Friday, October 19, 2012

Intrade, Ohio, and the State of the Race

This is meant as a follow up to my previous post where I suggested that Obama's odds on Intrade seemed rich relative to the existing poll numbers (it's worth reading first). Yesterday, Obama was trading at a 64% probability for re-election. Today, he's off a bit, at 61.5%. National polls are a dead heat, and swing states are essentially a toss up as well, the one exception being Ohio, where Obama has a slim but definable lead (up 2.4% in the RCP average).

Looking at things on the surface, Obama's lead in Ohio could arguably justify a slight advantage on Intrade, but 64%? This seems illogical. Interestingly, once again, the Buckeye State is shaping up as the state-of-states. It was once thought Romney couldn't win without it, but now, with Virginia, North Carolina, and Florida solidifying for him, it looks like neither candidate can win without it.

Before we delve further into Ohio, let's just step back and make some strong generalizations about this race:

  1. Democrats, with few exceptions, will vote for Obama
  2. Republicans, with few exceptions, will vote for Romney
  3. Independents will divide their vote, but right now are polling Romney +8.2

This last number is an average of the last dozen national polls, and it's a biggie. To put it in perspective, Obama won indies last time by the same amount - there's been a 16 point swing.

So, here's the game: you tell me who's showing up on election day, and I'll tell you who's going to win. If it's precisely the same electorate as four years ago, Obama wins by 2.4%, versus 7.2% in '08. (The change is entirely due to independents). But we all know that's not going to happen. Republicans were dispirited in '08, and Democrats had found their messiah.

What if the electorate is the same as 2010? In this scenario, Romney wins 51-49. But that was an off-year election, and we all know that's not going to happen either.

So, you tell me. Who's going to show? I would argue that the mood of the electorate is closer to 2010 than 2008, and certainly that's what the "intensity" polls say, but that still makes for a tight race. Really, there's no way to look at this thing and see Obama at 61.5%.

When analyzing polls, it's illuminating to see what party breakdown pollsters use. You see, pollsters don't just take all the answers people give and add them up; the data gets demographically massaged. Most importantly, the pollsters make assumptions about what the makeup of the electorate will be. The "electorate" is defined as the people who actually show up to vote. Here are some recent national breakdowns:

                                  Democrat     Republican     Independent
2010                               36%              36%                 28%
2008                               39                 32                   29
2004                               37                 37                   26

Even when you look at more elections, 2008 was a total outlier.

Which brings us back to Ohio. If some smart money knew something about Ohio that favored Obama, then the big Obama advantage on Intrade becomes plausible. So let's do a bit of a deep dive in Ohio to see what all the clever traders might know.

Obama won Ohio by 5 points. The breakdown of the electorate was 38-33-29, or D+5. In 2004 it was R+4 and in 2010 it was R+1.

So, what's the right mix for this time? Would you say it's D+5.4? No? Well, that's what the pollsters are using. That's the average of the last seven Ohio polls (hat tip to Numbers Muncher). They are actually assuming that Democrat turnout will be stronger than 2008! There is zero evidence to support this, and it deifies common sense.

A more reasonable assumption might be D+2. If I'm right about this, Romney wins 50.14% of the vote (assuming independents cut his way by the national average of 8.1%). That sure sounds like a toss-up to me. which means that the entire election can viewed the same way. This is not my view, mind you. I think Romney going to win by a surprising margin, and I've been on the record as such for months. But a cold, dispassionate look at the numbers says this thing's a toss-up.

Given this, it appears that the Intrade market is off by 10 points or so. Shorting Obama right now is a great trade. Not necessarily because he's a lock to win, but because the market should merge with the polling data soon. When it gets near parity, unwind, unless you just want to be a punter.


  1. The state to watch is Iowa. If he cannot win in lily white Iowa, you can turn out the lights. It will be close but no cigar. If O'Bomney leads in Iowa in the closing days, the other important swing states will fall like dominos. Most of the "swing" states are not swing states. The ones that matter are only Iowa, Nevada, Colorado, Ohio, Virginia and Florida.

  2. I've pulled up the interactive electoral map. On Romney's best night, it's Obama 271-267.

    Unless the "Great Awakening" that Republicans hope takes place on election day happens, it's game over. That's why Intrade has not moved.

  3. Correction: I forgot to click Ohio in the best case which would give a 285-253 win. So if the beautiful hallucination brought on by the Kool-Aid bender Scott has been on for over a year occurs, it is within the Dumber and Dumber "So you mean there's a chance?" scenario.

    This would require winning Florida, Virginia and Ohio which is still unlikely but could happen even without the "Great Awakening" occurring. InTrade has it about right.

    Don't put your money down on it. I recommend setting it on fire instead. At least you get the pleasure of watching it burn. It ain't easy to catch Romney fever.