ATR Channel

Average true range was introduced by J. Welles Wilder Jr. inNew Concepts in Technical Trading Systems (1978). True range is the greatest of three distances: the bar’s own high minus low, the high minus the previous close, and the low minus the previous close. Including the previous close is what makes it honest about gaps — a bar that opens far from yesterday’s close and barely moves has still travelled a long way, and a plain high-minus-low would call it quiet.

ATR measures movement, not direction. That makes it the natural unit for the two decisions that are about distance rather than direction: how far away a stop has to sit to survive normal noise, and how large a position can be for a given risk budget. Volatility-based position sizing — risk a fixed fraction of capital divided by some multiple of ATR — is the standard application, and the core of the Turtle rules.

Wrapping a trend line in ATR bands is the same idea Chester Keltner’s channels and their modern ATR variants apply to a moving average. Here the centre is aleast-squares fit through the bars you selected rather than an average, so the channel describes a specific stretch of trend rather than a rolling window.

The tool takes 2 clicks, which select a time range. It then fits a least-squares line through the chosen price source over the bars in that range, computes ATR over the same bars, and places a boundary a multiple of that ATR above and below the fit. A minimum of three bars is required before a fit exists.

Only the time of each anchor matters. Their vertical position is ignored, hence the horizontal resize cursor on both grips. Dragging an anchor changes which bars are in the sample; the lines then go where the data puts them. There is no width grip, because the width is a property of the bars — set the multiplier in the properties panel instead.

ATR is averaged arithmetically over the selected window, not with Wilder’s smoothing. This is a deliberate departure. Wilder’s recursion carries weight from bars before the window, which would make the channel depend on data the user did not select — two identical selections could then produce different channels depending on what preceded them. A drawing tool measures a bounded, user-chosen sample, so the average is taken over exactly that sample. The first bar of the slice has no previous close available and falls back to its own high minus low.

Each boundary end is converted from its own price independently, so a constant ATR offset stays a constant price offset rather than becoming a constant pixel offset that would land at the wrong price on a logarithmic axis.

Both tools fit the same line. They disagree about what the bands should measure, and the difference is not cosmetic.

A quiet, steady drift gives narrow bands on both. A choppy market that ends where it started gives wide bands on both. But a market that trends hard in small steady bars gives wide regression bands and narrow ATR ones — and that divergence is the case where having both tools pays for itself. It says the trend is strong and orderly: price is far from where it started but each individual bar is calm.

Stop placement: this is the primary use. A stop inside one ATR of entry will be hit by ordinary noise; the conventional distance is 1.5 to 3 ATR depending on how much room the strategy needs. The Band readout gives that distance in price directly, so the channel boundary is a stop level you can read off the chart.

Position sizing: divide the risk you are willing to take on the trade by the band distance to get a position size that carries the same risk regardless of the instrument’s volatility. This is what makes results comparable across a portfolio — a fixed share count risks wildly different amounts on a calm and a volatile instrument.

Mean reversion at the boundary: a close beyond a boundary means price has moved further than the recent bar range makes routine. In a range-bound market that is a candidate reversion trade back toward the fitted line. In a trending market the same event is more often continuation, so the higher timeframe context decides which reading applies.

Volatility regime detection: draw the channel over two adjacent periods of equal length. If the ATR figure has materially increased, the market has entered a higher-volatility regime and every distance-based decision — stops, targets, size — needs rescaling.

1. 1
   Select the trend

   Click the start and end of the stretch you want measured. Only the times matter. Include enough bars that both the fit and the ATR mean something — the minimum is three, but a few dozen is more useful.
2. 2
   Check the fit

   The dashed centre line should look like the trend you had in mind. If it does not, the selection is probably spanning two different regimes.
3. 3
   Set the multiplier

   Adjust it until the boundaries sit just outside the normal swings. That multiple is then the number of ATRs your stop needs.
4. 4
   Read the band

   The readout’s Band figure is the boundary distance in price. Use it for the stop distance and for position sizing.
5. 5
   Project forward

   Set Extend to Right to carry the boundaries past the sample. Remember they are extrapolated from the selected bars, not recomputed on new ones.

• A rising ATR across successive selections means larger bars: widen stops and reduce size, or the same strategy now carries more risk than it did.
• Bands much wider than the regression channel over the same bars mean the market is choppy at the bar level but the trend is orderly. Bands much narrower mean the opposite.
• Price rarely closing outside a 2 ATR channel is normal. Repeated closes outside mean either a volatility expansion in progress or a fit that no longer describes the trend.
• ATR is directionless. A wide channel says nothing about which way price will go, only how far it is likely to travel.
• Because ATR is priced in the instrument’s own units, values are not comparable across instruments. To compare, divide the band by price and compare the percentages — or use the Envelope Channel, which works in percent by construction.

Multiple channels for a trade plan: place two ATR channels on the same selection with different multipliers — say 1 and 3. The inner boundary is a reasonable trailing-stop distance while a trade is working; the outer is the point at which the move has gone far enough that the thesis is either confirmed or the position should be reduced.

Gap-heavy instruments: because true range includes the previous close, this channel is markedly wider than a high-minus-low measure on instruments that gap between sessions. That is correct behaviour and is precisely why Wilder defined it that way — a stop sized from intraday ranges alone will be gapped through.

Source selection for wide-ranged bars: fitting to HL/2 rather thanclose makes the centre line describe the middle of the trading range rather than the settlement path. On instruments with long wicks the two lines can sit visibly apart, and HL/2 is usually the better centre for a channel whose purpose is to bound the range.

• 1.New Concepts in Technical Trading Systems — J. Welles Wilder Jr. (1978)
• 2.Average True Range (ATR) — Investopedia
• 3.Average True Range — ChartSchool — StockCharts
• 4.Keltner Channels — Investopedia
• 5.Way of the Turtle — Curtis Faith
• 6.Technical Analysis of the Financial Markets — John J. Murphy