What ATR actually is

J. Welles Wilder introduced the Average True Range in his 1978 book New Concepts in Technical Trading Systems, the same book that gave trading RSI. He designed it with commodities and daily prices in mind.

The building block is True Range (TR), which is the greatest of three measurements:

• the current high minus the current low

• the current high minus the previous close, as an absolute value

• the current low minus the previous close, as an absolute value

Why three? Because plain high-minus-low misses moves that happen between bars. If a market gaps, the jump from yesterday's close to today's range is real movement that a simple range calculation never sees. True Range extends the measurement to the previous close whenever that close sits outside the current bar.

ATR is then a smoothed average of True Range, over 14 periods by default. The first value is a simple average of the first 14 TRs, and after that Wilder's own smoothing takes over:

 Current ATR = (Prior ATR x 13 + Current TR) / 14

This is the same 1/N smoothing Part 15 flagged inside RSI, with the same practical consequence: different platforms can show slightly different ATR values for the same chart.

Compute it by hand once

Three bars are enough to learn the whole indicator. Assume the previous close before bar 1 was 100.00.

Two lessons sit in that box. Bar 2 is the reason True Range exists: the range alone said 3, but anyone holding through that gap experienced a move of 7. And the smoothing line shows ATR's temperament: one wild bar nudges a 14-period ATR from 2.00 to 2.36, so the reading describes the market's recent character, not its latest outburst.

One honest note for perpetuals traders: crypto trades around the clock, so the gap methods fire far less often than on stocks or commodities that close overnight. On a 24/7 chart, ATR usually just equals the average bar range. The construction still matters, because wicks and violent single bars are exactly what the average is smoothing.

The two numbers ATR gives you

Everything practical about ATR comes down to two calculations, done in this order.

1. Stop distance. A stop-loss exists to detect that the trade idea failed, and a stop placed closer than one ATR sits inside the market's ordinary per-bar travel, which means routine noise can trigger it with the idea still intact. The common convention is a multiple: a stop 2 x ATR away from entry, placed beyond the structure level your idea depends on. The multiple is a convention rather than a tested constant, but the principle is arithmetic: the stop should sit outside normal movement, and ATR measures normal movement.

2. Position size. Once the stop distance is known, size stops being a feeling and becomes division:

 POSITION SIZE FROM ATR
   account risk  = account x risk per trade
                 = 10,000 USDT x 1%              =  100 USDT
   stop distance = 2 x ATR = 2 x 800             =  1,600 USDT per BTC
   position size = account risk / stop distance
                 = 100 / 1,600                   =  0.0625 BTC
   notional      = 0.0625 x 64,000               =  4,000 USDT

The order of operations is the entire lesson: volatility set the stop, the stop set the size, and leverage fell out at the end as a consequence (4,000 notional on a 10,000 account is 0.4x). Most traders run this pipeline backwards. They pick 5x leverage first, which on this account is 50,000 notional, about 0.78 BTC, and then the same ordinary 1,600-dollar move costs 1,250 USDT: 12.5% of the account on a single routine fluctuation. Same market, same volatility, same stop distance. The only difference is which number was chosen first.

What ATR is not

ATR's limits are as clean as its uses, and Wilder built them in deliberately.

It is not directional. The reference is explicit: ATR "does not provide an indication of price trend, simply the degree of price volatility." A rising ATR means bars are getting bigger. It does not say which way the next one points, and nothing in the formula could: every input is an unsigned distance.

It is not comparable across assets. Like MACD in Part 16, ATR is denominated in the asset's own price units. As the reference puts it, low-priced instruments have lower ATR values than high-priced ones, and "because of this, ATR values are not comparable." An ATR of 800 on BTC and 0.02 on a small altcoin may describe identical volatility. To compare, divide ATR by price: 800 on a 64,000 BTC is 1.25% per bar, and that percentage travels across assets when the raw number cannot.

It has no overbought level. There is no ATR equivalent of RSI's 70. High ATR means wide bars, low ATR means narrow ones, and neither is a signal to do anything except adjust your measurements.

What the evidence says, briefly

This series spent five parts feeding directional indicators through data-snooping corrections, and the machinery goes quiet here for a simple reason: ATR makes no directional claim, so there is no win rate to test and no crossover to backtest. It was not among the strategy families in the 2023 crypto study the earlier parts drew on, and no tested result for it appears in this series' evidence base; none is claimed either. The honest questions are narrower. Is the measurement well built? Yes: the gap handling is exactly what a naive range calculation misses. Do the uses built on it follow logically? They do, because both are restatements of "put the stop outside normal noise and size off the stop." ATR earns its place not by predicting anything but by being the input that makes other decisions computable.

