ATR — True Range, smoothing and a worked calculation

Imagine a hypothetical trader copying a thirty-pip stop from EUR/USD to GBP/JPY without checking current ranges. The same pip distance may represent different fractions of each pair’s ATR. This article explains the range calculation and how to turn an explicitly chosen stop distance into a position-size model; no fixed volatility ratio or actual trading history is assumed.

What the Average True Range really is

Average True Range averages a range measure that includes the previous close. TradingView attributes the indicator to J. Welles Wilder’s 1978 New Concepts in Technical Trading Systems, alongside RSI, ADX and Parabolic SAR. ATR describes historical volatility without direction; it is not the full distance travelled by ticks or a bound on the next bar. Record the price units, timeframe, period, initialization and smoothing method before interpreting the number.

ATR contains no sign specifying buy or sell. The same range can accompany rising or falling prices. The ADX trend-strength calculation answers a different question. A period of 14 is a common configurable setting, not a universal optimum: on H1 it refers to hourly bars, on D1 to daily bars. A raw EUR/USD ATR of 0.0030 equals 30 pips when one pip is 0.0001, or 300 platform points if a point is 0.00001. Do not assume that every symbol uses that conversion.

How true range and the indicator are calculated

For one bar, True Range is TR = max(H − L, |H − previous C|, |L − previous C|). H and L are the current high and low; previous C is the preceding close. Absolute-value signs make the distances nonnegative. Choose the maximum, not the sum. If previous C lies within H–L, TR equals that range; if it lies outside, a distance to an extreme may make TR larger.

A fully specified example. Use four consecutive closed bars from one hypothetical EUR/USD Bid feed, with previous C = 1.1000 before bar 1 and pip = 0.0001. These are invented teaching prices, not market observations.

Bar 1 compares 0.0030, 0.0020 and 0.0010, so TR is 30 pips. For bar 2, H − L is 20 pips, while previous C = 1.1010 gives distances of 50 and 30 pips: TR is 50. Its opening gap is 40 pips, which is not mechanically added to its 20-pip range. Bar 3 has TR 40 pips; bar 4 has TR 10 pips. Open is useful for checking the data but does not appear directly in the TR formula.

For a transparent Wilder-style ATR(3) model, initialize with the first three TR values: (0.0030 + 0.0050 + 0.0040) / 3 = 0.0040, or 40 pips. The next update is ((n − 1) × previous ATR + current TR) / n, giving (2 × 0.0040 + 0.0010) / 3 = 0.0030, or 30 pips. A rolling SMA(3) instead gives (0.0050 + 0.0040 + 0.0010) / 3 ≈ 0.003333, or 33.33 pips. Wilder smoothing retains earlier information through the previous average; it is not simply the latest three TR values. This example computes ATR(3), not ATR(14).

Period, smoothing and bar aggregation

TradingView uses RMA by default for its ATR and allows SMA, EMA or WMA instead. A shorter period changes the weight of a new observation; it does not stop being an average or fix a delay of exactly a week. The period is a parameter to test under specified rules. Aggregation also matters: with previous C 100, bars O/H/L/C 100/102/99/101 and 101/101/98/100 each have TR 3. Their combined bar is 100/102/98/100 with TR 4, whereas the sum of the smaller TR values is 6 and their mean is 3. Rebuild the larger OHLC bars and recalculate TR; multiplying an H1 ATR by 24 does not create a D1 ATR.

ATR as a volatility measure — and its limits

ATR values use the instrument's price units, so cross-market comparisons require context and normalization. Compare ATR with the same instrument's history, or express it as a percentage of price. A fixed point reading on an equity index does not by itself identify a crisis.

Wilder smoothing weights each new true range by 1/N. A large range can lift ATR, while later smaller ranges reduce it gradually; ATR alone does not establish wider spreads or routine slippage. If volatility changes, recompute the planned stop distance and position size using current instrument specifications. A rule to halve size at a chosen threshold is a strategy assumption to test.

An explicitly chosen ATR stop-distance model

One possible rule uses an ATR multiplier to choose a planned stop distance. As hypothetical inputs, an EUR/USD ATR of 80 pips gives 120 pips at 1.5× or 160 at 2×; a GBP/JPY ATR of 200 pips gives 300 or 400 pips. These are assumed readings, not current normal ranges. For a long EUR/USD entry at 1.0850 with pip = 0.0001, the two planned levels are 1.0730 and 1.0690. The distance is a multiple of the selected ATR, not necessarily a fraction of the actual daily range. No multiplier guarantees that ordinary fluctuations will not trigger a stop, and an ordinary stop does not cap the realized loss. Evaluate the rule and its costs under stated data and fill assumptions.

Position sizing computed from ATR

For a hypothetical USD 10,000 account, a 1% planned price-loss budget is USD 100. Assume EUR/USD contract size EUR 100,000 per lot, pip 0.0001, USD 10 per pip per lot and a 120-pip stop distance. Unrounded size is 100 / (120 × 10) = 0.08333 lots; rounding down to a 0.01-lot step gives 0.08 lot and USD 96 planned price loss. Add commission and other costs and allow for slippage. Other pairs require their own pip value and account-currency conversion; a stop does not guarantee a maximum loss.

Trailing stops based on ATR

A trailing stop needs an explicit reference price and updating rule. StockCharts documents its Chandelier Exit default as highest high over 22 periods minus 3 × ATR(22) for longs, or lowest low plus 3 × ATR(22) for shorts, and attributes the indicator to Charles Le Beau. That is a named platform convention, not a compulsory day-trading or swing-trading multiplier. A plotted candidate level and an actual broker stop are different objects.

An ATR distance can expand as volatility rises and contract as it falls. Specify whether the protective level is allowed to move away from price: a trail that only ratchets toward profit will not loosen simply because ATR increases. Neither an ATR trail nor a fixed-distance trail guarantees a profitable exit.

Practical takeaway: ATR supplies a price-distance measure. Position sizing still requires the stop rule and pip value in the account currency.

What to do tomorrow

You can reproduce the calculation without a funded account. Use a frozen set of closed-bar prices, then compare your result with the platform’s inputs and initialization. Four checks help distinguish a numerical discrepancy from a trading decision.

1. Record the instrument, feed, timeframe, period, smoothing and price units. Compare like-for-like values rather than assume a permanent volatility ratio between pairs.
2. Compute TR from H, L and previous C on a closed bar. With missing previous C, state an initialization rule; do not silently insert zero or mix Ask highs with Bid lows.
3. If you study a sizing rule, specify the chosen distance, monetary budget and pip value in the account currency. Keep costs and slippage separate from the planned price-loss calculation.
4. Recheck the same calculation after another closed bar. If you test a threshold or a rule to halve size, treat it as a strategy parameter with defined evidence, not a universal response to an ATR increase.