Several very interesting things happened in the political prediction markets on Super Tuesday. As you would expect, most of them had to do with Ohio.
Going into the day, we predicted that Mitt Romney would win seven states to Rick Santorum's two and Newt Gingrich's one. This was off by one: Santorum took North Dakota, which the markets had Romney winning with about 75 percent certainty. This assumes Romney can hold on to his thin lead in Ohio. (Since we report the likelihood of victories, not up-and-down predictions, we expect to miss one or two races in a big batch like this. In fact, if partial victories were possible, we'd anticipate getting 8.7 of 10 races right.)
We like to preach the gospel of political prediction markets as being the older and wiser cousin to polls, reacting more cautiously to flavor-of-the-week candidates and insurgent politicians like Santorum whose campaigns are less organized and wealthy than their competitors'. The markets were extremely confident in Romney winning the Buckeye State, but not so confident that they didn't momentarily abandon that pick as the early returns seems to suggest a narrow Santorum win. At around 10:30 p.m. ET, Romney's share of the market in Ohio had dipped to 20 percent. If you cast your lot with him then, you're probably going to make a nice wad of walking-around money.
With the full benefit of hindsight, one could argue that the markets reacted too quickly to Santorum's early lead. While he did better than expected in Ohio, a county-by-county analysis suggests the picture was always better in the state for Romney than it appeared around 9 p.m. ET. The returns from urban areas and other demographics that tend to support him came in more slowly than those in smaller counties. Curiously, the narrative very closely resembles that of a general election fight between a Republican and a Democrat, in which the conservative can often have an inflated sense of confidence earlier in the night for the same reason. (The New York Times website's graphic for Ohio used blue for Romney and red for Santorum, making it almost visually indistinguishable from a partisan election.)
There are two competing narratives in this primary. On one hand, the delegate math heavily favors Romney, to the point that his hold on the nomination is considered very secure. On the other, Santorum's continued successes, though smaller in share, continue to advance the storyline that Republican voters still cannot bring themselves to just give Romney the nomination already. We saw this play out briefly in the overall odds that Romney will win the nomination, which were in the high 80s yesterday morning. At the bleakest moment in the night for the former Massachusetts governor, those odds dipped about 5 points—a small but significant drop. The math was never in question, but the storyline was. To those driving the markets, it is clear that a win is still a win, regardless of its implications for the byzantine delegate politics that will determine the nominee.