Take-profit and stop-loss: how to set them correctly

Most blown accounts don't break on the entry. You can take an entry almost at random and still close the month green, if the exit was worked out in advance. Or you can perfectly catch a reversal off a density level, then drag your stop twice and ride into liquidation.

Take-profit and stop-loss turn a set of random clicks into a system with measurable expectancy.

Below: three ways to calculate the distance to your stop, how position size follows from it, and why a declared 1:3 R:R turns into 1:2 in practice.

What a take-profit and a stop-loss are

A take-profit closes a position in profit once a set price is reached. A stop-loss closes the same position at a loss when price moves against it by a distance you defined in advance. Both orders go in right after the entry and work without the trader.

The difference in mechanics matters more than it seems. The take sits in the order book as an ordinary limit order. The stop-loss sits on the exchange servers as a conditional order, invisible to everyone else, and turns into a market order only when it triggers.

What they're for

The first job of a stop is to cap the loss at the amount you decided to lose before entering. Not the one that still feels "bearable" in the moment.

The second job is less obvious. The stop sets the unit of risk measurement that everything else is counted in. One risk (1R) is your stop in money. A 2R profit means you took two stops. Drawdown and position size are counted in R, not in dollars: dollars depend on the account, R doesn't.

An account of 5,000 USDT, 1% risk per trade. So 1R = 50 USDT. After that it doesn't matter whether you trade BTC with a 0.3% stop or a small alt with a 4% stop. Size is fitted so that the stop costs exactly 50 USDT. How to tie this percentage to your overall capital was covered in the article "Risk Management".

The third job is to take the decision away from you at the moment when there's no time left to think. The position is in the red, the tape is accelerating against you, your hands are cold. An order placed with a cool head thirty seconds ago handles it better.

The take works symmetrically: it locks in the profit you planned, not the one greed will let you take ten minutes later. The market regularly gives 0.9% and takes 0.7% of it back a minute later.

How they work on the exchange

A futures stop-loss comes in two types. A stop-market triggers at the trigger price and closes the position with a market order: execution is guaranteed, the price isn't. On a thin alt it will easily slip 0.3-0.5% past the trigger. A stop-limit places a limit order at the stated price, and if the market flew through that level on an impulse, the order just hangs there and the position stays open. On a knife that ends badly.

For protecting a position you almost always take a stop-market. Slippage of 0.2% is unpleasant. An unclosed position during a liquidation cascade is a disaster. The mechanics of the other order types were covered separately in "Order Types on a Crypto Exchange".

A separate point is which price the trigger is measured against. Most exchanges use mark price (the index price) by default, not last price. Mark price smooths out local spikes and protects you from being taken out on an anomalous tick, last price reacts faster but takes you out more often. Do your stops keep triggering on moves that don't exist on other exchanges? Check this setting first.

Fees are part of the mechanics too. On Binance USDⓈ-M futures the maker fee is 0.02%, the taker fee 0.05%. A market entry plus a stop-market exit gives 0.1% of turnover. With a 0.3% stop that's a third of your risk handed to the exchange. A take as a limit order costs 0.02% and saves you the difference.

If the basic order mechanics are still fuzzy, watch the free lesson from the course "Trading From Scratch | free crypto trading and scalping course": it walks through connecting an exchange via API and working with orders step by step.

How to set a stop-loss correctly

How to set a stop-loss correctly, if you strip away all the particulars? The stop goes where your trade idea stops being valid. Not where the money feels comfortable. First you find the point where the scenario is invalidated, then position size is calculated from the distance to it.

There's one formula and it works for any style:

Position size = (Account × Risk in %) ÷ Distance to stop in %

An account of 5,000 USDT, 1% risk (50 USDT), a stop 0.8% away. Size = 50 ÷ 0.008 = 6,250 USDT, at 10x leverage that's 625 USDT of margin. The same account with a 2.5% stop gives a size of 2,000 USDT, and the risk in money doesn't change.

The reverse logic ("I want to enter with 10,000, so I'll put the stop closer") is the main reason for serial stop-outs. There's a separate breakdown on distributing capital across trades in "Money Management in Trading: Managing Your Capital".

