Nicolas wrote:

Edurecio wrote:

I am looking for an average of valid values.

“of valid values” from the past? Then we could also designated that as “Optimization” Don’t take it wrong, but valid values would have been the ones that you could have been robustly validated in a OOS period.

So the way you do is not bad, but it is the same as simply using optimized variables, if you don’t make a robustness test, you will never know that they were the good ones or the good way to calculate/define them.

I don’t take it wrong, I try to learn.

“… but valid values would have been the ones that you could have been robustly validated in a OOS period” We have the possibility to validate in a OOS Period every month (I’m in no hurry), comparing the data obtained with the previous accumulated average with the data of a theoretical optimization for only that month. Also see how much convergence deviates from the mean with each new monthly calculation as a robustness sample.

I do not consider myself with “the truth”, I know very little compared to many of you (Nicolas, you are really an expert and I am a rookie !!) but I think that way exists and can be combined with other forms of evaluation of variables

Greetings.