Linear regression
is used to predict future values from past values using statistics
 often showing when securities are overpriced. Using the least
squares method, a straight line can be drawn thru a trend having
the minimum distance from actual prices. As this resulting trendline
is in the exact middle of the prices, any move above or below suggests
the security is either overbought or oversold.
An extension to this
is 'Linear Regression Channels'  rather like Bollinger Bands, these
are two parallel lines above and below the trendline at a fixed
distance (the maximum distance in the timeframe under examination).
The bottom channel line tends to provide support and the top channel
line provides resistance, with any extended move outside the channel
suggesting a reversal in the trend. Probably more useful to swing
traders than those traders day trading, linear regression lines
nonetheless may have their place.The calculation for the linear
regression trendline is outside the scope of this article.
previous
next
