Stop-Loss Discipline: The Decision You Make Early

The short answer

The decision that matters about a stop is made before the position exists, when nothing is at stake and the level is a dispassionate judgement about where the idea would be wrong. Everything afterwards is execution.

This is why "stop-loss discipline" is a slightly misleading phrase. It suggests the skill is the moment of honouring the stop — the willpower to let it hit. In practice, if honouring it requires willpower, something has already gone wrong upstream: the level was placed to suit a position size rather than the market, or the size was too large for the distance, or the reason for the trade was never written down clearly enough to be falsified.

There is also a definitional problem that makes most writing on this subject unreliable. Two different things are called a stop-loss, and findings about one are quoted about the other.

Two different objects

A stop order on a single trade. A level, named in advance, at which the idea is abandoned. Its justification is logical rather than empirical: if you cannot say where you would be wrong, you have not made a falsifiable decision.

A stop-loss rule over a portfolio or account. A policy that reduces exposure after cumulative losses pass a threshold — stop trading for the week, cut size by half, go to cash. This is a different object: it operates on the whole account, over time, and it is something that can be backtested.

The second has been studied. Kaminski and Lo proposed an analytical framework to measure the value added or subtracted by stop-loss rules, which they define as predetermined policies that reduce a portfolio's exposure after reaching a certain threshold of cumulative losses, and tested the effect on the expected return and volatility of a portfolio strategy. Using daily futures price data, they analysed stop-loss policies applied to a buy-and-hold strategy in index futures.

Their finding is conditional, and the condition is the interesting part: at longer sampling frequencies, certain stop-loss policies can increase expected return while substantially reducing volatility, consistent with their objectives in practical applications.

Read that carefully. It is a result about particular policies, at particular horizons, applied to a buy-and-hold index strategy. It is not a finding that stop orders on individual discretionary trades improve outcomes, and it should not be cited as one. We have not found a study that establishes the latter.

Where the decision actually is

If you want the stop to hold, these are the things to settle first, in this order.

1. What would make this idea wrong? Answer in a sentence before you look at sizing. If the honest answer is "nothing in particular, I just think it goes up", you do not have a trade with an invalidation level — you have a directional opinion, and a stop placed on it is arbitrary.
2. Where is that, as a level? From the market: structure, a range boundary, a failure point. Not from your account, and not from a round number of points.
3. Given that distance, what size? The division is set out in how much a mistake is allowed to cost. If the resulting size is uncomfortably small, the correct response is a smaller position, not a nearer stop.
4. What is the response if it is reached? Written as an if-then, which is the form with evidence behind it — see what a plan has to decide in advance.

Everything people describe as needing discipline is downstream of getting steps one to three right. A stop you placed for reasons you could articulate is far easier to honour than one you placed because the size demanded it.

Why moving it feels rational at the time

Nobody moves a stop while thinking "I am abandoning my method". The reasoning at the moment is always local and always plausible: the level was a bit tight, the move was noise, the news is out now, it only needs a little more room.

What the evidence adds is a specific warning about when that reasoning happens. Imas found that risk-taking after losses splits on whether the loss has been realised: following a realised loss, people took less risk; where the same loss was left unrealised — still open, still on the screen — they took more. The moment you are deciding whether to widen a stop is exactly the unrealised-loss state. The research says that is when appetite for risk is highest.

That is the most useful thing we can tell you about stop discipline, and it is not about willpower. It is that the decision is being made in the one state the literature identifies as adverse, which is an argument for having made it earlier rather than for trying harder now.

The backstop that is not a stop

On a leveraged retail account in Europe, something will eventually close the position whether you do or not. ESMA's product intervention measures standardised the margin close-out at 50 per cent of the minimum required margin, and added negative balance protection on a per-account basis as an overall guaranteed limit on retail client losses.

Both are consumer protections and both are useful. Neither is a risk policy. If the close-out is what ends your position, the stop level, the size and the plan have all already failed; what is left is the regulator's floor. The context for those rules is that national authorities found 74 to 89 per cent of retail investor accounts typically lose money on CFDs, with average losses per client ranging from €1,600 to €29,000 — which is the population these backstops were designed for.

What to measure

The question "am I disciplined with my stops?" has no useful answer. These do, and they are all countable from a trade log:

• How many positions closed at the level you named, versus at a level you chose later.
• The distribution of your losses. A tight cluster means your stops are holding; a long tail means they are not, whatever you remember.
• Whether moved stops were moved in one direction. Almost always they are, and the asymmetry is the finding.
• What happened after. Not to prove the move was wrong — sometimes it was not — but because a habit that pays off occasionally and costs heavily otherwise is exactly the kind that survives on memory and dies on records.

None of that works without having written the level down before entry, which is the practical argument for keeping a record of decisions rather than of outcomes. The broader evidence on how losses change subsequent decisions is in the hour after a loss and what a losing run does to judgement, and the whole field is mapped in our survey of what is actually known about trader behaviour.

Trading leveraged products carries a high risk of losing money quickly. Nothing here is advice, and no level, distance or percentage is suggested anywhere on this page.