Most systems try to learn what winning looks like. LORENTZIAN spends more of its memory on what losing looked like — here's…