|
Linear
Regression Indicator
Description
The
Linear Regression indicator is based on the trend
of a security's price over a specified time
period. The
trend is determined by calculating a linear
regression trendline using the "least squares
fit" method. The
least squares fit technique fits a trendline to
the data in the chart by minimizing the distance
between the data points and the linear regression
trendline.
Any
point along the Linear Regression indicator is
equal to the ending value of a Linear Regression
trendline. For
example, the ending value of a Linear Regression
trendline that covers 10 days will have the same
value as a 10-day Linear Regression indicator.
This differs slightly from the Time
Series Forecast indicator in
that the TSF adds the slope to the ending value of
the regression line.
This makes the TSF a bit more responsive to
short term price changes.
If you plot the TSF and the Linear
Regression indicator side-by-side, you’ll notice
that the TSF hugs the prices more closely than the
Linear Regression indicator.
Rather
than plotting a straight Linear Regression
trendline, the Linear Regression indicator plots
the ending values of multiple Linear Regression
trendlines.
Interpretation
The
interpretation of a Linear Regression indicator is
similar to a moving average.
However, the Linear Regression indicator
has two advantages over moving averages.
Unlike
a moving average, a Linear Regression indicator
does not exhibit as much "delay."
Since the indicator is "fitting"
a line to the data points rather than averaging
them, the Linear Regression line is more
responsive to price changes.
The
indicator is actually a forecast of the next
periods (tomorrow’s) price plotted today.
The Forecast Oscillator plots the
percentage difference between the forecast price
and the actual price.
Tushar Chande suggests that when prices are
persistently above or below the forecast price,
prices can be expected to snap back to more
realistic levels. In
other words the Linear Regression indicator shows
where prices should be trading on a statistical
basis. Any
excessive deviation from the regression line
should be short-lived. |
