In order to actually understand true risk (to create a profitable model), you’ll actually have to experiment and lose money in order to bootstrap your own ML model. Taking data acquired elsewhere and hoping it can make your own model instantly profitable isn’t possible.