Average True Range: How to Size a Stop With ATR in Futures
Average true range averages a bar's range to size a volatility-scaled stop. Set the stop as a multiple of ATR, then size the position to a fixed dollar risk.
By Imperial Analytics
Average true range is the indicator most traders meet through a stop-loss setting and never look at again. That is a missed opportunity, because ATR answers a question a fixed-point stop cannot: how much room does this trade need right now, in this session, on this instrument. This primer defines true range, shows how the average is built, and walks the two steps that turn an ATR reading into a stop distance and a contract count.
What average true range measures
Average true range is a volatility measure. It reports the average size of a bar's full price range over a lookback window, default fourteen periods, expressed in the instrument's own price units. ATR tells you how far price typically travels in one bar. It does not tell you which direction price will go.
ATR was introduced by J. Welles Wilder Jr. in 1978 as part of a volatility system, alongside the directional movement indicators and the parabolic stop-and-reverse.1 The reading is an absolute number in points, not a percentage and not a bounded oscillator. A larger ATR means recent bars have been wide and the instrument is moving quickly. A smaller ATR means recent bars have been tight and the instrument is quiet.
Because ATR is expressed in the instrument's own points, the reading is tied to that contract. An ATR of eight points on the ES and an ATR of eight points on a different contract describe different dollar ranges and are not directly comparable. The number is most useful as a relative read against the same instrument's own recent history: ATR today versus ATR last week tells you whether conditions have widened or settled.
That single property, range size without direction, is what makes ATR a sizing tool rather than a signal. It cannot tell a trader to buy or sell. It can tell a trader how wide a stop has to sit before normal bar-to-bar movement would hit it by chance.
How true range is calculated
True range is the largest of three distances: the current high minus the current low, the absolute value of the current high minus the prior close, and the absolute value of the current low minus the prior close. The gap-inclusive definition keeps an overnight jump from understating how far price actually moved.
A plain high-minus-low range misses the gap between one bar's close and the next bar's open. True range closes that hole by also measuring from the prior close, so a bar that opens away from the previous close still records the full distance traveled.1 The true range for a bar is whichever of the three candidate distances is largest.
Consider an illustrative bar. The prior close sits at 5000. The current bar prints a high of 5012 and a low of 5004. The high-minus-low distance is 12 points. The high-minus-prior-close distance is also 12 points. The low-minus-prior-close distance is 4 points. True range takes the largest, 12 points. In this case the intrabar range and the true range agree, but on a gap-down open the prior-close measurements would have been the larger figure and the plain range would have understated the move.
Data note
The 5000, 5012, and 5004 figures above are an illustrative bar shown to demonstrate the three-way maximum, not a live quote. True range is always computed from the actual high, low, and prior close of the bars on the trader's own chart.
True range is computed bar by bar across the lookback window, and those individual true ranges are the input the average is built from in the next step.
How ATR smooths true range into an average
ATR seeds with a simple average of the first fourteen true ranges, then each later bar updates with Wilder's smoothing: the prior ATR times thirteen, plus the new true range, divided by fourteen. The recursion gives the most recent bar weight without discarding the accumulated history in a single step.
Written as a formula, the update is ATR = (prior ATR * (n - 1) + current TR) / n, where n is the lookback period, default fourteen.1 The first value cannot use the recursion because there is no prior ATR, so it is seeded with a plain average of the first n true ranges. Every value after that carries the seed forward, gradually diluted by newer bars.
The smoothing is the reason ATR reacts but lags. A single unusually wide bar nudges the average upward, but does not move it to the new level on its own. A genuine shift in volatility, such as a quiet session giving way to a fast one, moves ATR over several bars as the wide true ranges accumulate in the recursion. This is by design: the average is meant to describe the prevailing range, not to jump on one outlier.
The lag has a practical consequence for warm-up. Because the seed influences the reading for many bars after it is set, ATR is not stable on the first few bars of a fresh series. A common rule is to let several lookback lengths print before trusting the value, which for a fourteen-period ATR is roughly forty bars.
How to size a stop from ATR
A volatility-scaled stop sets the stop distance as a multiple of ATR rather than a fixed number of points. A common form places the stop one-and-a-half to two times ATR away from the entry. The multiple widens the stop in fast conditions and tightens it when the instrument is quiet, so normal movement does not hit it by chance.
A fixed-point stop has the same problem in both regimes. Eight points might be a reasonable stop in a quiet session and a coin-flip in a fast one, where price covers eight points in a single bar without the trade being wrong. The ATR-scaled stop adapts because its width is a function of recent range. When ATR is large the stop sits further away; when ATR is small the stop pulls in.
