How do you measure how much a stock moves on a typical day?
Average True Range, from the same 1978 Wilder book as RSI, answers one narrow question extremely well: how far does this instrument travel in a typical session? It says nothing about direction — and that restraint is what makes it useful.
True Range first
The naive answer is the day's high minus its low. That understates any session that gapped overnight, so Wilder defined True Range as the largest of three candidates:
- today's High − today's Low
- |today's High − yesterday's Close|
- |today's Low − yesterday's Close|
Example 1 — no gap. Yesterday's close 101; today High 103, Low 99. Candidates: 4, 2, 2 → TR = 4, just the day's range.
Example 2 — a gap. Yesterday's close 95; today High 103, Low 99. Candidates: 4, 8, 4 → TR = 8. The overnight jump counts. The plain range would have reported half the real movement — which is the entire reason True Range exists.
Both sessions are drawn below, and they are the same session twice: identical high, identical low. All that differs is the close before them.
Then average it
ATR(14) is the Wilder-smoothed average of those true ranges:
ATR_today = (ATR_yesterday × 13 + TR_today) ÷ 14
with the first value being a simple mean of the first 14 true ranges.
Reading it without embarrassing yourself
ATR is quoted in price units, not percent. An ATR of 5 on a $500 stock is a 1% daily wander; an ATR of 1 on a $20 stock is 5%. The second stock is five times livelier and shows the smaller number. Always divide ATR by price before comparing two instruments — this is the most common mistake made with the indicator.
It is also completely direction-blind. A violent crash and a violent rally of the same size produce the same ATR. That is a feature: it lets you ask "how big is a normal day here?" without contaminating the answer with an opinion about where price is heading.
What practitioners do with it
Described, not prescribed: because ATR scales with the instrument's own conditions, some traders express stop distances and position sizes as multiples of ATR, so the same rule automatically widens in wild markets and tightens in calm ones. Others use it purely as a yardstick — is today's move large by this stock's own standards? The Risk Management for Traders course develops the sizing arithmetic properly; here the point is only what the number means.
Limits
ATR is a Wilder-smoothed average — weighted heavily toward the last fourteen bars but never fully forgetting the earlier ones — so it rises after volatility rises, and then takes a while to come back down. The lag lesson again. Earnings gaps, halts, thin illiquid names and long holidays all distort it. And a high ATR is not a warning or an invitation; it is a measurement of recent restlessness.
Which is exactly the trouble: the thing being measured arrives in clumps, and the measurement is an average of the clump that just finished.
In the data
The pane under Apple's five years of candles is its 14-day ATR, built from the same true-range definition as above:
Read the pane's scale: it is in dollars, not percent. Part of any change in it over five years is simply the price having moved, which is why two stocks' ATRs, or one stock's ATR five years apart, say nothing side by side until each has been divided by its own price.
Try it now
- Compute the true range of both marked sessions on the first schematic. Same high, same low, and the answers differ by a factor of two — write down which of the three candidate terms won in each case.
- Now convert the measure into something comparable. The latest 14-day ATRs of the S&P 500 fund and the small-company fund are the first two tables below; the third holds each fund's latest price. Divide each ATR by its own price. Compare the percentages, not the point values, and notice how differently the two rank once the currency is out of the way. The year of each is under the tables, with its ATR in the pane under the bars.
- Below is a session where Apple opened outside the previous day's range, with the day before it. Verify that its true range was set by the gap rather than by the day's own high minus low.