Glossary term
Linear Regression Channel
A statistical trend channel constructed by regressing price on time, then plotting upper and lower parallel bands at one or two standard deviations above and below the regression line. Used for mean-reversion entries inside a trend. Closelook's Cointegration Monitor leans on LRC deviations to timestamp pair-trading entries.
Definition & Context
A Linear Regression Channel fits a least-squares line through a window of closing prices and projects two parallel bands at ±1 or ±2 standard deviations. The centreline captures the local trend; the outer bands capture statistically extreme deviations from that trend. Unlike Bollinger Bands (which use a simple moving average plus volatility), LRC respects the slope of the trend — so in a strong uptrend, both the mean and the bands slope upward together.
Practical use: traders buy near the lower band and sell near the upper band while the slope remains intact. A breakout through the outer band on elevated volume signals either an acceleration (trend-following entry) or regime change (mean-reversion trap). Because LRC is path-dependent — add one more bar and the entire regression shifts — it is best used on rolling windows (typically 50, 100, 200 bars) rather than as a fixed reference line.
Why It Matters for Investors
LRC is how Closelook’s Cointegration Monitor timestamps pair entries. Once two assets are identified as cointegrated, the spread between them is expressed as deviations from its own linear regression. Entries fire when the spread breaks the ±2σ band; exits typically at the mean or opposing band. The Pattern Scanner uses LRC on single-ticker price as channel-reversion and channel-breakout patterns from the Directional Alpha library.
Related Concepts
LRC pairs naturally with Cointegration analysis, uses Z-Score to express deviations, and interacts with Trend Following when bands break.