Donchian Channels (DONCHIAN)

Summary

Donchian Channels: three overlap lines built from rolling price extrema. The upper band is the highest high and the lower band the lowest low over the period; the middle line is their midpoint. Richard Donchian's original four-week rule — generally credited as the first published systematic trend-following system — buys a break above the high of the preceding weeks and sells a break below their low.

Formula

Window = the optInTimePeriod bars ending at the current bar

Upper = Highest High of Window Lower = Lowest Low of Window Middle = (Upper + Lower) / 2

Notes

• The window includes the current bar, matching TradingView (ta.highest/ta.lowest), NinjaTrader, ta4j, pandas-ta and every other library that ships Donchian Channels.
• A breakout rule compares the current bar against the previous bar's band — High[t] > Upper[t-1] — which is where the one-bar offset belongs. Reading Upper[t] against High[t] can never signal, because High[t] is inside the window that produced it.
• Upper, Middle and Lower are bit-identical to MAX(high, N), MIDPRICE(N) and MIN(low, N). DONCHIAN computes all three in one pass under the name users look for.
• The middle line is the channel midpoint, not a moving average of price.
• No smoothing or recursion is involved, so there is no unstable period: outputs are exact from the first bar.

Inputs

• inHigh — High price of each bar
• inLow — Low price of each bar

Outputs

• outRealUpperBand — Highest high of the window
• outRealMiddleBand — Midpoint of the upper and lower bands
• outRealLowerBand — Lowest low of the window

Parameters

Parameter	Type	Default	Accepted values	Description
optInTimePeriod	integer	20	2–100000	Number of bars in the extrema window

Properties

Numerical Stability: Start-Independent

✅ Overlap Input i
☐ Independent Y-Axis
☐ Candlestick
☐ Can Output NaN or ±Inf
☐ Identity at Period 1

Implementation

TA-Lib Definition: donchian.c · donchian.yaml

Native	File
C	ta_DONCHIAN.c
Rust	donchian.rs
Java	Core_DONCHIAN.java

TA-Lib is also available for Python, R and more using a wrapper.