Average True Range (ATR): Formula, Settings, and Trailing Stops

Average True Range (ATR) measures how much a market has been moving. A rising ATR can accompany a rise or a fall in price. It supplies a volatility scale, so a trading rule still needs a separate way to choose direction.

The familiar 14-period setting appears in many chart packages. Traders also use multiples of ATR to describe stop distances. Across 104,160 historical configuration evaluations, we examined a detail that can disappear behind that shorthand: a line described as an ATR trail can move farther away from price when volatility expands.

How ATR is calculated

Start with the true range of each candle:

True range = largest of:
    high − low
    absolute value of (high − previous close)
    absolute value of (low − previous close)

Including the previous close lets true range account for a gap. If a candle has a high of 101, a low of 99, and a previous close of 95, its high–low range is 2 but its true range is 6.

The customary ATR calculation smooths these true ranges using Wilder’s update. For ATR 14, after initialization:

ATR now = (13 × previous ATR + current true range) ÷ 14

TradingView’s ATR reference describes true range and its default Wilder-style smoothing. The number 14 counts chart candles, so the same setting spans different elapsed time on hourly and daily charts.

ATR is a distance; ATR percentage puts it in context

An ATR of 2 means two price units. It does not mean 2%, and it does not say that the next candle will move exactly two units.

To express the distance relative to price:

ATR percentage = 100 × ATR ÷ closing price

ATR 2 at price 100 → 2%
ATR 2 at price 1,000 → 0.2%

TradingView’s ATR percentage explanation uses this normalization. Raw ATR values across differently priced markets are difficult to compare; expressing each as a percentage puts their scale in context.

A longer ATR length changes how quickly the distance reacts. A larger multiplier changes how far away the rule places its line. Those are separate choices. The familiar 14-period length does not make a 2× or 3× distance universally appropriate.

Common uses of ATR

ATR is used to describe unusually large movement, scale a breakout threshold, choose a stop distance, or help relate position size to a specified risk budget. Fidelity’s ATR guide discusses volatility and distance-based rules.

Position size and stop distance remain different decisions. If the distance doubles while quantity stays unchanged, the amount exposed to that distance doubles too. Our study keeps the allocation rule fixed; it does not test an ATR position-sizing system.

A volatility reading also supplies no built-in directional signal. We used a separate moving-average entry so that the experiment could focus on the exit formulas.

An ATR trail needs a rule about moving backward

Consider this long-position exit line:

Anchor = higher of recorded entry price and highest close since entry
Adaptive line = anchor − multiple × current ATR

The anchor cannot decline while the position stays open. Including the entry price matters when an opening gap places the fill above the subsequent closes. The ATR distance can still grow. If ATR rises enough, subtracting that larger distance makes the line fall.

A ratchet adds a second rule:

Ratcheted line = higher of:
    previous ratcheted line
    current adaptive line

In the constructed example, the highest close stays at 104 while ATR rises from 2 to 4. At a 3× multiple, the adaptive line falls from 98 to 92. The ratchet remains at 98.

Both formulas use ATR. They implement different exit policies. One continually adapts its distance; the other preserves the tightest level reached so far. A name such as “ATR trailing stop” is insufficient to tell them apart.

We tested four explicit exit formulas

All four rules use the same entry and can also exit when price crosses below the entry average. They differ only in the additional ATR exit line:

The tests compare close-triggered market exits. A close below the line requests an order for the following candle. They do not simulate an exchange stop order that guarantees an intrabar fill at the drawn level.

Across the matched adaptive-versus-ratchet comparisons, 99.5% recorded the same number of executions. Their returns still differed: the ratchet produced a higher historical return in 32.4% of pairs, a lower return in 18.0%, and the same reported return in 49.6%. These shares describe this defined grid; they do not give probabilities for a future trade.

The chart changes only the ATR multiple, keeping the volatility length and entry rule fixed. The return surface has irregular steps because moving an exit past one candle can change its execution price. A wider multiple does not guarantee a higher return, and a ratchet does not guarantee a lower realized drawdown in every historical setting.

Why earlier exits need not mean fewer transactions

Our entry requires a fresh upward crossing of the trend average. After an ATR exit, price must complete another qualifying crossing before a new position can start. The ordinary downward crossing also provides an exit if an ATR line has not already done so.

At the fixed ATR 14/3×/EMA 50 setting, all four formulas also recorded the same execution count within each of the eight small-amount sample comparisons. Under that structure, a stop can change when the same cycle ends without creating or removing the next entry opportunity. This differs from a rule that re-enters immediately whenever price remains above the average. The entry specification is part of the stop experiment.

These observations concern a defined long-only rule on historical candles. ATR itself supplies a distance scale; the entry, anchor, update, ratchet, order type, and quantity determine what that scale does inside a strategy.

Try the tested rule

Defaults use ATR 14, a 3× distance, and an EMA 50 entry. stopMode selects the four formulas in the table. The script requests 99% equity sizing and a 500-candle startup; configure fees and market dates separately.

//@version=6
strategy("ATR stop laboratory", overlay=true, initial_capital=10000, default_qty_type=strategy.percent_of_equity, default_qty_value=99, process_orders_on_close=false)
atrLength = input.int(14, "ATR length", minval=2)
multiple = input.float(3, "ATR multiple", minval=0.1)
trendLength = input.int(50, "Entry EMA length", minval=2)
stopMode = input.int(2, "Stop: 0 fixed, 1 fixed-distance trail, 2 adaptive, 3 ratchet", minval=0, maxval=3)
a = ta.atr(atrLength)
trend = ta.ema(close, trendLength)
up = ta.crossover(close, trend)
down = ta.crossunder(close, trend)
var float entryATR = na
var float highClose = na
var float ratchet = na
float stopLine = na
if strategy.position_size > 0
    highClose := na(highClose) ? math.max(strategy.position_avg_price, close) : math.max(highClose, close)
    fixedStop = strategy.position_avg_price - multiple * entryATR
    fixedTrail = highClose - multiple * entryATR
    adaptive = highClose - multiple * a
    ratchet := na(ratchet) ? adaptive : math.max(ratchet, adaptive)
    stopLine := stopMode == 0 ? fixedStop : stopMode == 1 ? fixedTrail : stopMode == 2 ? adaptive : ratchet
    if close < stopLine or down
        strategy.close("Long")
else
    highClose := na
    ratchet := na
    if bar_index >= 500 and up
        entryATR := a
        strategy.entry("Long", strategy.long)
plot(trend, title="Entry EMA", color=color.blue)
plot(stopLine, title="Close-tested stop", color=color.orange)