By levels

The stop hides behind a structural element that protects your idea.

For a long off support — below the low of the candle that produced the reaction, plus a buffer. For a long on a breakout — below the breakout level or below the edge of the range price left. For a short it's all mirrored.

The buffer is mandatory. A stop exactly on the low sits in the same place as the whole crowd's stops, and the market collects those clusters regularly. A sensible buffer: 0.05-0.15% for liquid pairs and up to 0.4% for volatile alts.

An example. BTC/USDT, a range of 67 180 - 67 420, a long entry at 67 240 on the break upward. The low of the range is 67 180, buffer 40 points, stop at 67 140. Distance 100 points = 0.149%.

Round numbers are a separate conversation. Large limit orders pile up on them, and price is drawn to them like a magnet. Putting a stop exactly at 67 000 or 3 500 is a bad idea. Working with horizontal zones is covered in detail in "Support and Resistance Levels".

By ATR (Average True Range)

ATR shows the average candle range over a period, usually 14 bars. The method adapts to current volatility instead of a fixed percentage. The working formula: a stop 1.5-2 ATR away from the entry price on your timeframe.

An example. BTC/USDT, 5-minute chart, ATR(14) = 180 points. A stop at 2 ATR = 360 points, about 0.53% at a price of 67 000. An hour later volatility has risen, ATR = 310, and the same multiplier now gives 620 points. A position sized for that stop automatically becomes half as large at the same dollar risk.

The second use is more valuable than the first: ATR works as a sanity filter for a stop calculated by levels. A structural stop smaller than 0.5 ATR will almost certainly get taken out by ordinary noise, even when the direction was called right.

The weak spot of the method: it knows nothing about structure and can put the stop in the middle of a zone where a large limit order is sitting. The level gives you the place, ATR gives you the check.

Behind a density level in the order book

A method for those who trade order flow, not pictures.

A density level in the order book is a large limit order that can act as support or resistance. If there's a 10 million dollar buy order sitting under price while the average cluster volume on that coin is 3-5 million, that density won't be eaten in one move. So the stop goes behind it, not in front of it and not inside it.

An example from practice. ADA, a 5 million dollar buy-side density in the order book, sitting there for 30 minutes already. A long entry as price approaches that zone, stop 0.2-0.3% below the density. The idea is invalidated not when price touches the order, but when it's punched clean through.

Same with the liquidation map. A bright cluster at 95 000 — stop at 94 800 or 94 500, outside the zone where the impulse runs out of air. A stop inside the cluster is a voluntary donation to the market maker's fund.

The key check is whether the order is real. In Secret Terminal a lifetime timer is shown next to large volume in the order book. A density that's been standing for a while confirms a large player's interest. An order that appears and disappears as price approaches is spoofing (faking demand with orders that get pulled before execution). We also look at the clusters: price came in but the volume didn't go through — the order was simply pulled.

I usually keep the stop behind the density, but if it gets pulled before price arrives and the timer resets, I exit manually. The argument I entered on is gone.

How to calculate a take-profit

The take goes where someone else's liquidity is sitting, and then it's checked against the risk-to-reward ratio. Fails the check — the trade isn't taken. The reverse route ("I need a 1:3 R:R, so I'll put the take three stops away") gives you pretty numbers and unreachable targets.

Where to look for the target:

• A large density level in the order book on the opposite side. Place the limit order slightly in front of it, otherwise you'll be queued behind an order worth several million.
• A round number, if it's a reasonable distance away.
• The edge of the range or a previous extreme on a higher timeframe.
• The POC (Point of Control) of a higher-timeframe candle, meaning the level with the most traded volume, which price returns to.
• A liquidation cluster. Price reaches it on momentum.

The Risk:Reward ratio (1:2, 1:3)

R:R shows how many units of profit you take per unit of risk. A 100-point stop and a 300-point take is 1:3. On its own that number says nothing, it only works paired with your win rate. The minimum win rate to break even is calculated as 1 ÷ (1 + R).

This shows why a scalper and a swing trader have different systems. An order book scalper works with an R:R around 1:1 but runs a 60-70% win rate: he sees the order flow. A swing trader keeps a 30% win rate, but one trade pays for four losers. Both models work. What doesn't work is a 35% win rate with a 1:1 R:R.

