Beta #2

Optuma Desktop Feature Requests
1

Hi,

I am here again and hope I keep you at Optuma not too busy with my posts.

This time I have some thoughts about the tool “Beta”. I made some calculations in Excel and have recognized that Optuma uses the normal percent change calculation to calculate beta and not the LN calculation as it is common in finance.

Here is a chart of the weekly and monthly beta of Apple.

beta
beta1132×810 77.7 KB

These calculations are of course correct!!!

But if I would use the LN to calculate the weekly percent changes I would get for the weekly beta 1,1363 and not 1,1297 what I would get if I would use the normal percent change calculation.

If I would use the LN to calculate the monthly percent changes I would get for the monthly beta 1,4407 and not 1,4550 what I would get if I would use the normal percent change calculation.

These differences are small but notwithstanding great enough to have great effect when you use them to make calculations for portfolio optimizations.

On the other hand it is confusing if you get in Optuma a different value than the one I have calculated in Excel or I get from another source. It would be good to have the same values from all sources and as already mentioned in finance it is common to use the LN to calculate percent changes.

In the tool “Regression Analysis” the beta is also calculated. So it is necessary to consider this tool too when changing the beta calculation.

In this context, I would also like to mention my post in the script forum

https://forum.optuma.com/topic/beta-function/

I have attached an Excel file with all my beta calculations so you can verify my calculations.

Thanks

Thomas

Beta-Calculation-in-Excel.xlsx (48.2 KB)

2

Hi Thomas,

Just wanted to show you the updated chart. I’ve used Training Mode to bring my data back in line with the way it was when you created this chart. This will be in Optuma 1.6

All the best

Mathew

beta

3

Just thought I would add that I took the opportunity to make sure that the Regression Chart was updated to use a LN script instead of Change. I left it as scripting so there is more control over the dependant and independant variables.

AAPL

4

Hi Mathew,

thanks for updating all correlation and regression calculations using the option to make the calculations with the LN function (natural logarithm) e.g. LN(Close/Close[1]) instead of simple price %-change calculation e.g. (Close/Close[1])*100.

The LN calculation, also called Log Return, is the preferred return calculation in finance. The simple price %-change calculation is called Simple or Normal Return.

As far as I am able to verify the numbers of the many charts/calculations you have made all seems to be correct.

Bringing the correlation and regression calculations all in line using the option to make the calculations with the LN function removes all the confusion which so far have existed.

For those interested in this topic the new updated calculations also removes all the confusion discussed in the forum post “Regression R-Squared” (https://forum.optuma.com/topic/regression-r-squared/).

Thanks again Mathew for your great work.

Thomas