Average True Range (ATR) Explained

Table of Contents

    Average True Range (ATR) helps answer a practical question: how large have an asset’s recent price ranges been, including movement across the previous close? A $2 stop distance means something different for a stock whose recent ranges are around $0.50 than for one whose ranges are around $5. ATR supplies a common price-based reference for that comparison.

    It does not tell you whether to buy or sell. This guide calculates a 14-period ATR from complete hypothetical data, explains how to compare it across prices, and separates a planned stop-distance budget from the loss that an actual trade can produce.

    What Is Average True Range?

    ATR smooths a sequence of true ranges. Each true range considers the current bar’s high and low together with the previous bar’s closing price. Including that previous close lets the calculation reflect a price gap that a simple high-minus-low range can miss. A bar can represent a trading day, an hour or another specified interval. [1]

    Range Size, Without Direction

    For a stock quoted in dollars per share, an ATR of 3.00 means a smoothed true range of $3 per share for the chosen bars. It does not mean a 3% return, a $3 expected gain, or a maximum $3 move.

    Large upward and downward movements can both raise ATR. Falling ATR describes smaller recent true ranges, not a prediction that the next move will be upward. Fidelity’s indicator documentation explicitly distinguishes volatility from price direction. [1]

    The True Range Formula

    Let Ht and Lt be the current bar’s high and low. Let Ct-1 be the previous bar’s close. Calculate three distances and take the largest:

    True Range
    TRt = max[Ht − Lt, abs(Ht − Ct-1), abs(Lt − Ct-1)]
    abs means absolute value: a distance is nonnegative, whether the price moved up or down.

    This is the formula documented by Fidelity and ICE Market-Q. Do not add the three distances: their spans overlap. With valid high and low data, true range is equivalently the distance from the lowest to the highest of the current high, current low and previous close. [1] [2]

    If the previous close lies within today’s high-low interval, the ordinary high-low range wins. If the whole interval is above or below the previous close, one of the previous-close distances wins. This measures a span, not the sum of every price movement during the bar.

    A Complete 14-Period Wilder ATR Example

    Consider a fictional stock with completed daily bars, all on the same price and trading-session basis. The close immediately before day 1 is $100. Prices below are USD per share; there are no splits, dividends or missing bars. Days 1 through 14 initialize ATR, and days 15 and 16 show the updates.

    Hypothetical daily prices and true ranges, in USD per share
    Day Previous close High Low Close True range
    1 100.00 101.00 99.00 100.00 2.00
    2 100.00 102.00 99.00 101.00 3.00
    3 101.00 105.00 103.00 104.00 4.00
    4 104.00 105.00 103.00 104.00 2.00
    5 104.00 106.00 103.00 105.00 3.00
    6 105.00 103.00 101.00 102.00 4.00
    7 102.00 103.00 101.00 102.00 2.00
    8 102.00 104.00 101.00 103.00 3.00
    9 103.00 107.00 105.00 106.00 4.00
    10 106.00 107.00 105.00 106.00 2.00
    11 106.00 108.00 105.00 107.00 3.00
    12 107.00 106.00 103.00 104.00 4.00
    13 104.00 106.00 103.00 105.00 3.00
    14 105.00 107.00 104.00 106.00 3.00
    15 106.00 110.40 108.00 109.00 4.40
    16 109.00 109.70 108.00 109.00 1.70

    1. Check the gap-sensitive true ranges

    On day 3, the three distances are 105 − 103 = 2, abs(105 − 101) = 4 and abs(103 − 101) = 2. True range is $4, although the day’s high-low range is only $2.

    Day 6 shows the reverse situation: 103 − 101 = 2, abs(103 − 105) = 2 and abs(101 − 105) = 4. The true range is again $4. Equal range sizes do not reveal the direction of the gap.

    2. Average the first 14 true ranges

    For this finite starting dataset, use the arithmetic mean of the first 14 true ranges. ICE Market-Q explicitly documents this initial-value convention. [2]

    Initial 14-Period ATR
    ATR14 = (TR1 + TR2 + … + TR14) / 14
    Initialize once; do not apply the subsequent recurrence before an initial ATR exists.

    The sum is 2 + 3 + 4 + 2 + 3 + 4 + 2 + 3 + 4 + 2 + 3 + 4 + 3 + 3 = 42. Therefore, ATR14 = 42 / 14 = $3.00.

