Election prediction? Don't just say it, bet on it!

By Brett Arends for MarketWatch

A friend of mine bumped into two acquaintances earlier this week. They are die-hard Republicans. They are serenely confident. “Mitt Romney has this in the bag,” they said. Other friends of mine, also Republicans, are in a similar zone.

Meanwhile, oddly enough, my Democratic friends are also confident. “Obama’s going to win,” they keep telling me. Liberals tend to be a pessimistic bunch, but not this time. Apparently they are sleeping easily.

It’s surreal. Normally, with an election this close, everyone’s on edge. This time everyone has already ordered champagne. Half of the country is going to wake up next Wednesday is a state of total shock.

If you think the amateurs are confused, look at the professionals. The election nerds. The numbers guys. For them, it’s spreadsheets at dawn. But they shouldn't just offer an opinion. They should offer a bet.

Models vs. models

First Nate Silver, the baseball numbers guy turned election numbers guy at the New York Times, launched a scathing attack on the methodology and reliability of the polling methods used by Gallup.

The back story was that Gallup has pretty consistently shown Mitt Romney about even or slightly ahead in recent weeks, while Silver’s prediction “model” gives Barack Obama about a 77% chance of winning.

After Silver’s attack, the conservative National Review responded in kind, calling Silver’s own model flawed . MSNBC’s Joe Scarborough joined in, calling Silver something of a partisan joke, and saying in reality Obama had no better than a “50.1%” chance of winning.

Enter no less a beast than Paul Krugman. The Princeton professor and Nobel Laureate suggested the attacks on Silver’s model were part of an ideological attack on reason , and an ominous foretaste of the future.

To all of these people, I have four simple words: Show me the money!

If only pundits were required by law to bet on their predictions. Voters, as well as investors, would be spared an enormous amount of misleading blather. There is a way to do this, too. I have a sporting interest.

The bet is in Iowa

By now most people have heard of InTrade, a betting exchange run from Dublin which offers a futures market on election outcomes. (Find my book about this.)

How does it work? Take a look at the Mitt Romney victory contract. If Romney wins, it pays out a dollar. If he loses, it pays out nothing. How much would you pay for such a contract? If you think Romney has a 50% chance of winning, logically you’d pay up to about 50 cents for it. If you think Romney has a 30% chance, you’d pay no more than 30 cents. And so on. People buy and sell these contracts all day, and it gives a rough approximation to where "the market" thinks the right chances are.

As I write this, betting at InTrade gives Obama a 63% chance, Romney a 37% chance. The numbers have been fairly consistent.

Alas, you cannot bet on InTrade if you live in the United States because our federal government forbids it in the name of "freedom." There is, however, one exception.

Since 1988 the University of Iowa's Tippie College of Business has had a waiver from the federal government to run a small election futures market, the Iowa Electronic Market, which is open to U.S. citizens. It runs the market as a nonprofit experiment on the alleged wisdom of crowds. You're limited to a $500 account, so it isn’t going to make you rich. But you can have some fun. I've had an account for a few months.

Today in the Iowa market, the only place in America where you are allowed to bet on the election, Obama is given a 62% chance of winning, Romney 37%.

The efficient market hypothesis crowd will therefore tell you this is the "correct" price, the same way, I guess, Groupon’s (GRPN) $20 IPO was the correct price for that stock. (One year later: $4.50). Such people should be treated with great skepticism.

Presidenty Hillary Clinton?

I remember in the summer of 2007 the betting markets (mainly InTrade) were giving Hillary Clinton a 80% chance of winning the Democratic nomination and a 50% chance of winning the White House. I wrote an article at the time calling such prices a "bubble." The efficient-markets brigade was outraged.

One of them wrote a long screed somewhere saying I had written "the dumbest article of 2007." And, of course, they were proven right. Hillary Clinton went on to sweep into office. (Can you remember the name of that skinny guy who ran against her? You know, the one from Illinois with the funny name? It’s on the tip of my tongue.)

The Iowa market has a respectable track record of predictions. A study by Iowa professors found that from 1988 through 2004 it was more accurate than opinion polls 74% of the time.

But it is like any market. It is a reasonable starting point, and it has a reasonable record, but it is far from infallible.

And Iowa’s predictions are also undermined by the betting in a second market at the same exchange, where instead of betting on who is going to win the election, you just bet on the share of the vote each candidate is likely to get.

Right now, the betting at Iowa is predicting Obama will beat Romney by half a percentage point. The study by the Iowa professors found that from 1988 through 2004, in the final five days before elections, the market had an average margin of error of about 1.1%.

In other words, in terms of vote share the Iowa market is calling this nearly a dead heat -- yet simultaneously giving Obama nearly twice as high a chance of winning as Mitt Romney.

Silly.

Where I'm placing my bets

I am willing to stick my neck out and argue the markets are wrong -- and, unlike most pundits, I have put my money where my mouth is.

I happen to think Obama is slightly more likely to win next week than Romney. The prez has an edge. But it’s slight.

I’ve been active in the Iowa market since early September. At the time Romney was floundering, running the worst campaign in living memory, and yet the Iowa market was giving him a 40% chance of winning. I thought that was too optimistic. I bought up Obama contracts.

The market moved heavily toward Obama over the next few weeks. Then came the first debate. Ten minutes in I cashed out my Obama position at a small profit. The markets at that moment gave Romney a 31% chance of winning. I thought that was too low. I put all my money on Romney.

I could cash out today for another profit. But I still think this market is too low. I’d buy Romney contracts up to about 43%. I’d sell them if someone offered me over 50%. This means I happen to think fair value is about 46% or thereabouts. I may be right or wrong, but at least I’m putting my money on it.

(Paradoxically, I’m betting against my own interests. A Romney victory will do wonders for sales of my biography of the man, "The Romney Files." So I ought to be hedging my bets.)

Will other pundits? If not, why not? Thank heavens I spend more time writing about the markets than I do about politics. In the markets, all this blather is moot. Show me.

More at MarketWatch.com:

• Red states vs. blue: Who has bigger 401(k)s
• 14 tax issues to watch after the election
• A look at who’s funded the election