Reputation versus evidence

Using ATR honestly

• Use it as a ruler, not a signal. ATR answers "how much does this market move per bar on this timeframe." Every use flows from that answer; none flows from watching the line for crossovers.

• Set the stop first, the size second. Stop distance as a multiple of ATR beyond structure, then position size as account risk divided by stop distance. Leverage is the output of that pipeline, never the input.

• Compare volatility in percent, not points. Divide ATR by price before comparing assets or the same asset at different price levels.

• Match the ATR timeframe to the trade. A stop for a 4-hour swing trade should be measured with a 4-hour ATR. A 5-minute ATR describes a different market.

• Recheck it when conditions change. ATR from last month's quiet range understates today's expansion; the ruler only helps if it measures the market you are in.

Common mistakes

• Placing stops inside one ATR - that distance is the market's ordinary travel, so noise alone can spend your stop.

• Comparing raw ATR across coins - the units are the asset's own price; 800 versus 0.02 says nothing until you divide by price.

• Reading an ATR rise as a buy or sell signal - it reports bar size, not direction.

• Choosing leverage first and finding the stop later - that runs the sizing pipeline backwards and turns routine noise into large percentage losses.

• Using one ATR setting for every timeframe - the 14-period default measures whatever chart you apply it to, and a stop built from the wrong chart is the wrong stop.

ATR on ApeX Omni

ATR is arguably more useful on a perpetuals venue than anywhere else, because leverage punishes unmeasured stops. The workflow on ApeX Omni: read the ATR on the timeframe you are actually trading, put the stop a multiple of it beyond the structure level your idea depends on, then divide the amount you are willing to lose by that stop distance to get position size, as in the worked example. Two platform-specific checks follow. First, confirm the stop distance sits well inside your liquidation buffer: Part 9 showed how quickly the liquidation price closes in as leverage rises, and a stop that would trigger after liquidation is not a stop. Second, trigger the stop on the mark price rather than the last price, as Part 12 argued, so a single wick on the venue's own feed cannot spend it. Done in this order, ATR quietly fixes a classic account-killer in perps trading: position sizes chosen by feel in a market whose typical move was never measured.

The bottom line

ATR is a measuring tape. It tells you, in the asset's own units, how far the market has typically travelled per bar over the last fourteen bars, gaps included, spikes smoothed. That is all it does, and that is precisely why it is so practical: the two decisions that decide whether a trading account survives, where the stop goes and how big the position is, are both unanswerable without exactly this measurement. It will not tell you where price is going. It will tell you what surviving the journey costs, which is the question most traders never ask until the answer is taken out of their account.

Next in this series: ADX, Wilder's other 1978 invention, and the question the moving-average whipsaws of Part 14 begged: is this market actually trending at all?

Frequently asked questions

What does ATR measure? The average size of a bar's true movement over the look-back period, 14 by default. True Range is the greatest of: high minus low, high minus previous close, and low minus previous close (both as absolute values), so gaps between bars are counted as movement.

Who created ATR? J. Welles Wilder, in his 1978 book New Concepts in Technical Trading Systems, the same book that introduced RSI. He designed it with commodities and daily prices in mind.

How do I use ATR for a stop-loss? Place the stop a multiple of ATR beyond the level your trade idea depends on, commonly 2 x ATR. The principle is that a stop closer than one ATR sits inside the market's normal per-bar movement and can be triggered by noise alone.

How do I size a position with ATR? Divide the amount you are willing to lose on the trade by the stop distance. With 100 USDT of risk and a 1,600 USDT stop distance, the size is 0.0625 BTC. Leverage then falls out as a result rather than being chosen first.

Does ATR predict price direction? No. It measures the degree of movement only, and every input to the formula is an unsigned distance. A rising ATR means bigger bars, not a bullish or bearish signal.

Why is ATR different on every coin? Because it is expressed in the asset's own price units, so values are not comparable across assets. Divide ATR by price to get a percentage, which can be compared.

Explore more from this series: Part 9: Leverage and Liquidation | Part 12: Order Types | Part 17: Bollinger Bands | Part 18: Stochastic Oscillator

This article is for educational and informational purposes only and is not financial, investment, or legal advice. Do your own research before making any trading decision.