Home / Blog / Why an average true range channel and a deviation channel disagree
TRADING GUIDE

Why an average true range channel and a deviation channel disagree

Put two volatility envelopes on the same gold chart and they will sometimes disagree sharply about how wide the market is. That is not an error in either one. They are measuring different things: one looks at how scattered the closing prices have been, the other looks at how much ground each candle covered from high to low. On a market that produces long wicks, that distinction is not academic.

📅 October 8, 2026⏱ 8 min readBy XAUUSDLiveChart Research Desk
Track gold in real time on the live chartOpen Live Chart →
WHY AN AVERAGE TRUE RANGE CHANNEL
XAU/USD…
01

What a Keltner channel computes

The centre line is usually an exponential moving average of closing prices. The upper and lower lines are that centre plus and minus a multiple of the average true range over a lookback. True range for a single candle is the largest of three quantities: the high minus the low, the absolute difference between the high and the previous close, and the absolute difference between the low and the previous close. The average of those values over the lookback is the channel width input.

Read that definition again, because the second and third terms are the interesting part. They exist so that a gap is counted as range. If gold opens well away from where it closed, the distance travelled between the two sessions is included in the measurement even though no candle covers it. That makes the measure appropriate for markets with discontinuities, and spot gold has one every weekend. The multiple applied to the average, commonly between one and a half and two and a half, is a free choice and it determines how often price reaches the edge.

02

True range is not the same as close dispersion

Here is the practical difference. A deviation based envelope looks only at closing prices. A candle with an enormous range that opens and closes at nearly the same price barely registers, because the close has not moved far from the average. The same candle produces a large true range, because the measurement uses the high and the low.

Gold produces that candle regularly. A scheduled release arrives, price travels a long way in both directions within the hour, and the hour finishes close to where it started. Everyone who was involved experienced a violent market. A measurement built on closes experienced almost nothing. So an average true range channel widens and a deviation channel does not, and the two envelopes now disagree about the state of the market. Neither is lying. If your question is how far a position might be shaken around, the range based answer is the relevant one. If your question is how much the settled price has been moving, the close based answer is. Most traders care about the first and use a tool built for the second without noticing.

One wide wick candle, two different answers about widthhigh of the candlelow of the candleopenclosetinybodytrue range input: the whole high to low distanceclose dispersion input: barely any changeso the range based channel flares outand the deviation based one stays narrow
03

The squeeze, defined by band position

The best known use of the two envelopes together is a definition rather than a prediction. When a deviation based envelope sits entirely inside a range based channel, the scatter of closes has fallen below the typical travel of a candle. In plain terms, bars are still moving around but they keep finishing near each other. That is compression with activity underneath it, which is a more specific condition than either envelope reports alone.

It is worth being clear about what that condition is good for. It identifies a state, and states are useful for deciding which approach applies. A compressed state argues against a method that needs follow through and in favour of waiting. It says nothing whatsoever about which direction an eventual expansion takes, for exactly the reason given in the previous section: both measurements are built from absolute distances, and absolute distances have no sign. Any rule that converts a squeeze into a side has added an assumption that the measurement does not contain.

04

Channel edges as a trailing stop

Because the width is a multiple of average range, the lower edge in an uptrend is a volatility scaled distance beneath a smoothed price. That is the same construction used by most trailing stop methods, which is why the edge often looks like a sensible place to exit. It also inherits the same awkward property: the distance expands after volatility expands, so the edge is furthest from price immediately after the move that made you nervous.

In practice this means a channel edge stop gives back more in fast conditions and less in quiet ones, which may or may not be what you want. If the intention is to risk a consistent amount, a width that changes with recent range works against you unless position size changes with it. The arithmetic of matching size to a volatility based distance is covered in the average true range stop discussion and the same reasoning applies here without modification. The channel does not become a different kind of tool because it is drawn as a line rather than quoted as a number.

05

Two charts labelled the same can disagree

There is more variation in implementation here than most people expect. The centre line may be an exponential or a simple average. The range term may use the smoothing method Wilder described or a plain average of true ranges. The lookback for the range term may match the centre line lookback or be shorter. And the multiple is arbitrary.

Each of those choices shifts the edges, sometimes noticeably during a volatile stretch. The consequence is that a rule which references a channel touch is not portable: the same rule on two differently configured charts produces different trades. Before testing anything that depends on the edge, write down all four choices. This matters more on gold than on slower markets, because fast conditions are exactly when the smoothing differences compound and exactly when you are least willing to re-examine your settings.

06

Where it misleads on gold

Three recurring problems. The weekend discontinuity is handled by the formula, which is a strength, but a single large gap inflates the average range for the whole lookback and the channel stays unnecessarily wide afterwards. If a gap is sitting inside your window, the width you are reading is partly a memory of Sunday rather than a description of now.

Second, an edge touch in a trending market is routine. The channel is a measurement of typical travel, so price exceeding typical travel during a strong move is expected. Reading a touch as exhaustion repeats the mistake that envelope indicators invite in general. Third, the smoothing on the centre line means direction changes arrive late, so using the slope of the centre as a trend filter builds in a delay you should quantify rather than assume away. The sensible place to keep a channel is as a volatility reference you check before deciding how much room a trade needs, which you can judge on the live chart by comparing the current channel width against the distance to the level you intend to trade against.

Q

FAQ

What is the difference between a Keltner channel and a deviation based envelope?

The width input. A Keltner channel uses average true range, which measures the full travel of each candle including gaps. A deviation envelope uses the scatter of closing prices. On a candle with long wicks that closes near its open, the range measure widens and the close measure barely moves, so the two disagree.

Which one should I use on gold?

It depends on the question. If you want to know how much room a position needs, the range based measure is more relevant because it accounts for the full travel of each bar. If you want to know how much the settled price has been drifting, the close based measure is. Many traders keep both and compare them.

Does a channel touch mean price has gone too far?

No, for the same reason it does not with any envelope. The edge is placed at a multiple of typical range from a smoothed price, so exceeding it is what a strong move looks like. In a trend, repeated touches are characteristic rather than extreme, and fading them means trading against the condition that produced them.

Why do my channel and someone else's look different?

Because there are at least four free choices: the type of average on the centre line, the smoothing used for the range term, the lookback for each, and the multiple. Any of those shifts the edges. Before comparing rules or results with anyone, write down all four, because a channel touch is not a portable concept without them.

How does a weekend gap affect the channel?

The true range calculation deliberately includes the distance between Friday's close and the new open, so a large gap enters the average and widens the channel. That is correct behaviour, but it persists for the whole lookback, so for a while afterwards the width partly reflects the gap rather than current conditions. Check whether one sits in your window.

ⓘ See these ideas on real price: open the free XAUUSD live chart.

More from the blog

View all posts →
Join GroupChat