Assign a weight to every stat and rank the field by your model — the same build-your-own-edge workflow the sharps use. Crank up what wins at this week’s course, fade what doesn’t (negative weights allowed), and save as many models as you like. Every stat is normalized so the weights are directly comparable.