Expectancy = (Win rate × Average profit) − ((1 − Win rate) × Average loss)

Win rate 40%, average profit 2R, average loss 1R. Expectancy = 0.4 × 2 − 0.6 × 1 = 0.2R per trade. At 50 USDT of risk that's 10 USDT expected on every entry, around 200 USDT over twenty trades a week.

Partial exits

Closing the whole position with a single order isn't mandatory. In pieces it comes out steadier.

The classic scheme: half at the first target, drag the rest further. On knives and spikes, where price has shot out sharply and starts coming back, it makes sense to lock in at least half on a retracement of roughly 50% of the move.

Let's run the numbers. A 10,000 USDT position, a 0.5% stop (50 USDT of risk), the first target at 1R, the second at 3R. Closing the whole position at 1R gives 50 USDT, at 3R — 150 USDT, but it gets there far less often. A 50/50 scheme with both targets hit gives 25 + 75 = 100 USDT, and if price only reached the first one and came back to breakeven, you get 25 USDT instead of zero.

Average R:R drops with partial exits. Win rate and the smoothness of the curve go up. It's a good trade: a run of six losses in a row breaks your psyche faster than your account.

A mandatory addition: after the first partial the stop moves to breakeven. From there the trade either closes at zero or makes it to the second target, but there won't be a loss any more.

Trailing take-profit

Strictly speaking this isn't a take but a way of managing the position: the fixed target is replaced by a moving stop that crawls along behind price.

Three variants that work:

• By ATR. The stop stays 1.5-2 ATR away from the high of the move. Simple, mechanical, works well in a trend.
• By candles. After each candle closes, the stop moves under its low (for a long). More aggressive than ATR, gets knocked out more often.
• By density levels. As soon as price pushes through another density level in the order book, the stop is pulled up behind it. The most sensible option for order book trading, but it requires you to sit at the screen.

Where doesn't a trailing stop work? In chop. On the BTC 5-minute in a range it consistently eats more than it gives me: it knocks me out on every pullback, and there simply is no 3R move that day. On alts on listing day it's the opposite — there a fixed take leaves half the move on the table.

The rule: the trailing stop switches on after the position has covered at least 1R, and only when there's momentum context.

Stop-loss and take-profit: mistakes

Three mistakes account for most of a beginner's losses. All three are fixed with discipline, not with new indicators.

A stop that's too tight

The symptom: you get taken out, and then price goes exactly where you expected. Again and again.

The cause is almost always inverted calculation logic. The trader wants to enter with 10,000 USDT, but 1% risk on the account only allows 3,000. Instead of reducing size he cuts the stop from 1% to 0.3%, and the stop ends up inside ordinary market noise.

One formula to check it: divide the distance to your stop by the ATR of your timeframe. Below 0.5 — you'll get taken out almost guaranteed. From 1.0 to 2.0 — the working range. Above 3 — you're too far from the entry and losing R:R.

Specifics. A coin with a 1-minute ATR(14) of 0.6%, stop placed at 0.15%. An ordinary one-minute candle covers that stop four times over. Direction has nothing to do with it.

The second source of the problem is a wide spread. When the order book is empty and the gap between the best bid and the best ask is 0.3-0.5%, a 0.4% stop triggers on a single trade. On instruments like that either widen the distance or don't trade them at all.

Moving the stop "in hope"

The most expensive mistake. The position is in the red, ten points left to the stop, and your hand reaches out to push it "just a bit further, it's about to turn".

The math is merciless. Drag the stop once and you exit at a 3R loss instead of 1R. To get back to where you started you need three profitable trades in a row. One violation wipes out several days of results, and if you drag it all the way you get a liquidation: the position is leveraged.

The worst part is that sometimes it works. Price turns, you close in profit, and your brain records "dragging the stop pays". Five trades later that learned reflex takes the whole account.

I'll admit it honestly, in my first year I killed two accounts in exactly this way. Not with entries, not with a lack of strategy. With two mouse movements at the wrong moment.

The rule is simple. The stop moves only in the direction of reducing the loss. Arguments against the position show up — exit manually right now, don't wait for the stop.