    3. Apply Wilder smoothing to subsequent bars

    After initialization, retain 13 parts of the previous ATR and add one part of the new true range, then divide by 14. This is Wilder smoothing, not a fresh equal-weight average of the latest 14 rows. [2]

    Subsequent 14-Period ATR
    ATRt = (13 × ATRt-1 + TRt) / 14
    Equivalently, ATR changes by one-fourteenth of the difference between the new true range and the previous ATR.
    • Day 15: TR = max(2.40, 4.40, 2.00) = 4.40. ATR = (13 × 3.00 + 4.40) / 14 = 43.40 / 14 = $3.10.
    • Day 16: TR = max(1.70, 0.70, 1.00) = 1.70. ATR = (13 × 3.10 + 1.70) / 14 = 42.00 / 14 = $3.00.

    For comparison, a rolling 14-bar simple average on day 15 would remove day 1’s range and add day 15’s: (42 − 2 + 4.40) / 14 = $3.1714, rounded. That is a different smoothing rule. Carry full precision through calculations and round only the displayed results.

    How to Interpret ATR Values

    Within a consistent series, rising ATR means the smoothed true range is increasing. In the example, day 15’s $4.40 true range exceeds the prior $3 ATR, so the smoothed value rises by only (4.40 − 3.00) / 14 = $0.10. A single large bar influences later values through the recurrence.

    The period setting controls responsiveness. For a general n-period Wilder ATR, the new observation’s coefficient is 1/n. A shorter setting gives the newest range more weight; a longer setting smooths it more. The usual EMA convention instead uses 2/(n + 1). [8] TradingView, for example, permits multiple smoothing choices, so matching the number 14 alone does not establish matching calculations. [4]

    There is no universal dollar value that makes ATR high or low. Compare it with the same asset’s earlier values under the same conventions, and consider the price level. A quiet historical reading cannot rule out an upcoming earnings gap or other sudden change.

    ATR Percent: Comparing Different Price Levels

    Raw dollar ATR can mislead when security prices differ. StockCharts documents this absolute-price limitation. To remove the immediate price-level scale, define ATR percent here as ATR divided by the same bar’s positive closing price, multiplied by 100. [3]

    Normalized ATR Used in This Article
    ATR%t = 100 × ATRt / Ct
    Use a consistent price basis and state the denominator. A zero or negative price requires a different treatment.

    Reset to two fictional stocks, Cedar and Harbor. Suppose their completed daily 14-period ATRs use the same method and session convention. Cedar closes at $50 with a $2 ATR; Harbor closes at $200 with a $4 ATR.

    Hypothetical ATR comparison after price normalization
    Stock Close (USD/share) ATR (USD/share) ATR percent
    Cedar 50.00 2.00 100 × 2 / 50 = 4%
    Harbor 200.00 4.00 100 × 4 / 200 = 2%

    Harbor has twice the dollar ATR but half the ATR relative to its current price. The ratio supports this specific range comparison; it does not establish which investment is safer overall. Use the same bar frequency, period count, smoothing and data conventions before comparing normalized values.

    ATR vs Standard Deviation and Other Volatility Measures

    The word volatility covers several measures. Their units and inputs tell you which question they answer:

    Price-range measures and return-volatility measures differ
    Measure Main input Interpretation
    High-low range One bar’s high and low Price span within the bar; can miss the gap from the prior close.
    True range High, low and previous close Gap-aware span in price units for one bar.
    ATR / ATR percent Smoothed true ranges / current close Range scale in price units / normalized percentage under the stated denominator.
    Return standard deviation A sequence of returns Dispersion of returns around their mean, in return units for the stated frequency.

    Standard deviation in finance measures a different quantity. A 2% ATR ratio does not mean a 2% daily return standard deviation. It also does not create a statistical confidence interval: two ATRs is not automatically a two-standard-deviation move. Multiplying ATR or ATR percent by the square root of 252 does not automatically produce annualized return volatility; that would require an additional model connecting ranges to returns.

    Other historical volatility estimators use high-low or OHLC information to estimate return variance under stated assumptions. That task differs from smoothing true range. GARCH and EWMA volatility models instead model time-varying return variance. ATR alone supplies neither their variance forecast nor a conversion to it.

    Using ATR to Define a Hypothetical Stop Distance

    An ATR-based rule can express an intended stop distance as a chosen positive multiple, k, of a measured ATR. For an initial long position:

    Initial Long-Position Stop Reference
    Distance = k × ATR
    Stop trigger = Entry price − Distance
    A planning rule for the trigger level; it does not specify or guarantee the execution price.

    Start a separate hypothetical example: assume entry at $80, ATR measured from completed bars before entry of $2.40, and a chosen multiple of 2. The distance is 2 × $2.40 = $4.80, giving an initial stop trigger of $80 − $4.80 = $75.20. Freeze that ATR input and trigger for this example.

    The choice of 2 is illustrative. A larger multiple puts the initial trigger farther away and allows a larger price loss per share if execution occurs there. It does not make a strategy profitable, identify a fundamental breakdown or prevent a temporary move from triggering the order.

