The description for the RSquared indicator in WL8 says "Measures the strength of a trend, and is a component of the standard deviation calculation." That should read "Measures how well the given time series follows the simple linear regression indicator, LR and LRSlope" instead.
The RSquared computation is a merit comparison of "the model" (expected) against "the raw" (observed) data. In this case, "the model" is a simple linear regression, although it can be any empirical representation. It has nothing to do with the second moment (i.e. the variance) or the square root of the second moment (i.e. the standard deviation).
You have documented RSquared correctly here: http://www2.wealth-lab.com/WL5Wiki/RSquared.ashx Just copy that WL6 documentation into WL8. Also, remove the reference to the standard deviation as a parameter for computing the RSquared; it should not be there.
The RSquared computation is a merit comparison of "the model" (expected) against "the raw" (observed) data. In this case, "the model" is a simple linear regression, although it can be any empirical representation. It has nothing to do with the second moment (i.e. the variance) or the square root of the second moment (i.e. the standard deviation).
You have documented RSquared correctly here: http://www2.wealth-lab.com/WL5Wiki/RSquared.ashx Just copy that WL6 documentation into WL8. Also, remove the reference to the standard deviation as a parameter for computing the RSquared; it should not be there.
Rename
Wilco.
Thanks, we will get this done for B50.
The Standard Deviation is irrelevant to the RSquared calculation. The Standard Deviation parameter should be removed. I'm running WL8 Build 54.
Linear regression can take on many forms and dimensions. It might be more descriptive to call it "simple" (or "1st degree") linear regression in the description here since this special case is fitting only one term. You could even make reference to the LR(...) indicator in the description for this single-term case.
Linear regression can take on many forms and dimensions. It might be more descriptive to call it "simple" (or "1st degree") linear regression in the description here since this special case is fitting only one term. You could even make reference to the LR(...) indicator in the description for this single-term case.
The reason it's there is because the calculation uses an approximation that relies on standard deviation, which calculates a bit faster. At this point I think it's ok to just use the full RSquared calculation. Removing the parameter will be a breaking change though but I think it's justified.
Your Response
Post
Edit Post
Login is required