, organised by the event each one simulates.
Why an Ordinary Backtest Is Not Enough
A standard backtest tells you how a strategy would have performed across a historical period. That is useful, but it contains a structural bias: most historical periods are, by definition, ordinary. Average spreads, average volatility, average liquidity. A strategy optimised against average conditions can look excellent while being completely unequipped for the 5 percent of the time that determines whether an account survives.
A stress test asks a different question. Not what would this have earned? but what would have broken it? That reframing matters, because the failure modes of a trading system are usually not gradual. They are concentrated in specific market states, and if you have not tested those states you do not know your maximum loss — you only know your average one. Techniques like Monte Carlo testing attack the same problem from the statistical side; the three tests below attack it from the market side.
Stress Test 1: The Volatility Shock
The first event worth simulating is a volatility shock. It occurs when the market moves rapidly from a relatively stable phase into one of very wide swings. It can follow an unexpected central bank decision, a macroeconomic release far from expectations, a geopolitical crisis, a bank failure, a sudden collapse in equity indices, or an event nobody modelled at all.
The central point is that during a volatility shock, many variables change at the same time. Candles become far larger, habitual technical stops can turn out to be too tight, targets are hit or missed at high speed, and price can spike in both directions before settling on a clearer orientation.
To stress a strategy against this scenario, deliberately increase the volatility in your tests. Simulate wider bars, stops hit more frequently, worse fills, and more violent intraday movement. You can also select historical periods characterised by elevated volatility and check how the system would have behaved through them — the pandemic-era weeks of March 2020 and the 2022 rate-hiking cycle are both readily available and appropriately unpleasant.
A volatility stress test should answer a specific set of questions. Does the stop loss still make sense when the average daily range doubles? Is position size reduced automatically or does it stay the same? Does the system continue to generate reliable signals, or does it go into hyperactive mode and open too many positions? Are the volatility filters sufficient to keep it out of the most chaotic phases? Our guide to position sizing with ATR covers the mechanism that answers the second of those.
Stress Test 2: The Liquidity Crisis
The second event to test is a liquidity crisis, and it is a different condition from simple volatility. A market can be volatile while remaining liquid, meaning it still absorbs orders at acceptable spreads with reasonable depth. In a liquidity crisis, the problem is that execution itself stops working. Spreads widen, slippage increases, technical levels are crossed without sufficient trading against them, and orders can fill at prices a long way from where they were expected.
For a professional trader this is one of the most important scenarios to simulate, because it directly determines the gap between theoretical and actual results.
A liquidity stress test therefore needs to introduce deliberately worse transaction costs. Average spread is not enough. You have to evaluate what happens if the spread doubles, triples, or widens abruptly during a high-impact event. You have to simulate negative slippage on both entries and exits. And you have to ask whether your stop losses would genuinely fill near the intended level, or whether under difficult conditions they would produce larger losses than modelled.
This matters most for strategies with high trade frequency, scalping systems, intraday methods with small targets, or approaches on less liquid instruments. If average profit per trade is small, even a slight deterioration in execution can turn a profitable strategy into a losing one — a 0.8 pip average edge does not survive a spread that widens by 1 pip.
The questions here are practical. What happens if orders fill 2, 5 or 10 ticks worse than expected? What happens if stops are hit with above-average slippage? Does the strategy retain a positive expectancy once costs increase? Is the trade count high enough that costs become the dominant factor in the result?
Stress Test 3: Sustained Directional Movement
The third event to simulate is strong, persistent direction. This occurs when the market develops a move that is clean, sustained and unwilling to correct. It is a favourable scenario for some trend-following strategies, but it can be extremely damaging for contrarian systems, mean-reverting systems, or anything built on the assumption that price returns quickly to an average.
A great many fragile systems fail exactly here. They work well while the market oscillates, bounces, corrects and respects levels. Then a phase arrives in which price breaks a resistance, keeps rising, ignores overbought signals and offers no meaningful pullback. Or it breaks a support and continues falling with no significant recovery. In those conditions, strategies that keep looking for the reversal accumulate repeated losses — and because each individual loss is within the rules, nothing looks broken until the drawdown is severe.
A directionality stress test needs to verify how the system responds to markets that move in one direction for an extended period. Analyse what happens in the presence of persistent trends, gaps in the same direction, consecutive breaks of technical levels and a complete absence of mean reversion.
A thorough version of this test should include sequences of consecutive adverse moves longer than the historical average. If your worst historical losing streak was six trades, test eight and twelve. The relevant question is not whether that has happened before, but whether the account survives it if it does.
The questions to ask are: does the system allow entry without chasing price too far? Does it have rules for staying in a move, or does it force an early exit? Can it protect profit without exiting at the first pullback?
What to Do With the Results
A stress test that produces uncomfortable results has done its job. The purpose is not to confirm the strategy but to locate the conditions under which it stops working, so that you can either build a defence or accept the exposure knowingly.
In practice the responses fall into three categories. You can add a filter that keeps the system out of the hostile regime — a volatility ceiling, an economic calendar blackout, a minimum liquidity condition. You can adjust position sizing so the same adverse move produces a survivable loss rather than a fatal one. Or you can accept that the strategy has a known weakness and cap the capital allocated to it accordingly.
What you should not do is conclude that the scenario is unlikely enough to ignore. Strategies are rarely destroyed by the conditions they were designed for. They are destroyed by the ones nobody tested — which is the same underlying failure discussed in our articles on spotting a fragile strategy and on why a trading system stops working.
