ATR Indicator: Reading Volatility on the Chart and Sizing From It

What the line under your chart is counting

Every bar has three candidate distances, and the indicator takes the largest of them. Two of the three reach outside the bar, which is the part that surprises people who expect a range:

Take the last bar of the series above: high 2,480.00, low 2,420.00, previous close 2,420.00. High minus low is 60.00, the gap from the previous close to the high is 60.00, and the gap to the low is 0.00, so the true range is 60.00. The fourteen-bar reading going into that bar was 15.21. One new bar cannot replace it — the recursion takes one fourteenth of the jump and keeps thirteen fourteenths of what was already there:

That single step is the personality of the indicator. A news bar three times the normal size moves the reading by less than a third of the difference, and the same bar moves a seven-period reading by roughly twice as much as it moves a fourteen-period one. Slow is the design, not a defect: the reading is meant to survive an outlier, not to become one.

Nothing in that arithmetic has a sign. The indicator rises in a crash and it rises in a melt-up, because both are travel. It is a measuring tape, not a signal, and the only honest use of it is as an input to a distance you are about to pay for.

Picking the period: you are choosing memory, not sensitivity

The series behind this page spends 44 bars in a band about 10 wide, then 16 bars in a band about 18 wide with one 60-point outlier at the end. That shape is why the long window reads lower than the short one: it is still averaging bars that no longer look like the ones on the screen. Below is the same series read four ways, with a 3,000.00 budget, a stop at twice the reading, and the gold contract's 0.01 pip size, 100-unit lot and 1-unit step.

Read the two panels together, because that is the trade. The left falls as the window lengthens; the right rises. Nothing about the account changed between the columns — only how much of the recent past the reading is willing to remember.

So the period is not a dial you turn to make the number come out larger or smaller. It decides how long a change in conditions has to persist before your stops and your size reflect it, and each setting charges you somewhere:

• 7 to 10 bars. The reading follows the market almost immediately, which is what you want if you hold for a day or two. It also moves on its own when nothing has changed, so you re-size often, and every stop move costs a spread.
• 14 to 20 bars. The default in most chart packages for a reason: on daily bars it catches a change in conditions within a week or so and it is not thrown off by one wide bar.
• 50 bars and up. Barely moves at all, which is comfortable when you hold for weeks and dangerous right after a regime change — it is still describing the market you had, not the one you have.

Pick one period per instrument and write it down. Changing the period to justify a different size on a trade you already want is not analysis; it is the same as moving the stop until the size comes out right.

The reading is in price; the money lives in the contract

The indicator prints in whatever units the symbol is quoted in. That is a distance, and a distance is not money until two more numbers are attached to it: how big a pip is on that symbol, and how many units sit in one lot. Those two come from the contract, not from the chart, and they are exactly the numbers that stop a volatility stop from being portable between symbols.

Below is every symbol in this site's contract table where one lot is one unit — 57 of them. A lot and a unit are the same thing on all of them, so the sizing step is unusually simple. The pip sizes are not simple: they range from 1 down to 0.0001.

The two right-hand columns are the whole lesson. Take a reading of 20.00 of price and place a stop two readings away, 40.00 of price:

The pip count differs by a factor of 10,000 across those three rows while the money is identical to the cent. Both are true at once, and only one of them belongs in your sizing: a pip count is a distance once you know the pip size, and it is money once you also know the contract size. "I risk 40 pips" is not a risk statement. "I risk 40.00 USD per lot" is.

The minimum step in that table is the third contract number, and it is the one that never appears in a piece about volatility. It decides how much of the size you computed can actually be placed. Say a 1,000.00 budget meets a stop distance of 37.50 of price — 26.667 units on any of these symbols:

On the index the size has to be placed as 26 units, and 25.00 of the budget goes nowhere; on the crypto contract the step is 0.001, so 26.666 goes in and 0.02 is left over. The budget did not change and the reading did not change — the contract decided how much of your own risk you were allowed to use. That is the second reason a volatility stop has to be carried in money: the same distance is not the same trade on two symbols with the same price.

This is also why a stop copied from a forum post is meaningless across markets. Forty pips on US30 is 40.00 per lot; forty pips on a 0.0001-pip symbol is 0.0040 per lot. Same sentence, ten thousand times the difference.

When the reading rises, the size falls

Here is the same fourteen-bar reading taken at two points in the series: bar 44, at the end of the quiet stretch, and bar 60, after sixteen wider bars and one 60-point outlier. Same account, same 3,000.00 budget, same multiple, same contract.

