Moving the stop away from price to avoid taking the loss
The one modification that turns a controlled loss into an uncontrolled one. Every risk figure on the account rested on that number.
A stop-loss is a price level attached to a position at which the broker closes it automatically, capping the loss on that trade. It lives on the broker's server, so it works even when the terminal does not.
also: SL, hard stop, protective stop
Updated · Reviewed
The price at which you have agreed in advance to be wrong. Setting it turns an open position into a known cost.
A stop-loss fixes the worst case before the market gets a say. Position size, risk/reward and expected drawdown all follow from that one distance. Without it a strategy reports the losses that closed, not the losses it can take.
| Setting | Typical value | Note |
|---|---|---|
| ATR multiple (volatility stop) | 1.5–2.5 × ATR(14) | A recommendation, not a census — no record here is set this way. Below 1.5 the stop sits inside ordinary movement. |
| Shipped stop, 16 single-stop EAs here | 25–114 pips, median 100 | What these builds actually send. Five of the sixteen ship exactly 114. |
| Risk per trade | 0.5–2% of balance | This and the stop distance fix the lot size. Choosing lot size first inverts the logic. |
| Minimum distance from price | broker's stops level | Market Watch → Specification prints it. A stop inside it never reaches the server. |
| Break-even move | after 1 × the stop distance in profit | Common, and not free: it turns some winners into scratch trades on a retrace. |
Nobody picks a stop distance alone. It goes with the risk limit, and the pair produce the lot size.
Widen the stop to 80 pips and the size must halve to 0.125, or risk doubles.
Calculation 100 ÷ (40 × 10) = 0.25 lots
Result A 40-pip stop costing exactly $100 if it is hit
Read a stop as a pair — the distance and the size that goes with it. Either alone says nothing about the risk.
| Range | What it means |
|---|---|
| Stop inside 1 × ATR | Ordinary movement closes the trade; the win rate falls for reasons unrelated to entry. |
| 1.5–2.5 × ATR, size adjusted | Outside the noise, and sized so being wrong costs a planned amount. |
| Wide stop, unchanged lot size | Risk per trade grew in proportion to the widening. |
| No stop at all | The worst case is the account; drawdown covers only the losses that closed. |
Every EA here states its stop rule and publishes its trade ledger, so the realised loss sits next to the implied stop. The measured figures here come from those backtest ledgers (EXP-STOP-OVERSHOOT-001).
The one modification that turns a controlled loss into an uncontrolled one. Every risk figure on the account rested on that number.
It is a trigger, not a promise: gaps fill through the level and the position closes at the next available price. A stop-limit fixes the price instead, at the risk of no fill; MT5 offers one, MT4 does not.
Thirty pips is generous in a quiet market and inside a single bar in a volatile one.
A stop placed at the distance a too-large position tolerates sits inside the noise. The size should give way.
Every listing ships its closed-trade list, so each exit distance sits beside the stop it carries. Sixteen of the 27 records send one distance on every order; a loser below means price closed against the entry.
| 16 uniform-stop records, 2,719 losing trades | Figure |
|---|---|
| Losses closing within ±2% of the shipped stop | 81.98% (2,229) |
| Losses closing past it | 4.34% (118) |
| Records whose deepest loss beat the setting by half again | 9 of 16 |
| Median deepest overshoot per record | 55.65% |
| Deepest single loss | 316.6 pips against a 114-pip stop |
| Friday entry, non-Friday close: all losses vs overshoots | 12.69% vs 18.64% (22) |
Row one is the stop working: four losers in five land on the preset’s number. Row two is the part no setting controls. The last row is thinner than it looks: count every hold spanning a Saturday or Sunday instead and the shares converge, 33.36% against 35.59%. Rows three and four are thin too: Sundial’s deepest loss, 39.0 pips against a 25-pip stop, is the only zero-elapsed row that is a loss, and dropping all three makes them 8 of 16 and 44.06%. Sundial’s timed close usually gets there first, median loss 8.1 pips — a faster exit moves the typical loss, not the worst.
One sizing rule follows: multiply the shipped stop by 1.56, the median worst case across these 16 records, before testing it against your per-trade limit; half beat it. Scale the distance from measured volatility, and read the stop any expert advisor ships off its own page in the catalogue.