Who won the last presidential debate? Who's going to win the US election? According to the betting and political futures trading communities, it was -- and will be -- George W Bush.

Conservative debate watchers around the globe watched with considerable incredulity as US network commentators -- and small groups of poll respondents -- handed the last debate victory to presidential election challenger John Kerry, but in the realm where "polling" is done with money, there was no doubt that even that media-heralded victory would translate into defeat for the challenger at the polls.

Betting and political commodities communities have a strong record of predictive success when it comes to such things, not surprisingly, and Stanford Graduate School of Business faculty member (and native Australian) Justin Wolfers has an explanation for that, one most of those who understand markets can appreciate.

"The rationale for this happening comes from finance, which says that markets are efficient aggregators of information and equilibrium prices should reflect all available information," he told the Stanford Graduate School of Business two years ago.

In a July 2002 paper -- Three Tools for Forecasting Federal Elections: Lessons from 2001 -- published in the Australian Journal of Political Science, Mr Wolfers looked at two common tools used to predict Australian elections (polls and forecasts based on economic models) and compared them with a third: betting odds.

In all cases, he found, betting odds had an extraordinarily high success rate, especially in marginal call elections.

Similarly, betting odds on US elections have proven to have extraordinary predictive value over more than a century.

Paul W Rhode and Koleman S Strumpf, economics professors at the University of North Carolina, Chapel Hill, said in recent paper that election betting correctly predicted the popular-vote winner in every presidential race -- but for one -- between 1884 and 1940. That exception took place in 1916, when a handful of California voters handed the contest to Democrat Woodrow Wilson. The professors note that, although Wilson's challenger had been heavily favoured in the month running up to the election, odds had narrowed to even chances by the time the polls closed. In all other cases, the betting odds established within a month of the election were accurate predictors of the outcome -- even in the absence of quick access to information about factors that might influence local voting populations.

After 1940, betting on elections became legally difficult in America -- and there was a 48 year gap in which no American-hosted betting pool data could develop.

But since the election of 1988, one source, the electronic market run as a test lab by the business school at the University of Iowa, has accurately predicted presidential outcomes based on futures contracts.

Iowa Electronic Market

The Iowa Electronic Market has correctly predicted every US presidential election since 1988 and is picking George W Bush as the winner this year.

The IEM is a pedagogical tool, used for research and teaching about real-world markets and investors -- anyone can become an investor -- are limited to $US500 buys, partly in order to exclude speculative manipulation of the market. One market is directed toward the popular vote, another to the electoral college voting outcome -- something that would have led to a split ticket in the 2000 election, where Al Gore won the popular vote but George W Bush won the election.

In the popular election market, Mr Bush has held the lead almost without exception since the inception of trading in June. After taking a very wide margin following the Republican National Convention, however, the odds moved toward parity until the conclusion of the third debate, at which time they reversed course. At present, Mr Bush is rapidly gaining ground and Mr Kerry is rapidly losing investor confidence.

"The central premise is that markets can reveal information about future events," Forrest Nelson, an economics professor at the university and board member of the Iowa Electronic Markets, told Wired Magazine two years ago.

Mr Nelson told Wired that in the last four presidential contests, traders' predictions for candidate vote totals have come within 1.5 per cent, on average, of the actual results.

Political races as markets: TradeSports

The IEM is hosted in America and is an elegant model, but real markets are considerably more rough and tumble.

Over at Dublin-based TradeSports.com traders can deal in a range of election-related propositions like the number of electoral votes Mr Bush will receive after the election and whether or not he'll win at all. In the 24 hours after the second debate, his chances of winning increased by 1.7 on the betting board while Mr Kerry's dropped 0.1. That increased the odds of a Bush victory only marginally, but probably only because Bush was already a heavy favourite, with bids at 57.1 to Kerry's 41.6. Asks for Mr Bush were at 57.7 and for Mr Kerry, 43.9, following the debate.

At TradeSports, bids are lodged in $10 blocks and values are established by bids and asks. A "0" value indicates the exchange members think an event will not happen and a "100" value indicates they think it's a certainty. When an event transpires, the value of the commodity settles at 100 and returns $10. The higher the value assigned to a proposition, the greater the likelihood it will occur.

If you bought 100 “Bush to win in ’04” contracts at 57 (57 being $5.70) you would be investing $570 (100 times $5.7) to win $430 (ie. 100 times $10-$5.70).

But because it is an open market with no limits on the amounts investors can risk, there has been considerable speculation that a series of recent sharp dips -- beginning on September 14 -- in Mr Bush's futures was caused by a rogue speculator attempting to send a signal to the electorate. Investors need not disclose identities on TradeSports, so some in the conservative political camp dropped heavy hints that the surge for Kerry futures was the doing of George Soros, a well-known, anti-Bush trader who did something very similar with the British pound 12 years ago.

But some very astute commentators don't go along with this proposition, noting that the "push" may simply have been a move by a trader to generate profit and that it levelled out within hours.

Others note that the Soros move against the pound is widely believed to have netted him over $US1 billion in profit -- and that Mr Soros is not the only person in the world capable of making large, speculative gambles on futures markets.

As of today, Bush is moving at 56.5 on asks of 57 (volume: 749.7k) -- up 1.9 for the day -- and Mr Kerry is moving at 43.6 at asks of 44 (volume: 48.1k) -- down 1.8 for the day.

Bookies online

More straightforwardly, there are dozens of betting services that offer odds on political contests like the US presidential election -- and most of them show Bush as the odds on favourite -- that is, if you sink x dollars into a bet for Bush and win, your profit will be less than the amount you bet. (You'll also get your stake back, of course.)

Right now, betting on a Bush victory is running around the 4/7 mark at most betting agencies. Simply put, that means a successful $7 bet stands to make the bettor a $4 profit.

Mr Kerry is gathering odds that range from 11/10 to around 5/4 -- not exactly a dark horse, but definitely the underdog. At 5/4, for example, an $4 bet would return $5 if successful -- $1 of which would be profit.

Put another way, betting $100 each way would net a punter 14 chances of winning $4 ($56 + $100) on Mr Bush and 25 chances of winning $5 ($125 + $100) on Mr Kerry.

Closing odds

In recent US presidential elections, a candidate holding anything above a 60 per cent chance of winning a month out from the election, won.

Mr Bush is holding at that threshold with about two weeks to go and betting momentum is clearly signalling an increasing likelihood for his November win.

Ultimately, there's always the possibility of a Wilson-like victory from the outside for Mr Kerry -- except the small pool of voters involved this time would be more likely to come from Ohio or Florida than California.

But gambling newsletter Online Casino News says London bookmaker Ladbrokes -- which is trying to recover from a staggering bet placed on Mr Bush to win -- predicts President Bush will win in 31 states and Senator Kerry in 19, plus Washington DC.

What are the odds?