Linear Regression...

[uphill]

Hi Alex,

Thanks for all the great help in the past! I have another question: The default Regression Trend Channel tool has a source setting of “H-L Flip.” The Linear Regression Study under Basic Studies does not have this option. Could you please provide me with information on how I can include this option in the Basic Studies Linear Regression area and also how I can “call it” from within an efs program. Or, please direct me to an efs that may already have this and all of the options included.


Thank you,

JR

[uphill]

Linear Regression...

Alex,

Also, if you could just provide me with the definition of “H-L/Flip” then I could go from there. I assume the “H” is for “High” and the “L” is for “Low” but I do not know what the “Flip” means. Also, please verify that the other options have the somewhat obvious mathematical average meanings. For instance: (O+H+L+C)/4 is just the moving average of the (Open + High + Low + Close)/4. Correct?

Thank you,
JR

[ACM]

JR
For the definition of H-L Flip and of (O+H+L+C)/4 right click on the Regression Trend Channel icon in the Line Toolbar and select Help in the window that will open.
The H-L Flip is not an available option in the Basic Studies Linear Regression or in the Builtin efs of the same.
I do not know if there is an efs that includes that function.
Alex

[uphill]

Linear Regression...

Thanks Alex!

BTW: Standard Dev. typically does not consider trend, yet the s.d. on the reg. channel tool seems to take that into account. e.g., if you were to use the tool to do a linear reg between two candles right next to each other but with different closes, the s.d. appears to be zero. Whereas, in pure mathematical terms, the s. d. would not be zero. So, is there a variation of the s.d. in this tool that you could share? Better yet, is there a chance that I could just have the efs code for the linear regression tool?

Thanks again,
JR

[gspiker]

JR,

The built-in indicators and tools are not written in EFS, they are done in C or C++.

I doubt eSignal will share their code with you, esp not one that might have some propietary components. AFAIK Alex doesn't work for eSignal, so he couldn't help you there anyway.

There are a number of LR codes out there written in EFS, of you do a search you should be able to find them.

G

[uphill]

Thanks G. eSignal seems to be very open and encouraging with development. I would like to see the code even in C++ to assist with my own efs writing...

I appreciate the feedback.

JR

[Guest]

JR,

if you were to use the tool to do a linear reg between two candles right next to each other but with different closes, the s.d. appears to be zero. Whereas, in pure mathematical terms, the s. d. would not be zero.

It depends how it is calculated.

Standard deviation from the regession line is always zero with two points in the regression calculation. This is because the regression line passes through both data points.

Standard deviation from the average will be some number unless the line is horizontal. In the case of a horizontal line, the standard deviation from the average would also be zero.

There are a number of LR codes out there written in EFS, of you do a search you should be able to find them.

As Garth indicated, there are many pages of linear regression posts out there. When I did a search using the term linear regression, I had 8 pages of links come up. I know there are several examples of code that performs the base regression calculations similar to eSignal, and they are already written in EFS javascript.

I am unaware of a version posted which includes H-L Flip functionality.