Interface RollingRegression

One bar's rolling ordinary-least-squares fit — the three arrays linearRegressionValues returns, row-aligned with its input.

NaN marks a bar the fit does not cover ([PND-STUDYBOX]).

interface RollingRegression {
    intercept: Float64Array;
    r2: Float64Array;
    slope: Float64Array;
}

Properties

intercept: Float64Array

The fitted line at the window's first bar (x = 0) — TA-Lib's LINEARREG_INTERCEPT, not the value at the window's last bar.

r2: Float64Array

Coefficient of determination, 0 … 1 — the fraction of the window's variance the line explains. A flat window is NaN, not 0.

slope: Float64Array

Change in the fitted line per bar — the units are the column's own per bar, so it scales with the price and is not a percentage.