Looking through his history, his predictions are kind of all over the place. He missed Colorado by 18 points, Minnesota by 28 points, Vermont by 30 points, Massachusetts by 14 points, Kansas by 20 points, Louisiana by 15 points, and while he was one of the only people to accurately predict Michigan, he still missed the margin by about 5 points.
Going through his whole blog, it seems more like he got lucky than anything else. I mean, maybe it could be something if he used it to weigh a model that also incorporated polling data.