Hi! I am trying to forecast some consumption data using exogenous price feature (spot). Twist: I am using leaded price features since I known power prices a day ahead. For the training it is trivial. I just add the exogenous features to the training DataFrame and set static_features=[] for the model. Let's say training ends October 15th. Note: I am using a model per forecasting step. (see viz 1) Now for the forecasting part - I want to use my model to forecast starting from November 1st. To my understanding this should be done this way: 1) Update the model with consumption between October 15th and November 1st 2) Use mlforecast to make future frame based on forecasting horizon length 3) Since having non static exogenous features these should then be added to the future frame prior to forecasting Well, I cannot just take the future frame and simply extend with the exogenous features. This will result in missing data. So the question is: "Can I understand the future frame format as each row 'being fed' to each of the model. Hence, I would need to replicate the spot prices, since every model at the forecasting point only have the same access to future spot prices" (see viz 2)