    Actual trading execution is a separate issue. A stop-market order seeks execution after its trigger is met; the price can be materially worse than the trigger. A stop-limit order constrains the execution price but may remain unfilled. FINRA describes both risks. [5]

    This fixed initial reference is also different from a moving stop. An ATR-based trailing strategy needs explicit rules for its reference price, recalculation timing and whether the stop may loosen. Do not assume a broker’s ordinary trailing stop order automatically calculates an ATR trail.

    ATR Position Sizing: Work From a Planning Budget

    Continue the $80-entry example. Assume account equity of $50,000 and an independently chosen planning budget of 0.5%, or $50,000 × 0.005 = $250. This percentage is an illustration, not a recommended allocation or maximum possible loss.

    Ignoring all trading costs first, divide that budget by the $4.80 planned loss per share and round down to whole shares:

    Whole Shares Before Costs
    Shares = floor(Planning budget / Stop distance)
    floor means round down; actual capital, account and liquidity limits can require a smaller position.
    Separate Planned Loss From Position Value

    Shares: floor($250 / $4.80) = 52 shares.

    Price loss if filled at the assumed stop: 52 × $4.80 = $249.60. Buying 53 would give 53 × $4.80 = $254.40, above the planning budget.

    Position value at entry: 52 × $80 = $4,160. This is the capital exposed to the stock, not a $250 investment.

    Allow for commissions and other costs

    Suppose fixed entry and exit commissions/fees total $2, with no other costs in this simplified calculation. The adjusted share count is floor[($250 − $2) / $4.80] = 51. At the assumed stop fill, the planned total is 51 × $4.80 + $2 = $246.80. Keeping 52 shares would instead give $249.60 + $2 = $251.60.

    Actual spreads, slippage and fees may depend on share count, order type and market conditions. Model those terms explicitly rather than assuming the $2 illustration covers them. A cost allowance still cannot guarantee the fill.

    Stress the gap before trusting the budget

    Keep the cost-adjusted 51 shares and the $75.20 trigger. Suppose the next eligible market price is $70, the broker’s trigger condition activates the order, and the entire position fills at $70. The loss is 51 × ($80 − $70) + $2 = $512. The position exceeded the $250 budget even though its initial sizing arithmetic was correct.

    A Stop Is Not a Maximum-Loss Contract

    Gaps, trading halts, thin liquidity and partial fills can separate the trigger from the actual sale. ATR records historical ranges; it does not cap the next move or guarantee an exit. Check the broker’s trigger, session and order-handling rules. [5]

    A larger ATR can imply fewer shares

    Reset costs to zero for a clean comparison. Keep the $80 entry, $250 budget and multiple of 2, but compare ATRs of $2.40 and $4.80.

    Hypothetical sizing with a fixed budget, before costs
    ATR (USD/share) Stop distance (USD/share) Whole shares Planned price loss (USD)
    2.40 2 × 2.40 = 4.80 floor(250 / 4.80) = 52 52 × 4.80 = 249.60
    4.80 2 × 4.80 = 9.60 floor(250 / 9.60) = 26 26 × 9.60 = 249.60

    Doubling ATR halves the share count in this example. Widening the distance to $9.60 while retaining 52 shares would instead create a planned price loss of 52 × $9.60 = $499.20. A change to the stop policy and a change to position size must be considered together.

    The sizing formula requires a positive stop distance. A zero ATR does not justify dividing by zero or taking an unlimited position. Whole-share arithmetic also needs a cash or buying-power cap. If the result is zero shares, the proposed setup does not fit that budget. For futures, options or other contracts, dollar exposure per quoted price point differs from one stock share; contract multipliers and other risks require separate treatment.

    Data Conventions That Can Change ATR

    Two tools can display different ATRs without either having a simple formula error. Before comparing them, match the following inputs:

    • Bar interval and session. Fourteen daily bars differ from fourteen hourly bars. Define the trading session, time zone and bar boundaries; extended-hours inclusion can change intraday highs, lows and previous closes. TradingView documents separate extended-hours display settings and availability. [7]
    • Corporate-action adjustments. Use highs, lows and closes on one consistent basis. Alpaca’s documentation, for example, distinguishes raw, split-adjusted and dividend-adjusted bars. These are provider choices, not interchangeable inputs. [6]
    • Smoothing and starting history. Match Wilder versus simple/other smoothing and the seed. A platform that starts much earlier can retain a different accumulated state from a short manual example. StockCharts specifically notes history and rounding differences. [3]
    • Completed versus live bars. A current bar’s high, low and close can change until it ends. State the observation time and use information actually available at a historical decision point.

