Since model has to ultimately run on weekly data, would it make sense to also include a set of weekly data in validation set?
It does not run on “weekly data” in the sense that it only lasts one week. It is monthly data still, just overlapping with a new round starting each week (but each round still lasts 4 weeks).
Agrew that we should reduce the data in tournament data set. 2.5G data is too big for newbie.