There's one problem with pundits. Alright, there are many. But one of them is the following: When they make prognostications, they seem to always make them with the requisite qualifications so that they can't be proven wrong. "I said Mitt Romney had a 60 percent chance of winning! Which meant he had a 40 percent chance of losing! So I got it right!"

I think we can do better. So let me stick my neck out there and make some specific predictions for the Republican primary.

Iowa: Ted Cruz first, Marco Rubio second, and Donald Trump third.

New Hampshire: Rubio first, someone else second.

Those predictions don't line up with the polls, so here goes.

I think Trump will peter out at the polls. Several writers I respect, including Ezra Klein, Tim Carney, and Ross Douthat, have laid out the case for believing, even six months later, that Trump's support will eventually decline.

First, it's important to realize, with regard to this election, that polls suck. It's a widely reported fact that opinion polls have been increasingly inaccurate. The proliferation of cell phones and the decline of land lines means that people are harder to reach. Those who do participate in polling tend to be older than the average voter. Second, the Trump phenomenon is probably more akin to a howl of anger than a real decision. Countless voters or would-be voters are telling pollsters they support Trump as a way to send a message but they actually do realize he would be a terrible president, and they're not going to pull the lever for him. Third, caucusing is different from voting. Caucusing is a complicated, involved process, and the demographic that Trump draws upon in polls may not turn out for him, especially because Trump doesn't have a strong turnout operation in the state. And once Trump does worse than expected in Iowa, it's going to change the momentum of his campaign. The soufflé is going to come down. Doing badly in Iowa will lead to him doing worse in New Hampshire, and so on.

I also think that Rubio is a perennially underestimated candidate, and my colleague Michael Brendan Dougherty lays out nicely the case for underestimating Rubio. It's convenient to describe him within the framework of what's sometimes known as the "two lanes" theory of primary voting. The idea is that in the Republican primary there are only two types of candidates: the establishment politician and the base politician, and Rubio is the candidate of the establishment. There's definitely some truth to this: Many in the establishment feel that Rubio is their candidate and has the best chance to defeat both Trump and Cruz, and later on Hillary Clinton.

But Rubio is also a Tea Party conservative with a 100 percent rating from the American Conservative Union. And conservatives know him, which is not the case with Jeb Bush or Chris Christie, nor was it the case with Mitt Romney. Rubio might not be the first choice of many conservatives, but he is still appreciated and trusted. Of course, everyone brings up his major heresy, his support for comprehensive immigration reform. But even though the immigration issue is powerful this year, experts tend to overestimate the importance of anti-immigration voters in the GOP primaries. lt was widely predicted that John McCain's support for immigration reform was going to cost him the GOP nomination, and it didn't. Earlier Pat Buchanan's anti-immigration stance was supposed to lead him to victory, and he went down to miserable defeat. All other conservatives — as well as moderates — feel good about Rubio. And the history of Republican primaries shows that the ability to win a general election is important for Republican voters. Rubio is probably the most electable candidate this time.

I think all of these factors together mean that Trump will do much worse than polls suggest, and Rubio will do better. This means a Cruz win in Iowa, because he's been camping there for months and has a great ground operation, but a stronger-than-expected showing by Rubio, which will propel him to a win in New Hampshire, where he's already doing better than Christie, Kasich, or Bush.

We'll soon see if I'm right.