    For a concrete adjustment check, suppose an unadjusted prior close is $100 before a hypothetical 2-for-1 split and the next high and low are $51 and $49. The raw formula gives max(2, 49, 51) = $51. Restating the prior close to the same post-split basis, $50, gives max(2, 1, 1) = $2. An inconsistent price basis can create a misleading range spike. Dividend treatment should likewise be declared rather than mixing an adjusted close with unrelated raw highs and lows.

    How to Evaluate an ATR-Based Rule

    1. Document the indicator. Record bar interval, session, price adjustments, period count, seed and smoothing. Reproduce a few true ranges before relying on a chart label.
    2. Define the decision before testing. Specify entry rules, the ATR observation used, multiplier, stop updates, order type and holding period. Use completed data available at the time.
    3. Check execution and costs. Include plausible gap, spread, slippage and nonexecution scenarios. A backtest that assumes a fill at every stop price can understate losses.
    4. Assess the whole portfolio. ATR does not measure correlation and covariance. Positions exposed to the same industry or event may lose together.
    5. Challenge the settings. Compare nearby periods and multiples across different market conditions, including data outside the sample used to choose the rule. An attractive fitted multiplier is not proof of a durable advantage.

    For example, three simultaneous $250 planning budgets sum to $750, or $750 / $50,000 = 1.5% of the example account before costs. That total is already larger than one position’s 0.5%, and common gaps can push realized losses beyond it. Neither the sum of budgets nor the sum of ATRs is a portfolio-volatility estimate.

    Common Mistakes With ATR

    • Using high minus low every time. Compare both previous-close distances too; the gap-sensitive days in the example show why.
    • Restarting a simple average each day and calling it Wilder ATR. Initialize with the mean, then apply the recurrence consistently.
    • Comparing raw ATR across price levels or bar intervals. Normalize where useful and match the method and timeframe first.
    • Treating a multiple as a probability or direction signal. ATR supplies neither a normal-distribution confidence level nor a forecast return.
    • Increasing a stop distance without reconsidering size. More distance means more planned dollar loss at the same share count.
    • Calling the stop budget the worst possible loss. Execution, gaps and correlated positions can invalidate that interpretation.

    Limitations of Average True Range

    ATR describes past price spans and responds to new information with smoothing. A low reading before an announcement can coexist with substantial event risk; a high reading after a shock can remain elevated as quieter bars arrive. A different market regime can make the recent scale unrepresentative.

    It also leaves out several dimensions of risk: the probability distribution of future returns, valuation, leverage, trading depth, the size of an executable order and relationships among holdings. A brief bad print or inconsistent corporate-action adjustment can distort its inputs. Use ATR to make one aspect of a trading rule explicit, then assess those other assumptions separately.

    Frequently Asked Questions

    No. Rising ATR means recent true ranges are increasing under the chosen smoothing rule. Large upward or downward price moves can both produce that result. Direction requires separate information.

    The initial value in this example averages 14 true ranges. Subsequent Wilder values also retain older history indirectly through the previous ATR. They are not rolling simple averages that discard the oldest observation completely at each step.

    Obtain an earlier close when possible. Some implementations start their first true range at high minus low when no earlier close is available; StockCharts documents that convention. Declare the fallback because it changes the starting dataset. This article supplies the prior close for every row. [3]

    No. Here it is 100 times ATR divided by the same bar’s closing price. That normalizes a range measure by price; it does not turn the result into a standard deviation of returns or an annual measure.

    There is no universal profitable multiple. Results depend on entry and exit rules, timeframe, instrument, costs, execution and market conditions. The multiple of 2 in the worked example explains distance and sizing arithmetic; it is not a recommendation or a guaranteed loss limit.

    Sources

    1. Fidelity: Average True Range (ATR). Definition, gaps, nondirectionality and true-range calculation.
    2. ICE Data Services / Market-Q: Average True Range. Initial average, subsequent smoothing and bar-period convention.
    3. StockCharts ChartSchool: Average True Range (ATR). Absolute units, initialization and history/rounding caveats.
    4. TradingView: Average True Range (ATR). Indicator settings and alternative smoothing methods.
    5. FINRA: Stop Orders: Factors to Consider During Volatile Markets, March 26, 2025. Stop-price, execution and stop-limit risks.
    6. Alpaca: Historical bars. Provider-specific adjustment, interval and feed options.
    7. TradingView: I want to access Extended Hours data. Intraday session settings and availability.
    8. StockCharts ChartSchool: Moving Averages – Simple and Exponential. Conventional EMA weighting multiplier.

    Sources accessed November 14, 2026. All numerical price, account, cost and loss examples are hypothetical teaching calculations, not actual security data or backtested results.

    Disclaimer

    This article is for educational and informational purposes only and does not constitute investment advice. ATR settings, stop levels and share counts are illustrations, not trading recommendations. Actual losses can exceed a planned stop budget. Consider your circumstances and consult a qualified financial professional before making investment decisions.