The stop got 69% wider and the size fell by 41%. Those two numbers are not independent: size is the budget divided by the distance, so the size ratio is the inverse of the distance ratio. 10.91 divided by 18.41 is 0.593; 81 divided by 137 is 0.591. The gap between them is the rounding down to the 1-unit step, nothing else.

What happens in practice is the opposite move. The trader who sized 137 units during the quiet stretch keeps 137 units when the reading widens, because 137 units is what they are used to and the chart still looks like the same market. At the wider stop that is 137 x 36.82 = 5,044.34 — 168.14% of the 3,000.00 budget, or 2,044.34 more risk than the plan allowed, on a trade where the setup did not change and the market simply got noisier. No platform warns about this. The margin figure looks fine and the ticket goes through.

The same relationship runs in reverse, and it is the direction people resist: when the reading falls, the size rises. A quiet market is exactly when a larger position feels safest, and it is also the only time a fixed budget genuinely pays for one. The budget has not moved; the distance it has to cover has.

One limit worth writing into the rule. An ATR stop is a floor, not a target. If twice the reading is smaller than the spread plus a tick on that symbol, the stop is inside the cost of getting in and the trade is not available at your risk — the fix is a different symbol or a different trade, never a tighter stop that the noise will take out for free. Costs belong in the distance, and they are covered on the spread and commission page.

Where the number enters the order

The reading belongs at one specific point in the sequence, and putting it anywhere else is how it turns into a story about the market rather than a distance in money:

1. Read it on the symbol and the timeframe you are trading. A fourteen-period reading on a five-minute chart and a fourteen-period reading on a daily chart are different distances with the same label.
2. Multiply by your multiple to get a stop distance in price, and round to the pip.
3. Turn that distance into money per unit using the pip size and the contract size.
4. Divide the risk budget by it and round down to the minimum step.
5. Multiply back to see the worst case, and compare it to the budget you started with.
6. Only then look at margin and at any per-symbol limit.

A trailing stop at a fixed multiple does the same thing repeatedly: the money at risk per lot stays put while the pip distance breathes with the market. That is the appeal. It does not make the stop a guarantee — gaps and spikes fill wherever the market is, so the realised loss can exceed the distance the reading suggested.

Before you send the order

• One period per instrument, written down before the trade, changed when your holding period changes and not when a size needs justifying.
• The stop distance travels in money per unit. A pip count only travels with the pip size and the contract size attached.
• Re-size when the reading moves, not when the profit and loss moves.
• After every re-size, re-check the worst case against the budget, then against margin.
• If twice the reading is inside the spread plus a tick, the trade is not available at your risk.

Size the trade from the reading

Questions traders ask

One you can hold to, matched to how long you are in the trade. On the series on this page a 7-bar window reads 22.83 and a 50-bar window reads 13.60 from exactly the same prices — a 68% difference with no new information, only a different amount of memory. Fourteen is the common default on daily bars; shorter if you are out within a day or two, longer if you hold for weeks.

Because the reading moved, and size is the budget divided by the distance. At ATR(14) of 10.91 a 3,000.00 budget with a stop at twice the reading pays for 137 units; at 18.41 the same budget pays for 81. Nothing about the account changed — the stop got 69% wider, so the size fell by 41%.

No. It means price is travelling further per bar, in either direction. The true range is the largest of three distances and none of them carries a sign, so the reading climbs in selloffs and in rallies alike. It is a measuring tape; the direction has to come from somewhere else.

It depends on the pip size, which is a property of the contract and not of the indicator. A reading of 20.00 of price is 20 pips on US30 with a pip size of 1, 2,000 pips on BTCUSDT with a pip size of 0.01, and 200,000 pips on a symbol with a pip size of 0.0001 — and all three are 40.00 USD per lot at a stop two readings away. Convert to money per unit before the number means anything.

It keeps the money at risk per lot constant while the distance breathes with volatility, which is the main argument for it. Two things it does not do: it does not cap the loss, because gaps fill worse than the level, and it does not stop the distance from drifting far from where you placed the trade if the reading keeps climbing.

Only if your stop is actually going to sit that far away. The reading belongs to the timeframe it was measured on: a daily reading is the travel in a day, and using it for a stop you intend to honour inside an hour gives you a distance the trade will not survive. Measure on the timeframe whose bars you are risking against.

Related guides

• Average true range calculator — how the number is computed, bar by bar: the three distances and Wilder's recursion. This page reads the result; that page derives it.
• Stop loss calculator — the distance a budget can pay for, once the reading has been turned into money.
• Lot size calculator — units, lots and minimum steps across the contract table.
• Crypto futures position size — sizing on the contracts that make up most of the table above.
• Spread, commission and position size — the costs that belong inside the stop distance.