What is a stop-loss order?
Most explanations end at “decide where you will get out.” The harder question is which chart that level comes from. A stop borrowed from a smaller timeframe than your idea is not a tighter version of the same trade — it is a different trade, with different leverage.
What does a stop-loss actually do?
It converts an unknown loss into a known one. You place the order where your reason for being in the trade stops being true; if price touches it, the position closes — usually at market. Because the maximum loss is a number before you enter, position sizing becomes arithmetic: risk = distance to stop × position size. No stop, no distance, no size, no plan.
How is a stop-loss different from liquidation and a “mental stop”?
A stop-loss executes at your level and costs a normal fee. Liquidation executes at the margin maths’ level and costs your margin. A “mental stop” executes at your emotions’ level, which under pressure means late or never. Livermore’s confession applies: the loss costs little; refusing to take it is what does the damage.
What happens between the stop price and the fill?
Two different prices, and it costs beginners money to assume they are one. The trigger is a condition you set; the fill is a transaction that happens afterwards, against whatever liquidity is actually there at that moment. Until the trigger prints, nothing of yours is resting in the order book — the exchange is only watching a number on your behalf.
| What you set | What the trigger releases | Guarantees | Does not guarantee |
|---|---|---|---|
| Stop-market | A market order | You get out | The price you get out at |
| Stop-limit | A limit order at a price you choose | The worst price you will accept | That you get out at all |
| Trailing stop | A trigger that follows price by a set distance | The distance is maintained | That the distance survives ordinary noise |
| Price alert | A notification | That you are told | Everything else — you still have to act |
Put the same trade through both. Entry $60,000, invalidation $58,200, position 0.0556 BTC, budget $100 — the trade used throughout this page.
- Stop-market. The trigger prints at $58,200 and the market order fills at an average of $58,080, two-tenths of a percent lower. Loss: 0.0556 × $1,920 = $107. You paid about $7 over budget for the certainty of being out.
- Stop-limit, $100 band. Trigger $58,200, limit $58,100 — a band of 0.17%. Price goes straight through both and keeps going to $56,000 without ever trading back up to your limit. Nothing fills. You are still long, and the loss on screen is 0.0556 × $4,000 = $222.
Read those two numbers together. The stop-limit did exactly what it promised — it refused a bad price — and the refusal cost 2.2× the risk budget, in precisely the fast move the stop existed for. A limit band protects you in the ordinary case you did not need protecting in, and steps aside in the violent one where you did. That is the argument for leaving a protective stop as a market exit, and it is arithmetic rather than preference.
Which price is the exchange even watching?
A setting most people never open. On derivatives venues the trigger can reference the last traded price, an index price, or a mark price built from prices across several exchanges — Bybit’s help centre lists all three as trigger references for take-profit and stop-loss orders (as of September 2026; check what your own venue defaults to, as these rules change). The mark price exists specifically to resist single-venue wicks, so a thin spike that prints on last price may never touch the mark. Same level, same market, two settings: one fires, one does not. Neither is wrong, but you should know which one you picked.
Bybit documents a sharper edge on paired orders too: when a take-profit and a stop-loss are set as a limit pair, the other order is cancelled the moment the first one is triggered — not when it fills. If that triggered limit order then never fills, the position can be left with neither a stop nor a target still attached (as of September 2026). It is small print with a real cost, and one more reason to prefer a market exit for the protective side. Setting one on Bybit, step by step.
Where this advice stops being right: if you are exiting a large position into a thin book, the market order is the thing that hurts you — it walks the book and prints the bad fill itself. Size decides which risk is bigger. Below the depth sitting in the book, take the fill; above it, a limit band and a plan to work the order start to earn their keep. Judge that from the order book, not from a rule.
A related question, answered elsewhere. This section is about which timeframe the level is read from. Which object on that timeframe it should be anchored to — trendline, swing low, zone edge or invalidation price — and what it costs when those four disagree, is covered in Lesson 43 on stop placement.
Which chart should the stop come from?
