How to Use the ATR: Work It Out by Hand, Set the Stop, Size the Trade

Michael Whitman · Trading since 2017, full-time since 2020 · YouTube creator, 137K subscribers · Published 2026-09-07 · Updated 2026-11-02 · Reviewed 2026-11-01

To use the ATR, treat it as a ruler: it tells you how many dollars this market covers in a normal bar, and you place your stop a set number of those rulers away, then let the stop distance decide how many shares you buy.

Below, SPY’s 30 Oct 2026 reading is rebuilt by hand from the real bars, turned into a stop and a share count, and then checked against 32 years of how often stops like it were reached.

Before you start

The indicator’s period chosen on purpose, with 14 as Wilder’s default. Every number on this page uses 14 daily bars. Changing the period changes every figure that follows, so decide once.

Your account risk per trade in dollars. The example uses a hypothetical $10,000 account risking 1%, which is $100. Without a dollar figure the ATR has nothing to divide into.

A stop multiple of the ATR written down before the trade. 1.5, 2 and 3 are the multiples tested further down. Write yours on paper before looking at today’s chart, so the chart cannot talk you into a different one.

The steps

1. Write down the last 15 bars: high, low and close

2. For each bar, take the largest of high minus low, high minus prior close, and prior close minus low

3. Update the ATR: prior ATR times 13, plus today’s true range, divided by 14

4. Multiply the ATR by your written multiple to get the stop distance

5. Subtract that distance from a long entry, or add it to a short entry

6. Divide your dollar risk by the stop distance and round the share count down

7. Divide the ATR by the close and compare that percentage with the instrument’s own history

8. Leave the stop where it is once the trade is open

Steps one to three are the calculation, which a charting platform does for you; doing it once by hand is how you find out which version of the formula your platform is running.

Steps four to six are the part that changes your order ticket. Step seven tells you whether today’s distance is wide or narrow for this instrument, and step eight protects the risk you set in step six.

A worked example: SPY’s ATR by hand, 23 to 30 Oct 2026

Start from a known reading. SPY’s 14-session Wilder ATR at the close of Friday 23 Oct 2026 was $7.2197, carried forward from February 1993 through every session since.

From there each new session needs only its own true range and one line of arithmetic.

Check one row yourself. For 26 Oct: ($7.2197 x 13 + $7.76) / 14 = $101.6161 / 14 = $7.2583. For 30 Oct: ($6.8563 x 13 + $19.00) / 14 = $108.1319 / 14 = $7.7237. Carried at four decimals, each row lands on the full-precision value.

Look at what the gap did on 30 Oct. The day’s own range was only $11.10, from $761.46 down to $750.36. SPY had closed at $769.36 the session before, so the bar actually sat $19.00 below where the prior day ended. Using high minus low alone would have recorded $11.10 and understated the move by $7.90. That one bar lifted the ATR from $6.8563 to $7.7237.

Now the stop and the size. SPY closed at $755.72 on 30 Oct, 1.022% of which was the ATR.

At 2 ATRs the trade is 6 shares, about $4,534.32 of SPY, with $92.68 at risk because the count is rounded down rather than up. Wider multiples buy fewer shares for the same money at risk, which is the whole point of the division in step six.

Run the same rule on a louder day. On 31 Mar 2026 SPY closed at $650.34 with an ATR of $10.6715, 1.641% of price and the highest share of any 2026 session through 30 Oct. Two ATRs came to $21.34, the stop went to $629.00, and $100 bought 4 shares. Same rule, same dollar risk, a third fewer shares.

Why your platform’s ATR may not match yours

There are several formulas wearing the same name. On the 30 Oct bar:

Wilder’s version gives each new bar a weight of 1/14, so a single wide bar fades slowly. A plain 14-bar mean drops an old bar all at once, and an EMA weights the newest bar at 2/15, nearly double. The 30 Oct gap bar pushes the last two higher because they lean harder on recent sessions.

The starting point matters for a few weeks, then stops mattering. Restarting Wilder’s formula fresh at 165 points across SPY’s history, the restarted line came within 1% of the full-history value after a median of 24 sessions and at most 66. If your hand calculation is a few cents off, check how far back it starts before checking the arithmetic.

Reading the ATR against SPY’s own past

A distance means more next to its own history. SPY’s 1.022% on 30 Oct was higher than 37.13% of the 8,263 daily readings since 3 Jan 1994; the median across that span was 1.195%. Within 2026 the reading ran from 0.779% to 1.641%.

Across the whole sample the spread is far wider. The lowest reading was 0.375% of price on 18 Oct 2017 and the highest 7.913% on 27 Oct 2008, about 21 times apart. A dollar stop that suited one of those days would be absurd on the other, which is why the stop gets measured in ATRs at all.

