Calculations behind FORECAST.ETS function in Excel

I don't know if you've found anything over the last few days, but here's what little I found.

Your description of the expected calculation ("the alpha parameter multiplied by the last available historical value plus 1 minus alpha multiplied by the previous forecast value.") sounds like what Wikipedia calls "Holt linear", which it also describes as the simplest of many different possible exponential smoothing algorithms (https://en.wikipedia.org/wiki/Exponential_smoothing ). If your attempts to mimic Excel's function with a Holt linear algorithm does not replicate Excel's behavior, I can only assume that Excel's function does not simplify to Holt linear.

This page (https://support.microsoft.com/en-us/...5-21d464333f41 ) claims that Excel's function uses an AAA method, with no further explanation. According to this page (http://freerangestats.info/blog/2016/11/27/ets-friends ), this could be Hyndman's taxonomy where a three character string is used to describe an exponential smoothing algorithm where AAA would mean "additive error, additive trend, additive season". But, then, one must know what those elements mean.

Microsoft is not always good at documenting this sort of thing. At this point, it looks like one would need to know enough about exponential smoothing algorithms in general to know what AAA means in terms of exponential smoothing algorithms.

Good luck in your search.