A stop is a bet that ordinary movement on your chart stays outside it, and ordinary movement scales roughly with the square root of time — the assumption behind our timeframes lesson. A 15-minute candle covers about √(15/240) = 25% of the ground a 4-hour candle covers. Take a 4-hour idea invalidated 3.00% below entry, $10,000 account, 1% risk ($100), and move only the chart the level came from:
| Stop taken from | Share of 4-hour range | Stop distance | Position at 1% risk | Position ÷ equity |
|---|---|---|---|---|
| 4-hour the idea’s own chart | 1.000 | 3.00% | $3,333 | 0.33× |
| 1-hour | 0.500 | 1.50% | $6,667 | 0.67× |
| 15-minute | 0.250 | 0.75% | $13,333 | 1.33× |
| 5-minute | 0.144 | 0.43% | $23,095 | 2.31× |
| 1-minute | 0.065 | 0.19% | $51,640 | 5.16× |
Read the last column. The idea never changed — only the chart the level was copied from — yet the position went from a third of the account to five times it. A tight stop is a leverage decision wearing a safety costume. The multiple is the inverse of the range ratio, so every chart down roughly doubles the position.
When should you move a stop, and what does moving it cost?
Move it only after the pullback testing your idea has finished and price has turned back up. While price is still falling there is no low yet to put the stop under, so an early trail is a guess dressed as risk management. And trailing is not free: it does not shrink the position you hold, it shrinks that position’s room.
Same trade: entry $60,000, invalidation $58,200, position $3,333 (0.0556 BTC). Price runs to $61,800 and you trail to the last 15-minute push low, $61,337 — 0.75% below. The position is unchanged; the give-back you tolerate is now $25 from current price instead of the $100 you budgeted. If it fires you bank about $74 — a fine outcome, just a smaller trade than the one you sized. Keep the trailing chart adjacent to the entry chart: a 4-hour idea trails on the 1-hour and still keeps half its room. On the 1-minute it keeps 6.5%, and a 1-minute break says nothing about whether a 4-hour trend is over. If you would rather the exchange do the moving, a trailing stop automates it — at the price of fixing your give-back in advance, before you know how deep this trend’s pullbacks will be.
When is “always use a tight stop” the wrong advice?
When the small chart is your chart. Enter on the 5-minute and there is no mismatch — 0.43% is the right stop for a 0.43% idea. The error is mixing charts, not using small ones.
When fees dominate. Tight stops force large positions, and fees scale with position while the risk budget stays fixed. Behind a 0.5% stop, round-trip fees eat about 40% of the budget before the market moves — arithmetic on the position sizing page.
When the stop stands in for size. Halving the stop to double the position is the same risk with less room. If the trade is not worth taking at 1% behind a 3% stop, 0.75% does not improve it.
A stop is also not a guarantee of price: in a gap or a cascade it fills wherever the book is — slippage — so leveraged positions keep the stop far inside the liquidation distance.
FAQ
Can a stop-loss fail? It can fill worse than its trigger price in gaps or cascades — that is slippage, not failure. On liquid markets in normal conditions it will not be skipped entirely.
Stop-market or stop-limit? Beginners: stop-market — a bad fill beats no fill. A stop-limit can be skipped entirely in a fast move, which defeats the purpose.
Should I move my stop to break-even? Only once the chart gives you a structure to hang it on. Moving to break-even because you are up “enough” puts the stop at a price the market has no reason to respect, usually inside ordinary noise. Break-even is a location on your profit and loss, not on the chart.
Which price triggers my stop — last price or mark price? Whichever your venue is set to, which is a setting most people never open. A thin wick can print a last price that the mark price never reaches, so the same level fires on one setting and sits still on the other. Check the default before you rely on the level.
Placing one takes an account and thirty seconds
A stop-loss is only protection once it is resting on the exchange. Where it sits relative to liquidation depends on the venue’s margin rules.
Compare the three exchanges we use →
That page carries referral links and says so up front — including what each venue does badly.
Every key term, one roadmap
The whole slide course — ten free PDF parts, 351 pages.