Results for prognostication #4: The Florida primary

It was a good night to do things simply. My methods predicted the final numbers more accurately than did Nate Silver's.  There were no surprises, as the basic order of Romney winning big over Gingrich, with Santorum and Paul trailing was what all polls said in the last few days.  The nearest thing to drama arose from polls that said the race between Santorum and Paul be close. It wasn't.  Both Silver and I overstated what Ron Paul would get, which brought our overall grade down. If this was a strict grading policy, we'd both get A-, since I was 91.0% accurate he was at 90.0%.  

Here are the results, candidate by candidate.

 Romney 46.4%(Both of us low, but I beat Silver by 0.5%)

 Gingrich 31.9%(Again, both low. I beat Silver by 1.7%)

Santorum 13.4%(I gain 0.1%, he was high and I was low)

Paul 7.0%(I lose 0.3% to Silver, both of us too high)

Other 1.3%(I lose 1.0% to Silver, he was high and I was low)

This battle was close due to the Other prediction.  My numbers just happened to dole out all the votes to the four candidates, while Silver's method said Other would get 1.6% of the vote. Reality was 1.3% for the stragglers, so Silver was 1.0% closer.

In the prediction contest, Silver and I are tied 2-2. I eked a win in Iowa and won last night. Silver's numbers were much better in South Carolina and he won in New Hampshire because I got silly and caught Buddy Roemer fever based on a single poll.

There are four caucuses in the next week, Nevada and Maine this Saturday, Minnesota and Colorado next Tuesday. As of right now, fresh data is scarce in all these places, but I expect that to change in the next few days. I reserve the right to make no prediction without at least one poll that is taken within a week of the actual vote. Barring that, I will be ready to toe the line twice early Saturday morning and the same on Tuesday.  After that, February has only two primaries, Arizona and Michigan, both on February 28, so the crystal ball will have a few weeks off.

While I'm interested in politics, this is really an excuse to fiddle with numbers, and longtime readers will know how much I enjoy that.  So gimme some data and let's get back in the ring!