Exponential smoothing functions are returning #NUM! values for certains kinds of data sets


While using NumXL for forecasting using single double or triple exponential smoothing we observed that it is returning error value (#NUM!)

as an example if for previous 24 months the data was equal to 1 then forecast for the 25th month is returned as #NUM! using exponential smoothing. The issue is not just limited to the above example.

However the error is not reported for all calculations and only some of them for the analysis that we are carrying out.

We have checked the optimization option in smoothing.


1 comment
  • Hi Amit,

    What is happening is (1) optimizer for smoothing values does not converge because the data does not have enough variations.

    Regardless of the cause, this is a defect on NumXL side, and a new release (hot-fix) has been just released (1.59.41298.1). This release include a fix for the default values for the optimization values in the exponential smoothing functions.

    See the NumXL 1.59 TUCSON release notes

    Please, download/install the new version (of our website), and give it a spin. Let me know your thoughts.

    Comment actions Permalink

Please sign in to leave a comment.

Didn't find what you were looking for?

New post