A stop loss is an exit rule that closes a position when a predefined loss boundary is reached. That boundary can limit a trade's planned loss, but it does not guarantee the fill price or make a strategy safer by itself.
For stock stop orders, Investor.gov explains that the trigger becomes a market order and is not a guaranteed execution price; FINRA likewise describes execution at the next available market price, which can differ from the stop (SEC Investor.gov, FINRA). Order mechanics vary by instrument and broker, so check the applicable rules.
The useful question is not whether stops are good. It is whether this stop improves this strategy after its exits, costs and market behaviour are specified.
What a stop-loss rule actually does
A stop-loss rule turns a loss boundary into an instruction to exit. It can be fixed by price, distance, volatility, time or a trailing reference; each version changes when the position closes and therefore changes the strategy being tested.
A stop is not a promise that the loss ends exactly at its trigger. Execution can occur at a different price, and a backtest that assumes a perfect fill may overstate what the rule controls. Define the trigger and the fill model separately (SEC guidance; FINRA order-type guidance).
Why the same stop can help one method and hurt another
Stop-loss results depend on the behaviour the strategy is trying to capture. In a trend-following system, an early exit may cut a losing position; it may also eject a position during a temporary retracement before the trend resumes. In a mean-reverting spread, a stop can cap exposure to a spread that keeps moving away, but it can also close a position before the modelled reversion.
That distinction appears in research using different setups. Clare and co-authors report that popular stop-loss rules did not add value to their S&P 500 200-day moving-average strategy, while Li's mathematical study of a mean-reverting spread models entry and exit jointly with transaction costs and a stop-loss constraint (Clare et al., Breaking into the blackbox, Li, Optimal Mean Reversion Trading with Transaction Costs and Stop-Loss Exit). These are not universal verdicts: one is a particular historical index-rule comparison; the other derives results inside a specified stochastic model.
Common stop designs, different assumptions
The label “stop loss” hides several distinct rules. Before comparing performance, write down what information the rule uses and when it can act.
| Design | Exit condition | Assumption to challenge |
|---|---|---|
| Fixed-price stop | Price crosses a chosen level | The distance remains meaningful as volatility changes |
| Volatility-scaled stop | Price moves a chosen volatility-based distance | The volatility estimate is available at decision time and stable enough to use |
| Trailing stop | A reference follows favourable movement and exits after a reversal | The trailing reference does not turn ordinary retracements into premature exits |
| Time stop | The position closes after a defined holding interval | The strategy's edge should have appeared within that interval |
The table describes rule families, not a ranking. A tighter stop may reduce the loss on some trades while increasing the number of exits and changing the distribution of subsequent returns. You need the complete trade sequence to see the net effect.
What the evidence can—and cannot—settle
The evidence is conditional, not a licence to add a stop to every system. Clare and co-authors compare rules on the S&P 500 and find that popular stops did not improve their moving-average strategy. Li studies a different question: an optimal mean-reversion entry and exit problem with transaction costs, where the stop-loss level affects the modelled entry and take-profit boundaries (City Research Online study, arXiv paper). The second result is a model-based result, not proof that a live strategy will behave the same way.
The studies differ in market, rule and method, so they do not identify one best stop distance. Historical results also depend on how the rule handles gaps, intrabar movement, spreads, slippage and the choice of test period. A stop selected after inspecting the losing trades is especially hard to distinguish from hindsight.
How you'd actually test it
Test the stop as a complete exit policy against a baseline with the same entries, data and execution assumptions. Keep the comparison fair: changing entry rules or costs at the same time makes it unclear what caused any difference.
- Specify the entry and baseline exit before looking at the results.
- Define the stop trigger, order type and fill rule. State how gaps and intrabar touches are handled.
- Compare no stop with a small, predeclared set of stop variants; choose parameters using only the design period.
- Charge spread, commission and plausible slippage consistently, including on stop exits.
- Measure more than the average trade: inspect drawdown, tail losses, exit frequency, missed recoveries and sensitivity to nearby settings.
- Freeze the chosen rule and evaluate it on data not used to select it. Report if the benefit disappears out of sample.
For a cTrader workflow, the same discipline applies to symbol data, spread and execution assumptions. The checklist in how to verify a cTrader backtest helps make those assumptions inspectable. realbacktesting is a trading-software studio for cTrader built around results a trader can reproduce and check, not a stop setting that is presumed to work.
Takeaway
A stop loss makes an exit rule explicit. Only a controlled comparison can show whether that rule improves the strategy it belongs to.