Standard Error

Click to Enlarge

Description

Standard Error measures how closely prices congregate around a linear regression line. The closer prices are to the linear regression line, the higher the r-squared value and the stronger the trend. 

For example, if each day’s closing price was equal to that day’s regression line value, then the standard error would be zero. The more variance or “noise” around the regression value, the larger the standard error and the less reliable the trend.

Interpretation

High standard error values indicate that the security’s prices are very volatile around the regression line. Changes in the prevailing trend (over the number of time periods specified) are usually preceded by a rapidly increasing standard error.

Standard error can be used effectively in combination with the r-squared indicator. Changes in trend are often signalled by a high downward moving r-squared, a low upward moving standard error, or a low upward moving r-squared and a high downward moving standard error. In other words, when the two are at extreme levels and begin to converge, look for a change in trend.

Note that a change in trend does not necessarily mean that an upward trend will reverse to a downward trend. Sideways movement is also considered a "change".