Oisigma/Blog
Research & notes

Donchian Channels vs an Expected-Range Band

Donchian Channels and an expected-range band both draw an envelope around price, but they answer different questions. A Donchian Channel reports the highest high and lowest low of the last N bars. An expected-range band reports how far the next close has historically tended to land from the current level, with a stated coverage rate. Only the second makes a claim that can be scored.

What does each one measure?

Indicator Measures Output Coverage claim
Donchian Channel (20) Highest high and lowest low over 20 bars, price space Upper line, lower line, midline None attached
Bollinger Bands® (20, 2) Standard deviation of closes around a 20-bar average, price space Three lines ±2σ, nominally ~95%
Expected-range band (BTM) Standard deviation of recent returns (~60 bars), return space Inner and outer band Inner and outer rates published, historically

A Donchian Channel is, by construction, a record of the extremes already seen in the window. Price sits inside it until it makes a new N-bar high or low, so touching the channel edge is the definition of a new extreme rather than a statistical event.

Which one can be scored for accuracy?

Only an indicator that attaches a stated rate to a range can be scored on coverage. A Donchian Channel attaches none, so there is nothing to test it against, and Oisigma has not published any Donchian test. Nothing here says how a Donchian Channel would score. Bollinger Bands® carry a nominal rate, and that has been measured: the Bollinger accuracy audit covers the (20, 2) result. The expected-range band's own rates are on the Proof page, as historical calibration; past behavior is not a guarantee of future results. The scoring procedure is explained in how to compare band indicators.

Why do they react differently?

A Donchian Channel changes only when a new extreme is set or an old one leaves the window, so it moves in steps. A standard-deviation band changes every bar as the window's dispersion changes. The lookback period article covers how the window length affects a volatility band; the same trade-off between reactive and steady applies to the Donchian N.

What does this comparison not show?

It does not show that either tool helps a trader. The paper validates the range's calibration, not the profitability of any use, and whether a use delivers value after costs is an open question. The expected-range band has stated limits too: it runs too narrow in the first days of a fast volatility jump, and the outer band is slightly optimistic in the deep tails. Donchian Channels serve purposes outside this comparison, such as marking the visible range of a lookback.

Frequently asked questions

What is the default Donchian length on TradingView?

The built-in uses 20 bars. The length is an input; changing it moves the extremes the channel records but not what the channel is.

‍

Do Donchian Channels use highs and lows or closes?

Highs and lows. The expected-range band uses closes only; why closing prices matter explains the difference that makes.

‍

Can both be shown on one chart?

Yes. No published test examines any combination, so this post makes no claim about one.

‍

The method is in the working paper. The Behavioral Transform Model runs on TradingView with a 30-day free trial: start free, cancel anytime.

‍

Now, your charts

Curious how this looks on your charts?

Try it free for 30 days and see the range update as new bars print, on whatever symbols and timeframes you actually trade.

Start your free trial
Complete checkout
Read the paper

30 days free, then $15/mo. Cancel anytime from your account.

Pick up where you left off.