The way we pick presidential nominees is relatively stupid. A random handful of states entertain candidates for months on end, weeding out a couple who were likely to be weeded out anyway. Then everyone else rushes to weigh in, and it’s a mess for a month or two until it gets sorted out. It’s a system rife with waste and built on a fairly rickety foundation, as may be proven this summer in Cleveland at the Republican National Convention. But it’s what we’ve got.

Since the modern primary system emerged, there haven’t been that many presidential contests. Recognizing that small sample size, though, there are a lot of states that have spent a lot of time voting for candidates who didn’t end up winning — and a few that almost always hit the mark.

The goofy primary system means that states that vote earlier are more likely to vote for someone who doesn’t get the nomination. After all, they have more candidates to choose from, and by the last few contests, no one is putting up much opposition.

On average, states vote for the eventual nominee 71.3 percent of the time — just less than three-quarters. That’s a bit unfair for another reason. Usually (but not always) states will also vote for a candidate who is from their state, despite the odds. If we exclude those home-state votes, states get it right 73.4 percent of the time.

But again: There’s a wide range. The six states that are worst at picking are mostly ones that vote early, as above.

51. Alaska, 38.5 percent (excluding home-state candidates)
50. Iowa, 50
49. Michigan, 53.3
48 (tie). Arizona, 53.8
48 (tie). Massachusetts, 53.8
48 (tie). Nevada, 53.8

(It goes to 51 because we included D.C.)

Alaska gets it right slightly more than a third of the time. Contrast that with the top tier.

1. Wisconsin, 93.8 percent (excluding home-state candidates)
2. Illinois, 93.3
3 (tie). Idaho, 92.3
3 (tie). Missouri, 92.3
5. Virginia, 91.7

Wisconsin has the best track record. In 1984, it picked Gary Hart as the Democratic nominee. In every other contest, it’s been spot-on.

The complete tabulation is below. Remember, this should not be seen as a guide to predicting this year’s outcome, because we simply don't have that much data at-hand. And also because this year promises to stay messy for a while, because our nomination process is a sloppy mess.