Excuse me, did OP solve the problem?
I just got the same issue and finding no way out.
No matter what I tried in R, etc(model = 'AAA'), HoltWinters(x, gamma = T)
There is no way like excel prediction is so close to Raw Data.
Btw, according to Microsoft Excel FAQ, forecast.ets in excel is model AAA.
Hope somebody can share own experience about this.