The chart shows that idea on the four months leading into the worked example. The line under the closes is each session’s close minus two ATRs, the level a fresh long would have used that day. On 2 Jul the gap between close and line was $20.31; by 10 Sep it had narrowed to $12.07, because the ATR had fallen from $10.16 to $6.03 over that stretch.

How to tell it worked

Pull your last 30 trades and write each stop distance in ATRs at entry. If you planned 2 and the list runs from 0.9 to 3.4, you are choosing the stop by eye and labeling it afterwards.

Compare each stopped trade’s dollar loss with your planned risk. Every one within 10% of the plan means the sizing step held; a loss at 2 times the plan means the stop moved or the size was guessed.

Check one ATR value a month by hand against the platform. The two should agree within a few cents once you match the method and start date from the sections above. Data vendors can also differ by a cent on a high or low: one other source shows SPY’s 30 Oct low at $750.37, which moves that day’s ATR by under a tenth of a cent.

Count stops hit within 2 days of entry. If that is more than a third of all stops, compare your multiple with the reach rates in the next section before changing anything else.

The original data

Sample: SPY daily bars from Yahoo Finance, 29 Jan 1993 to 30 Oct 2026, 8,497 sessions after dropping bars with no range (raw split-adjusted prices, no dividend adjustment).

True range and Wilder ATR exactly as in steps two and three, first value the mean of the first 14 true ranges. Stop test: for each session from 3 Jan 1994 with ten later sessions available (8,253, the last on 16 Oct 2026), did any low in those ten reach the close minus the multiple times that day’s ATR.

Then split by how busy the market was. Ranking those sessions by ATR as a share of price gives three groups of 2,751: calm (0.375% to 0.979%), middle (0.979% to 1.515%) and busy (1.516% to 7.913%). A 2-ATR stop was reached 35.33%, 33.22% and 32.21% of the time in those groups. A stop a fixed 2% below the close was reached 26.54%, 40.02% and 60.31% of the time across the same groups.

What this shows and what it does not. A reach rate is how often the low traveled that far; it is not a loss rate, because a position can be stopped and the price can recover afterwards, and it is not the result of any trading rule.

It is one index fund on the daily chart. Individual stocks jump between sessions more than an index fund does, so expect their reach rates at each multiple to differ from these. Every session’s true range, ATR and ten-session low is in the SPY ATR file.

When it fails

The stop is reached by a gap, so the fill is worse than the level. Of the 2,772 two-ATR stops that were reached, 408 (14.72%) were first reached by a bar that opened at or below the stop, so a stop order would have filled near the open instead.

The median overshoot was 10.96% of the stop distance. The worst came after 19 Feb 2020: close $338.34, ATR $2.90, stop $332.54. On 24 Feb SPY opened at $323.14, $9.40 under the stop and 1.62 stop distances beyond it.

The reading can be lowest right before it is needed. It averages bars already printed. SPY’s smallest ATR share of the sample was in October 2017, and a quiet stretch shrinks every stop sized from it.

It is read as direction. A rising ATR in a falling market means big bars, not a turn.

A stop 1 ATR away is reached more often than not. Within ten sessions, 58.89% of them were.

The stop gets widened on an open trade. Today’s ATR belongs to the next trade; moving an open stop raises a risk already fixed in step six.

Questions people ask

What is a good ATR setting? Fourteen bars is Wilder’s original and the default on most platforms. Whatever you pick, use the same length every time, or a stop of “2 ATRs” means a different distance each week.

How many ATRs away should a stop go? That is a choice, not a fact. On SPY daily bars since 1994 a 1.5-ATR stop was reached within ten sessions 45.27% of the time and a 3-ATR stop 18.13%; the wider stop buys fewer shares for the same dollar risk, as the 30 Oct table shows.

Does a rising ATR mean the price will go up? No. It went from $6.8563 to $7.7237 on a day SPY fell $13.64. It measures size only.

Why is my ATR different from the chart’s? The platform may use a simple mean or an EMA instead of Wilder smoothing, or start from a different date; the four-method table above shows how far apart they land on one bar.

Can I use it on intraday charts? The arithmetic is the same on any bar size. The figures on this page are daily SPY only, so measure your own timeframe before borrowing any of these rates.

Related

The ATR page explains what the indicator is and why a gap belongs inside true range. Stop loss covers where a level belongs before you check it against the ATR distance from step four.

Position sizing and risk per trade cover the dollar figure that step six divides. For a stop that moves with the trade instead of staying put, see the ATR trailing stop, and for why calm and loud stretches arrive in runs, see volatility clustering.

This page is educational, not financial advice. Test every idea on your own charts before risking money.