Political Models Update

As Naked Dollar readers know, we have two proprietary political models that we follow. From here out we will update both each time. Interestingly, the models are saying two very different things right now. More on that in a moment.

Our first model, the Naked Dollar National Pulse Index, follows all polls for House and Senate seats as well as the presidency. It aggregates poll-over-poll changes, counting them for either the Republicans or Democrats, depending on the direction. The idea is to look for short and medium term trends in the national political news. (For a more complete understanding, click here.)

Here is the update:

An upward trend means Republicans are gaining ground, and vice verse. Since the presidential polls have changed very little, what you see above is a major trend favoring all the Republicans running for House and Senate seats. This is a story that not a single news outlet has reported. It means the gound is shifting under the presidential race, only people haven't realized it yet. (One caveat: this does not reflect any bounce for Democrats in the wake of their convention.)

Here's what the numbers look like on a non-cumulative basis:

Lines pointing up are good for Republicans, pointing down good for Democrats. You can observe longer and deeper upswings for Republicans.

Our other model is The Naked Dollar Electoral Model. This one focuses entirely on the presidential race and the only thing that really matters, the electoral college. This is the third presidential election cycle in which we have deployed this model, and it has been very accurate. The brief explanation is that we take the odds of each candidate winning a given state, as observed in the political futures markets on intrade.com, and we multiply the odds times the electoral votes available in that state, awarding a proportional amount to each candidate. (For a more complete understanding of this, click here.)

Here's what the electoral model looks like right now:

It has Obama winning a very close election, 288 votes to 250. The last election was 365 to 173.

As you know, I actually think Romney is going to win with surprising ease. I've made this case for a number of months, and Pulse Index is part of my reasoning. But I also believe in the "wisdom of crowds" approach inherent in the electoral model, all of which puts me in a position of some cognitive dissonance. These models, not to mention my own reasoning, are saying two different things.

So, what to believe? I will stick by my prediction of an easy Romney win. The prediction markets are useful, but the further away from an election they are, the less accurate they are. Sometimes, they get it wrong in August or September, but then nail it in November. This was the case in 2004, for instance:

The last reading taken in 2004 got the electoral count exactly right, but you can also see there were times when the model had Kerry winning.

What's my case for why the markets are mispriced right now? First, I don't think anyone is noticing the incredible trending that's going on in all the other races, a trend that will absolutely have an effect on the presidential race. Second, the presidential polling remains, essentially, a tie. However, the samples being used by most pollsters resemble the electorate in 2008, and there's no way that's going to look like the electorate in 2012. It's possible this is the sort of poll-geek minutia that participants in the Intrade markets don't analyze.

I predict that Intrade will start reflecting these realities over the next 2-3 weeks. Better yet, I'd bet on it.