Standard Error Channel

The linear regression line is a straight line that minimizes the distance between itself and every data point in the series you are working with. Astandard error measures the variance from the linear regression. Subtracting the standard error from the linear regression line yields the bottom of the standard error channel, and adding it to the linear regression value gives you the top of the channel.

The standard error channel is a parallel concept to Bollinger Bands, which use the standard deviation calculation to set boundaries above and below a moving average to capture variance away from the average. Because the moving average is a wavy line, the Bollinger Bands are wavy, too, and also widen or narrow as variability rises or falls. The standard error does the same thing, only with straight lines.

The critical difference is that you don’t need to choose a starting and ending point for Bollinger Bands, because they track a moving average that constantly discards old data and refreshes itself with new data. To construct a useful linear regression channel, however, you have to pick reasonable starting and ending points. It’s still “mathematics” and thus a better way to draw a trendline than using your eye alone, but your choice of starting and ending points is inherently judgmental. Most practitioners chose an obvious lowest low or highest high.