Why Betting Markets and Polls Don't Agree

//

According to election polls, the presidential race is neck and neck. According to election polls, our country is split right down the middle. According to election polls, it's anyone's guess who will be the next president of the United States.

But if you follow betting markets like Intrade and Betfair, Obama's win is a better than two in three chance of winning. Both of these websites allow users to bet money on everything from sporting events to Academy Award winners and everything in between, including the presidential election. Players who guess correctly about the outcome of some future event win money. Because a lot of people play and because money is involved, the predictions tend to be accurate.

DNews Nugget: Predict the 2012 Elections

Intrade, which has been around since 2001, began allowing investors to bet on political outcomes in 2004. It gained notoriety when in 2008, it not only predicted that President Obama would win the election, it predicted that he'd receive 364 electoral college votes. He received 365. (To win, a candidate needs 270, by the way.)

DNEWS VIDEO: The Future of Voting

This year, Ireland-based Intrade (which accepts bets from Americans) shows that Obama has about a 67 percent chance of winning over challenger Mitt Romney. Britain-based Betfair (which doesn’t accept Americans' money) thinks Obama has a 77 percent chance of winning.

So why are the betting markets so different from the polls?

For starters, polling isn't perfect: people change their minds or the sample might not reflect the population as well as you'd like.

That's why polls have a margin of error of typically 3 to 4 percent. So a

candidate who is ahead in the polls isn't a lock.

A candidate could poll very badly in some states, and do well in

the states that add up to 270, even if he is behind in the national

polls. And

if the polling is good in crucial states, then the probability of

winning can

be quite high, even if the national polls are tied. Obama, for example,

polls extremely poorly in Utah and Idaho and Romney doesn't crack 40

percent in Massachusetts or Rhode Island. None of those states matters

much as a result — they are not part of the electoral college math.

Randomly Selected Leaders May Make Politics More Efficient

But polling is all about asking people who they'll be voting for. Prediction markets like Intrade and Betfair ask people to put money on who they think will win. This is where the wisdom of the crowd comes in. Because now you're asking thousands of people to put money on the answer to a question: Who will be the next President? And the bettors aren't betting because they want to lose. And the majority of them aren't betting from their hearts.

Not everyone's perfect though. Nate

Silver at FiveThirtyEight points out that in everything from sporting events to presidential elections, people like to bet on the underdog. Silver's own analysis of the presidential election, incidentally, doesn't just measure the popular vote.

He also takes into account the way a candidate will do in various states. Given that in the United States we have the Electoral College, that makes a huge difference in the calculated odds. Right now, Silver has Obama's probability of winning the election at 92 percent, even though he is only projecting a popular vote margin of 2.7 percent for the incumbent.

Candidates with minorities of the popular vote have won before. George W.

Bush beat Al Gore in 2000, Samuel Tilden lost to Rutherford B. Hayes in

1876 and Grover Cleveland lost 

to Benjamin Harrison in 1888. Bill Clinton won with only a plurality of the vote in 1992. That doesn't look as likely this time around, though SIlver has Romney with a 4.9 percent chance of winning the popular vote and losing the electoral college. (The chances of this happening to Obama are smaller because of the configuration of states where he polls well). 

PHOTOS: Shockingly Close Presidential Elections

This is why in the United States, the news coverage focuses a lot on

battleground states.The two biggest states up for grabs are Ohio and Virginia,

and the former has shown small, but consistent leads for Obama. If Obama

carries Ohio, then Romney would have to win at least four of the six remaining

battleground states: Colorado, Iowa, Nevada, New Hampshire, Virginia and

Wisconsin. That's not impossible, but it would be difficult to do — Wisconsin

hasn't gone for the GOP since 1984, leaving Romney with even fewer winning

combinations.

Obama, meanwhile, could win without Ohio. There are simply more plausible combinations that work for him. All this is to say it isn't just about the headline

percentage in the polls. But while the probabilities are with Obama, that doesn't mean it's a sure thing.

Credit: John Moore/Getty Images

DISCOVERYnewsletter
 
Invalid Email