ATR position sizing in crypto: turn a range into quantity

ATR position sizing in crypto converts a volatility-based stop distance into a trade quantity: divide the planned loss budget by the ATR-based distance plus a cost allowance, then apply exchange constraints. Use a clearly identified ATR observation and recalculate size before entry when that observation changes. The result is a modeled exposure, not a guarantee that an exit will cap the loss.

This AI-assisted educational guide uses an original hypothetical sizing ledger. It answers a narrow question: how does an observed range become a quantity you can audit, and what breaks when the range changes? No ATR length, multiplier or risk percentage below is a recommendation or a demonstrated trading advantage.

Define the ATR reading before using it

Average True Range measures the magnitude of recent price movement rather than its direction. For each candle, true range is the largest of the high minus low, the absolute high minus previous close, and the absolute low minus previous close. ATR smooths that sequence. TradingView's ATR documentation describes these inputs and its default RMA smoothing, with other smoothing options available.

Record the venue, instrument, candle interval, length, smoothing method and observation time. A length setting of 14 on hourly candles uses hourly bars; it does not become 14 daily candles. Different smoothing, initialization history or data sources can produce different readings even when both displays say ATR 14.

ATR is expressed in the instrument's price units. An ATR of 30 on an ETH/USDT chart means 30 USDT per ETH; it does not mean 30 percent. For comparisons across differently priced assets, define a normalized measure such as ATR / reference price × 100. That ratio still measures past movement, not the chance of hitting a stop.

Use the multi-timeframe availability worksheet to identify which completed observation was available at the decision. A developing candle can change the ATR before it closes. Mixing its latest live reading with a historical completed reading creates a different experiment.

Convert the distance into units, then apply limits

For a simple long position quoted in USDT, define:

• R: the total planned loss budget in USDT.
• A: the frozen ATR reading in USDT per unit of the asset.
• k: the chosen ATR multiplier.
• D = k × A: the planned entry-to-stop distance.
• C: an estimated combined entry-fee, exit-fee and slippage allowance in USDT per asset unit.

The initial quantity calculation is q = R / (D + C). With no allowance, it simplifies to R / (k × A). This is basic unit arithmetic: USDT divided by USDT per ETH gives ETH.

The model applies to an unleveraged spot holding or a linear contract when the quantity and contract multiplier have been translated into asset units. Bybit's documentation, for example, calculates USDT-contract P&L from quantity multiplied by the entry-to-exit price difference; its closed P&L also accounts for trading and funding fees. Bybit's USDT-contract P&L guide. Inverse contracts need their own payoff calculation.

C is a planning allowance, not an exchange fee quote or an upper bound on execution cost. Avoid counting the same expected slippage in both the entry price and the allowance. Funding, borrowing costs and any fixed charges require additional treatment if relevant. For perpetuals, the basis-versus-funding guide explains why a price premium is a different observation from a funding payment. The slippage and depth guide explains why a fixed allowance can fail in a moving book.

A three-state sizing ledger

Assume a hypothetical 10,000-USDT account, a chosen loss budget of 50 USDT, an entry at 2,000 USDT per ETH, a multiplier of two and a combined cost allowance of 4 USDT per ETH. Assume a fictional quantity step of 0.001 ETH. These are constructed inputs, not current ETH prices or venue specifications.

For the middle row, 50 / (60 + 4) = 0.78125 ETH. Rounding down gives 0.781 ETH, and 0.781 × 64 = 49.984 USDT. Its entry notional is 0.781 × 2,000 = 1,562 USDT.

All three rows target approximately the same modeled loss, but their notionals differ substantially. If a separate exposure cap permits only 2,000 USDT of this asset, the lower-range row is restricted to 1 ETH. Its modeled loss becomes 1 × 34 = 34 USDT. There is no requirement to spend the unused risk allowance.

Apply the strictest valid quantity allowed by the loss budget, available balance, exposure cap and venue rules. Round the actual stop to a valid price tick before recomputing its distance and quantity. Then check minimum quantity and minimum notional. If the valid minimum exceeds the budget, the outcome is no valid trade under these rules; rounding upward changes the risk. See the exchange order-minimum guide.

Freeze the entry decision; define later changes separately

Suppose the middle-row position of 0.781 ETH is already open. ATR later rises from 30 to 50. Moving its stop from 1,940 to 1,900 while retaining the quantity increases the modeled entry-to-stop loss to 0.781 × (100 + 4) = 81.224 USDT.

The higher-range row's 49.920-USDT loss belongs to a different starting quantity: 0.480 ETH. It does not describe the 0.781 ETH position. Updating the indicator is not the same as resizing an existing trade.

Write the management rule in advance. A fixed entry stop, an ATR-based trailing rule, and a rule that periodically adjusts exposure are different strategies. Each needs separate testing, cost accounting and confirmed order-state handling. A trailing rule also needs an anchor and a direction constraint; a fresh ATR value alone does not define it.

If a tested policy calls for reducing exposure, model the reduction's execution and reconcile the remaining position and protective instruction. A local calculation does not establish that either order has filled. Never treat an intended amendment as confirmed protection.

Where ATR sizing can fail

An abrupt repricing can outrun a smoothed historical indicator. A quiet period produces a small denominator and therefore a larger calculated quantity immediately before a sudden move. The exposure cap in the ledger addresses one consequence; it does not make the next range predictable. A volatility regime filter instead changes whether new entries are eligible; that permission decision should be specified separately from the quantity calculation.

A stop price is also not an assured fill price. Bybit distinguishes conditional triggers from the market or limit instructions activated afterward, with execution behavior depending on the resulting order. Bybit's order-type overview. A fast market can exceed the modeled allowance, while a stop-limit can leave exposure open. The stop-market versus stop-limit walkthrough demonstrates that distinction.

For leveraged positions, independently examine margin and liquidation conditions. The same quantity can have the same price-movement loss while different collateral arrangements change liquidation risk. ATR sizing does not establish that a planned stop will execute before liquidation.

Reject missing, zero, stale or malformed ATR inputs rather than allowing an invalid denominator to produce a huge size. Very short data histories and inconsistent candle feeds also deserve a rejection rule or an explicitly documented fallback.

Keep a reusable sizing record

Replay the three ledger states with identical rules, then add a sudden jump, an invalid ATR input and a quantity below the venue minimum. Verify that future candles do not change the stored entry calculation; the look-ahead-bias audit provides a useful reproducibility check.

These checks establish calculation integrity, not profitability. They also do not establish that KLINEO implements this particular sizing rule or every cited venue capability. The method is useful when its observation, arithmetic and subsequent position management remain separately inspectable under the editorial policy.