ATR: what measures the range of a candle and why the indicator is silent on direction

ATR (Average True Range) answers a single question: how widely an instrument moves over the period you have set for it. It is a smoothed average of the "true range", and nothing more. ATR does not show where the price will go, is not divided into overbought and oversold zones, and gives no entry points. Its place of work is the calculation of the stop and of position size, which is where abstract "volatility" has to be turned into a concrete distance on the chart. If you expect a buy signal from ATR, you are holding a ruler and trying to hammer in a nail with it.

What exactly ATR averages

The ordinary range of a candle is the distance from its high to its low. The problem is that this range loses gaps: if a stock opened far from yesterday's close and then spent the whole day in a narrow corridor, "high minus low" will show a calm that was never there.

Wilder, who proposed the indicator in the book "New Concepts in Technical Trading Systems" (1978), solved this through True Range. For each candle, the largest of the following values is taken:

• the distance from the high of the candle to its low;
• the distance from the high of the candle to the previous close;
• the distance from the low of the candle to the previous close.

The absolute value is always taken: True Range is non-negative by construction and has no sign. The last two values are the protection against a gap. They tie the candle to the previous close, so an overnight gap enters the calculation instead of disappearing from it.

True Range is then smoothed. Wilder used his own smoothing, close to exponential: each new value is blended into the accumulated average with a small weight. This gives an important property: ATR has a long memory. A spike in the range does not drop out of the indicator abruptly; it fades gradually, and this has to be kept in mind when reading it.

Why ATR is silent on direction

ATR is symmetrical: a fall and a rise of the same amplitude produce the same value. This is not a flaw, it is the design. Practical conclusions follow from it.

A rising ATR does not mean a "reversal" or a "breakout". It means only that the range is widening, and the market can widen in either direction, including inside a sideways range with wide candles. A falling ATR does not mean "safe". A contraction of the range regularly precedes a sharp move, and it is at the quietest moment that a stop calculated from a narrow ATR sits closest to the price.

ATR is a lagging indicator lagging indicator: it describes what has already happened to the range. To expect a forecast from it is to confuse it with leading constructions leading indicators. ATR does, however, have a merit that many fashionable tools lack: it does not allow repainting indicator repainting, because a closed candle no longer changes the history of the indicator.

The period: what changes when you touch it

Wilder's value for the indicator period indicator period has become the industry standard, and most terminals fill it in by default. Changing it makes sense, but not arbitrarily.

A short period makes ATR responsive: it reacts quickly to a spike and calms down quickly. A stop built from such an ATR breathes with the market, and is knocked out by noise more often. A long period gives a stable estimate of the regime: the stop becomes steadier, but after a change of regime the indicator still remembers the earlier, no longer relevant volatility for a long time.

The choice between them is not a search for the "right" number but a matter of matching the horizon of the trade. It is sensible to keep the ATR period comparable to the time you plan to hold the position. Intraday trading on an ATR with a long period produces stops calculated for someone else's timeframe.

Where ATR really works: the stop and position size

The main value of ATR is that it turns risk from a percentage of the account into a distance in points.

Step 1. You set the share of capital you are prepared to risk on the trade. This is your decision; the indicator takes no part in it.

Step 2. You take the current ATR of the instrument and multiply it by a multiplier: how much "further than the usual range" the stop has to stand so as not to be knocked out by routine noise. You set the multiplier yourself as well and keep it constant: the point of the method is that ATR changes, not the rule.

Step 3. Position size is obtained by dividing the acceptable risk in money by the distance to the stop. The wider the stock moves, the fewer lots you take for the same risk.

This scheme gives comparability: a position in a quiet stock and a position in a wide-ranging stock carry the same risk, although the stops stand at different distances. The same technique underlies the ATR trailing stop, where the level is pulled up behind the price at a fixed multiple of the current range.

Compare ATR with tools that measure not range but speed: RSI answers a different question and is useless for calculating a stop. It is worth keeping in mind the difference between a measurement and an indicator signal indicator signal. ATR has no signal at all.

Limitations worth knowing in advance

ATR is expressed in price points, so it cannot be compared directly between instruments whose shares cost different amounts. For comparison it is normalised: divided by the price, so that you look at the relative range. Otherwise an expensive stock will always look "more volatile" than a cheap stock.

On illiquid instruments ATR is distorted: infrequent trades produce ragged candles, and the indicator shows a range that cannot be traded through without slippage. Look at liquidity in the list of stocks before trusting the calculation; an instrument card such as SBER gives an idea of the character of trading.

A separate caveat: the platform supplies market data, but its set of metrics does not include a ready-made ATR value for instruments. The indicator is calculated in a terminal or a charting service, from the very same quotes.

Range expansion around events

The range does not live in isolation; it grows around news triggers. Financial reports, dividend decisions, record dates, corporate meetings: on such dates ATR widens after the fact, once the move has already happened.

This leads to a practical rule: do not trust a stop calculated from a calm ATR if a report or a record date falls within the horizon of the trade. Check the events calendar, issuers' financial reports and the dividend calendar. The nearest payouts:

The news flow is convenient to follow in the news section, and the definitions used in this article are collected in the glossary.

What ATR does not do

It does not predict direction. It does not set entry levels. It does not say whether an instrument is expensive or cheap. It does not distinguish a range that grows on panic from a range that grows on euphoria. And it does not replace the decision about risk; it only translates that decision into a distance on the chart.

That is why ATR sits well alongside indicators of a different nature: trend indicators are responsible for direction, oscillators for speed, ATR for scale. An attempt to get everything from it at once ends with the ruler starting to forecast the weather.