. . .more on Formulas 

The Standard Deviation is computes as follows:
- Sum the squares of the differences between the raw prices, and the linear regression line.
- Divide this sum by number of bars in the regression series data range.
- Take the square root of the result, giving the standard deviation.

See Also...
   ...Linear Regression Forecast / Bands
   ...Linear Regression Slope
   ...Linear Regression Acceleration

. . . more on Charts 

Above is a Daily Candlestick Chart of an DKWD (D&K Healthcare Resources).  The blue center line represents the regression line, based on the preferences selected below.  The red lines represent channels based on the standard deviations specified in the preferences below.

. . . more on Preferences 

• Price - Price data to be used as input in calculating the regression line.

• Pre-Smoothing Type - Smoothing type used to pre-smooth data before performing the regression.

• Pre-Smoothing Period - Smoothing Period used to pre-smooth data before performing the regression.

• Data Range - This group box contains the options used to specify where the regression analysis will begin and end.

• Automated using the last X bars - This option results in an automated moving regression line, always using the most recent X bars.

• From dd/mm/yy hh:mm:ss to dd/mm/yy hh:mm:ss - This option allows the user to specify a specific start and end point in time.

• From dd/mm/yy hh:mm:ss to Present - This option results is an automated moving regression line, always beginning at a specific fixed starting bar, and always ending at the most recent bar.  The data range will expand as time moves forward.

• Standard Deviation(s) Above - Number of standard deviations above the regression line to draw a parallel channel line.   (0 for no line)  You may specify multiple channel levels here, simply by comma separating the values (eg. 1, 1.5, 2).

• Standard Deviation(s) Below - Number of standard deviations below the regression line to draw a parallel channel line.  (0 for no line)  You may specify multiple channel levels here, simply by comma separating the values (eg. 1, 1.5, 2).

• Raff Channels - Developed by Gilbert Raff, this method finds the maximum distance between any closing price and the regression line. This distance is then used as the basis for the channels. The channels are drawn parallel to the regression line, at a distance above and below the line equal to the maximum distance computed. The upper channel is then used for support, while the lower channel is used for resistance.

• Extend Line to Present - Check this box if you would like your regression line (and channel lines if applicable) extended to the right of the chart.

• Regression Color - Color and style of the regression line in the chart.

• Channel Color - Color and style of the channel line(s) in the chart.

Several options exist for selecting the range of prices that will be included in the regression analysis (Data Range).  The first option "Automated using the last X bars", results in an automated moving regression line, always using the most recent X bars.  The second option, "From dd/mm/yy hh:mm:ss to dd/mm/yy hh:mm:ss", allows the user to specify a specific start and end point in time.  The third option, "From dd/mm/yy hh:mm:ss to Present", is another automated moving regression line, always beginning at a specific fixed starting bar, and always ending at the most recent bar.  The data range will expand as time moves forward.

Click the linear regression icon in the charting toolbar to enable the linear regression drawing tool.  Mouse down in the window at the point you'd like to begin your regression line.  Hold your mouse button down while you drag the drawing tool between the two points of interest. When you release the mouse button Investor/RT draws the regression line. As an option, you may choose to draw straight band lines parallel to the regression line. The bands are drawn a user-specified distance above and below the regression line. The distance is specified by the user as the number of standard deviations away from the regression line.  You are not limited to drawing only one channel line above and below the linear regression line. You may specify multiple levels simply by separating them with a comma. For instance, you may have:

Standard Deviation(s) Above: 1, 1.5, 2
Standard Deviation(s) Below: 1, 1.5, 2

And three channels will be drawn, at 1, 1.5, and 2 standard deviations from the regression line respectively. The standard deviation value is computed using the same range of data values used in determining the regression line (Data Range).  The values specified in the preferences are multipliers of this standard deviation value, used to compute the distance(s) of the channels from the regression line.

If the Raff Channels checkbox is checked a different method will be used to compute the bands.  Developed by Gilbert Raff, this method finds the maximum distance between any closing price and the regression line. This distance is then used as the basis for the channels. The channels are drawn parallel to the regression line, at a distance above and below the line equal to the maximum distance computed. The upper channel is then used for support, while the lower channel is used for resistance.  Again, you can specify multiple levels, separated by commas.  The levels will be used as multipliers of the standard Raff channels.  To simply draw the standard Raff channels, specify multipliers of 1 for both the above and below channels. 

The Linear Regression Study can optionally be extended to the right with the "Extend Line to Present" option.  This will project the line forward to the right edge of the chart.  The begin and ending points of the trendline may still be clearly seen as small dots in the line.  These points can be dragged to a new position in the price data.  When a drag occurs, the regression is recomputed and a new regression line is drawn considering the new data range.  When the "From dd/mm/yy hh:mm:ss to Present" data range option is being used, then only the begin point can be dragged to a new fixed location.  The begin point will then remain fixed while the end point will adjust to the most current bar as new bars form.

The following approximations offer a few rules of thumb for using the standard deviation settings:

• Plus or minus one standard deviation takes in 68.3% of all expected outcomes (historical price moves)
• Plus or minus two standard deviations takes in 95.4% of all expected outcomes (historical price moves)
• Plus or minus three standard deviations takes in 99.7% of all expected outcomes (historical price moves)

For example, excursions of price more than 2 standard deviations above or below the regression line represent an unlikely event (less than 5% probable). Such excursions usually represent overbought or oversold conditions.