Volatility is different from direction

Average True Range, or ATR, summarises the size of recent price movement. A rising ATR can accompany an upward breakout, a downward sell-off or a violent two-way market. It supplies a scale for movement, not a buy or sell forecast.

That scale can make trade planning more consistent. A fixed ten-pip stop means something different in an hourly market that has recently moved eight pips per candle than in one moving forty. ATR allows a study to express distances in volatility units while keeping the monetary risk budget separate.

ATR is quoted in price units. On EUR/USD, 0.0020 is commonly 20 pips; on a JPY pair, a 0.20 move is commonly 20 pips. Check the instrument’s pip convention and tick size rather than treating every last displayed decimal as a pip. MetaQuotes’ ATR definition.

True Range and the averaging choice

For a completed candle with high H, low L and previous close Cprev, TR = max(H − L, |H − Cprev|, |L − Cprev|). It captures a gap from the previous close even when the candle’s own high–low range is small.

Suppose EUR/USD’s previous close is 1.0805, the next high is 1.0860 and its low is 1.0825. The three distances are 35, 55 and 20 pips, so True Range is 55 pips.

Wilder-style ATR seeds with the mean of N true ranges, then updates as ATR(t) = [(N − 1)ATR(t−1) + TR(t)]/N. A rolling simple ATR instead averages only the latest N true ranges. Both are legitimate volatility summaries, but they are different calculations.

MetaQuotes’ published ATR example uses the rolling simple form, updating by adding the newest TR and removing the oldest. Do not assume every platform’s ‘ATR14’ uses Wilder smoothing. The study below uses SMA of 14 true ranges. Inspect the official ATR source.

Convert volatility into a trading plan

An initial stop at two ATR units is a distance rule, not a probability statement. It does not mean the stop is safe from ordinary movement, nor that only a fixed percentage of trades should reach it. A structure-based stop may be preferable in some studies, with ATR used only as a buffer.

Position size should respond inversely to the chosen stop distance when the monetary risk budget stays fixed. Doubling the stop distance normally halves the pre-cost exposure. Conversely, a quiet period can imply a very large position unless a separate notional, margin or maximum-lot constraint is applied.

Normalised ATR, defined as ATR divided by price and multiplied by 100, can help compare relative movement across instruments. For execution decisions, also examine spread divided by ATR: a spread that consumes a large fraction of ordinary movement can make a short-horizon strategy costly.

A reproducible breakout and ATR-trailing study

This hypothetical EUR/USD H1 framework illustrates using ATR for risk while price supplies direction.

1. At each completed bar t, calculate SMA-ATR14 and the highest high of the preceding 20 completed bars, excluding t. Require close(t) above that high and above EMA50(t).
2. Enter at the next available ask. Skip if spread exceeds the smaller of 2 pips and 10% of ATR, or if the fill is more than 0.2 ATR above the signal close.
3. Place the initial stop two signal-bar ATR units below the actual fill. Size for a maximum planned loss of 0.5% of equity including estimated costs, subject to a separately chosen maximum exposure and margin constraint.
4. After every later candle closes, calculate a trailing candidate as the highest completed close since entry minus three current ATR units. Set the new stop to the greater of the previous stop and this candidate. Never move it downward for a long position.
5. If the new candidate is already above the executable bid, exit at market instead of submitting an invalid stop. Otherwise respect broker stop and freeze distances when updating the order.
6. Use no fixed profit target in this study. Exit on the active stop or after 48 completed candles, whichever occurs first. Hold one position and require a new 20-bar breakout before re-entry.

Mirror the framework for shorts with lowest prices and a stop that can only move downward.

Worked stop, sizing and trailing calculations

Assume signal ATR14 is 0.0020 and a valid breakout fills at 1.0902. The initial stop is 1.0862, a distance of 40 pips. With USD 10,000 equity, a 0.5% budget gives USD 50. At USD 10 per pip per standard EUR/USD lot, pre-cost size is 0.125 lots. Reduce for commission and an execution allowance, then round down to the volume step.

Later, the highest completed close since entry reaches 1.0960 while current ATR is 0.0025. The trailing candidate is 1.0960 − 3 × 0.0025 = 1.0885. Because this is above the old stop, the rule raises the stop to 1.0885, if permitted by the current quote and broker constraints.

If volatility subsequently jumps to ATR 0.0060 without a new highest close, the raw candidate falls to 1.0780. The rule retains 1.0885. A volatility increase is therefore not permission to expand an existing trade’s loss allowance.

Where ATR-based risk can disappoint

ATR is backward-looking. After an unusually quiet period, a news surprise can cause movement many times larger than the prior average. After a large shock, ATR may stay elevated even as the market calms. Simple averaging can also change abruptly when an old large candle leaves the window.

An ATR stop does not guarantee its requested fill. Weekend gaps, thin liquidity and rapid announcements can create slippage beyond the planned distance. A calculated position size therefore represents an estimate of risk under specified execution assumptions, not a maximum possible loss.

Validate the exact ATR implementation and compare several reasonable stop multiples without selecting only the historical winner. Use chronological holdout periods with quiet and stressed conditions. Report drawdown, net expectancy, tail losses and exposure, and model spread, commission, financing and slippage. Trailing stops require realistic intrabar prices; evaluating only candle closes can miss stop hits that occurred before the close.

MT4/MT5 workflow and calculation checks

Add Average True Range from the desktop indicator list and set period 14. Read the completed candle’s value in the Data Window. Convert price units to pips explicitly before entering a stop distance or using a position-size calculator.

Confirm the platform or custom indicator’s averaging method with a small manual example. Keep that method unchanged between the chart, risk script and backtest. If a strategy requires Wilder smoothing, use a verified implementation rather than assuming the default name guarantees it.

For automated trailing, read the newly completed bar, calculate the candidate once and record every requested and accepted stop modification. A terminal-side trailing mechanism may depend on the platform staying connected; verify the actual order-management behaviour used by your implementation. Preserve the initial protective stop even when later updates cannot be submitted.