The multiple is a choice the trader fixes in advance and then measures. One-and-a-half to two times ATR is a common starting band, but the right multiple is the one the trader's own results support, not a number borrowed from a screenshot. The stop should also respect chart structure. If a multiple-of-ATR distance lands just inside an obvious swing level where noise routinely trades, the trade is better skipped or the stop placed beyond the level, not parked where it is most likely to be clipped.
Data note
Any ATR reading used in a worked example below, such as an ATR of eight points, is illustrative. It is shown to demonstrate the arithmetic of stop and position sizing, not drawn from a live chart. The trader's own ATR comes from the actual true ranges on the instrument and timeframe being traded.
How to turn a stop distance into a position size
Once the stop distance is set in points, position size follows from the risk budget. Divide the dollars you are willing to lose on the trade by the per-contract dollar value of the stop distance. The result is the contract count that holds dollar risk constant whether the market is calm or wild.
Take an illustrative ATR of eight points on the ES and a chosen multiple of one-and-a-half, which sets the stop distance at 12 points. The ES is worth 50 dollars per index point, so 12 points is 600 dollars of risk per contract. The MES is worth 5 dollars per point, so the same 12-point stop is 60 dollars per contract. The point values are fixed contract specifications; the ATR reading is the illustrative part.
Now apply a risk budget. A trader willing to lose 300 dollars on the trade cannot size the ES position, because one contract already risks 600 dollars, double the budget. The same 300-dollar budget supports five MES contracts at 60 dollars of risk each. The ATR-scaled stop has already absorbed the volatility of the session, so the dollar risk on the trade stays at the budget regardless of how wide or narrow conditions are.
This is the same fixed-dollar risk that defines one R in the R-multiple framework. Sizing every trade to a constant dollar risk is what makes outcomes comparable across setups and across volatility regimes, because the only thing that changes between a calm trade and a fast one is the contract count, not the amount at stake.
What ATR does not tell you
ATR is a volatility measure, not a direction signal and not a forecast. It reacts to bars that have already printed, so it lags a regime change by several bars. A rising ATR says bars are getting wider, not that a trend is starting or that a reversal is near.
The first limitation is direction. A high ATR is equally consistent with a strong uptrend, a hard downtrend, and a violent two-sided range. Reading a rising ATR as a directional signal is a category error; the direction has to come from a separate part of the process.
The second limitation is that ATR reacts rather than anticipates. It is computed from bars that have already closed, so it cannot move ahead of price. The most it can do is summarize the range of what just happened. Any claim that ATR predicts the next move is really a claim about how the coming bars will compare to the recent ones, which the indicator alone has no way to know.
The third limitation is the absolute scale. An ATR of eight means something different at a price of 2000 than at a price of 5000, and it means something different across two contracts with different dollar multipliers. Traders who need to compare across instruments or across long price spans often convert ATR to a percentage of price or to dollars per contract first. The honest use of ATR is local and relative: how wide are this instrument's bars now, compared with this instrument's bars recently.
↳ Note
A stop measured in points ignores the session. A stop measured in ATR asks the market how much room the trade needs before it draws the line.
The way to keep ATR honest in a process is to log it. Tag every trade with the ATR reading at entry and the multiple used to place the stop, then check whether the stop distance was matched to the outcome. If trades stopped at one-and-a-half times ATR keep getting hit and then reversing to target, the multiple is too tight for the setup. That check turns ATR from a default setting into a measured input, held to the same twenty-trade-per-setup minimum any pattern claim needs.
Frequently asked questions
- q: What ATR period should a futures trader use? a: Wilder's default is fourteen periods. Shorter periods react faster and produce a noisier ATR; longer periods are smoother and slower to move. The period matters less than fixing it in advance and holding it, so stop distances stay comparable across trades on the same setup.
- q: Does ATR tell you which direction price will move? a: No. ATR measures the size of recent ranges, not their direction. A high ATR can occur in an uptrend, a downtrend, or a choppy range. Direction has to come from a separate read; ATR only sizes how much room the trade needs.
- q: How many bars does ATR need before the reading is stable? a: The first ATR is a simple average of the lookback's true ranges, and Wilder's smoothing carries that seed forward for many bars. A common rule is to let several lookback lengths print before trusting the value, which for a fourteen-period ATR is roughly forty bars of warm-up.
- q: What ATR multiple should a stop use? a: There is no universal multiple. One-and-a-half to two times ATR is a common starting band, but the right multiple is the one a trader's own results support across at least twenty trades in the matching setup. Fix it, log it next to each trade, and measure whether the stop distance fits the outcomes.
- q: Can ATR be compared across two different futures contracts? a: Not directly. ATR is expressed in each instrument's own points and dollar value, so an ATR of eight on one contract and eight on another describe different dollar ranges. Convert to dollars per contract, or to a percentage of price, before comparing across instruments.