I am forecasting hourly power prices by applying Gradient Boosting to, among other variables, average monthly gas prices. The algorithm is currently trained on historical nominal monthly average gas prices beginning in 2012 and will ultimately applied to the Henry Hub futures curve (with the time value component stripped out). My question is this: What, if any, inflation adjustments do I need to make to my underlying data?
Do futures prices inherently contain inflation information? Or do I need to tack that on as part of my estimate? Do I need to use real prices to calibrate my regressor?