A model is crappy because it struggles with 1 player? That is ridiculous. If I developed a model that could correctly pick which number a roulette wheel would fall on with 100% accuracy for ever number except 5, that would be a flawed model. Any time the roulette wheel would land on 5, it would give me another # and I'd lose. But I'd win every other time. Is that a crappy model? I don't think so, because it would make me a ton of money (and get me banned in every casino I use it). Just because Dom's model struggles in areas does not make it worthless. That is a very, very myopic viewpoint to think so.