A trading journal helps you track violations: open a losing week in the performance calendar and count how many times the final loss exceeded the planned one. It usually sobers you up more than any advice.

No take-profit at all

The mirror mistake. The stop is set, the target isn't, because "let profits run".

In practice "let it run" looks like this. The position goes 1.2% into profit, the trader waits for 2%, price pulls back to 0.4%, he decides to wait for it to come back, price goes negative, the trade closes on the stop. A profitable entry turned into a loss without a single wrong call on the analysis.

The problem isn't even the money lost. Without a target defined in advance the trade has no success criterion, which means there's nothing to count: no average R:R, no win rate, no expectancy.

The minimum solution is an auto-take placed at a set distance right after the entry fills, together with an auto-stop. Even a rough target beats not having one.

Tempted to hold longer? Use partial exits. Half goes according to plan, the rest goes on a trailing stop.

A breakdown of how large participants hunt exactly these clusters of stops is in the free lesson from the course "Trading From Scratch | free crypto trading and scalping course". Useful if your stops get taken out regularly.

FAQ

• What percentage should I set my stop-loss at?

  There's no universal number, the instrument's volatility sets the reference. For BTC intraday the working range is 0.3-1%, for mid-cap alts 1-3%, for fresh listings 3-7%. You should calculate from the point where the idea is invalidated, not from a percentage, and then check the result against ATR. Below 0.5 ATR of your timeframe — the stop is too tight.

• Can you trade without a stop-loss?

  On leveraged futures, no. A position without a stop gets closed by liquidation, and the exchange decides the price of that close, not you. On spot you formally can, because the asset stays yours, but then the role of the stop is played by your willingness to hold a 50-70% drawdown for an unlimited time. Most people don't have that willingness, they only have the illusion of it until the first serious dump.

• Which is better, a stop-market or a stop-limit?

  For protecting a position, a stop-market: it will always fill, even if with slippage. A stop-limit can go unfilled if price flew through the level on an impulse, and then you're left in the position in the middle of a move against you. It only makes sense for entries in a calm market, where the price matters more than the fact of execution.

• Why does my stop trigger and then price immediately goes my way?

  Most often the stop sits in the same place as everyone else's: under an obvious low or on a round number. Those clusters serve as liquidity for large players, so price collects them and turns around. It's cured by a 0.05-0.4% buffer behind the structural level and the habit of checking the liquidation map before entering. The second reason is simpler: the stop triggers on last price during a local spike, switching to mark price helps.

• How do I calculate a take-profit if there are no levels nearby?

  Start from the minimum acceptable R:R for your style and from ATR. A working 1:2 R:R with a 0.4% stop gives a target at 0.8%. Then check against ATR: when 0.8% fits inside one average candle range on your timeframe, the target is realistic. If you need four ATR in a row in one direction, the trade is better skipped.

• What R:R counts as normal?

  The one where your real win rate produces positive expectancy. For order book scalping it's fine to work at 1:1 with a win rate above 55%. For swing trading you need 1:3 and up, because the win rate there rarely exceeds 35%. More important than the number itself is that R:R has to be counted after fees, not before: on short stops the difference reaches 30%.

• Should I move the stop to breakeven?

  After the first target is hit or 1R is covered, yes, it protects the result. Before that, no. Moving to breakeven early turns normal trades into zeros, because price almost always pulls back to the entry before continuing the move. Check your own journal for how many trades closed at zero and where price went afterwards. The number is usually unpleasant.

Take control of your exits

A stop-loss and a take-profit stop being a formality once you can see what's happening in the order book at the moment they trigger. Whether the density is real or spoofing. Whether volume is going through in the cluster. Who's more aggressive on the tape right now.

Secret Terminal brings this together in one window: an order book with an order lifetime timer, clusters with POC highlighting, a tape with a minimum trade size filter, and a density monitor across every connected exchange. Protective orders are placed with a hotkey straight from the order book and dragged around the chart with the mouse, snapping to candle extremes, and the journal will later show you where you strayed from the plan. Works with Binance, Bybit, OKX, MEXC and WhiteBIT via API keys, Windows 64-bit version, free.