Nate Silver @NateSilver538
@tangotiger @IneffectiveMath @MoneyPuckdotcom @domluszczyszyn @36Hobbit @mchernos @matt_barlowe I'm not sure that 2 works because there's a point at which if a model is *extremely* different from the betting line, you might become *less* inclined to bet it because the chance goes up that (i) the betting line accounts for some factor you're missing or (ii) your model is bad. — PolitiTweet.org