more of a question , rather than an issue

[cristianrep]

Thanks for the package, it looks very powerful and promising.

I just started reading through the articles , examples and I am a litle confused about how exactly the multistep is achieved in direct method.
My understanding from "A review and comparison of strategies for multi-step ahead time series
forecasting based on the NN5 forecasting competition" cited is that every single day horizon (for direct method) ends up being a different model.

But i see examples where horizon is defined as bellow:

horizons <- c(1, 3, 6, 12) # 4 models that forecast 1, 1:3, 1:6, and 1:12 time steps ahead.

stating it needs only 4 models , and last one somehow takes care of delivering a 12 days horizon (shouldn't this one only, need 12 models?)

I am sure I miss something, but I don't see it now.

Thx
rep