It is the movement or fluctuations that interest me most. I mean, what external events induce the movement? Well, it seems published polls do, to some extent but every time and the odds move without there being published polls.

Starting on March 30, Coalition was shorter at Centrebet at 1.85 to Lab’s 1.95.

There was tiny bit of fluctuation but then the roles reversed to ALP 1.87 and Coal. 1.95 (may 11).

The event when this happened? Costello’s Budget on May 9.

The trend then continued with, again, small movements - ALP 1.83, Coal. 1.95 to reach almost exact polar opposite on May 17.

On May 21 the biggest spread for Rudd thus far: ALP 1.75, Coalition 2.05.

From here on ALP went backwards and Coalition clawed back some ground.

On June 13, Labor slid to 1.85 and Coalition shortened to 1.90.

Over this last weekend, Labor again shortened to 1.83 and the Coalition blew out a bit to 1.93. This, I am guessing, was as a result of money going on on the back of the AC/Nielsen.

I am wondering what the point would be of a “wealthy party” trying to load the odds in its favour by putting money on itself? To stimulate the commentariat into saying something positive? A moot exercise I should think.

The best explanation or speculation I can offer, although it has been derided by MB and indirectly by Quiggin, is that some rogue elements in the Liberal Party who were privy to marginal electorate, private, post-Budget polls, went out and quietly put on a few bucks on the ALP when it was good value, putting the money in series of tranches in order not to spook the bookmakers, but eventually blowing the odds on or by May 21.

The polls are now returning to equilibrium with speculative and uninformed money going on more chaotically.

I have observed this phenomenon on the horse races, with many a plunge taking place, the most infamous example being the Fine Cotton plunge.

(In August 1984, the racehorse Bold Personality was substituted for a yak called Fine Cotton in a race at Eagle Farm. Fine Cotton went from from 33/1 to 7/2 in some 7 minutes.)

PS Presumably, the closer we come to the election, the bigger the pool. Conversely, a long way from the election, the betting markets were irrelevant because there wasn’t much money driving the odds (in a small pool, small bets affect the price) and anyway, the bettors were mad.

That assumes odds are newly framed, or at least purely rationally framed. Eg fixed price odds.

But floating dividend style odds - such as we see in election markets - on any given day must also be a way of the bookmaker seeking to hedge their losses. Eg they can be influenced, perhaps disproportionately, by large bets placed well out from an election.

In a worst case scenario, if Dr Leigh had his way and pundits focused on betting odds over polls, a wealthy party could tilt odds in its favour with carefully placed bets (like stacking an online poll).

Curiously, in the old days when electorates were smaller, there used to be laws outlawing waging on elections for the simple reason that it was a disguised form of bribery (a candidate wanting your vote would bet you he would lose)

It seems that betting markets are largely affected by published polls because the corresponding causation effect closest to the change in relative odds has been the Herald/AC Nielsen poll just published, which brings unfavourable news to the Coalition - furthermore, it heralds (he he) a trend, that voters have made up their mind, that they are not listening, and the Rattistas are riding the tumbrel.

Since Saturday, when the price was Lab 1.85, Coalition 1.90, the odds now are Lab 1.83, Coal 1.93.

A small change but the punters are a cautious lot.

Jason Koutsoukis in The Sunday Age on superiority of betting markets over opinion polls writes:

“For the first time in more than a decade, bookmakers have for the past few months steadily rated Labor as favourite to win.”

Not true Jason. During the first half of 2001 Labor was ahead in the betting markets - often further than at any time this year.

[Update: Jason’s probabilities for both parties add up to more than 100%, which is obviously silly.
He needs a formula that allows for bookmaker’s commission, such as: Probability of Coalition win = Labor price divided by (Labor price plus Coalition price). Swap them around for Labor probability. (Simon Jackman gave me this once; it’s presumably how Bryan Palmer does it as well.)]

http://www.mumble.com.au/

A Wharton prof of public policy, Justin Wolfers, examined the various methods of predicting the outcome of our 2001 federal election and found that betting odds from Centrebet “not only correctly forecast the election outcome, but also provided very precise estimates of outcomes” in a range of electorates and note this, “particularly in marginal seats,” Wolfers reckons.

Other interesting stuff I gleaned from the piece was that Iowa Electric Markets run an online futures exchange where people can trade on elections and other economic and political events. I must check it out.

Furthermore, betting odds on elections were very popular in the 19th century and were published in the same way the polls are published: as headlines in newspapers. Betting odds on election s pre-date polls, of course.

Here is the link .

However, it is pointless to test online betting against betting on elections when there wasn’t an internet and published polls for obvious reasons - online betting on elections has changed fundamentally because of those factors:

a. People who would have never bet then (because you had to be part of a milieu such as knowing bookmakers, etc.), now do;
b. The internet allows people to bet easily and relatively anonymously, as per my point about insiders.
c. Opinion polling itself has become more sophisticated, and obviously betting on elections uses opinion polling as one of its main input sources of information analogous to punters using horse form.

The main issue is whether election betting markets are more accurate indicators than opinion polls, or whether one is simply an amplification of the other.

http://johnquiggin.com/index.php/archives/2007/05/26/polls-pundits-and-pollsters/

http://www.johnquiggin.com/archives/001232.html

The idea that speculative markets can be used to forecast political events hit the headlines a while ago with the furore over terrorism futures. This idea is still around and the general claim that political events can be forecast by futures or betting markets is still being pushed hard. The main source of data is at the Iowa Electronic Markets, but there’s plenty more. Reader Jack Strocchi sent me this report on a study of Australian betting markets and elections.

As it happens, I’d already looked at this and come fairly rapidly to the conclusion that the betting markets weren’t much good, so I was struck by the money quote from author Justin Wolfers

The data suggests the Australian betting market is extraordinarily efficient. And why not? There’s a huge incentive for participants to do their homework, collect reliable information and make sure the price is right.”

Looking at the report and also the Iowa studies, the evidence in support of this claim still seems very weak to me. In 2001, for example,

The night before the election, Howard was ahead in two of three major polls ….[on Centrebet] Howard was the favorite with odds of $1.55, suggesting a 64 percent probability of his winning the election,”

That is, on the crudest possible use of the polls, two out of three suggested a Howard win, giving odds almost identical to Centrebet. In fact, I doubt that any serious analyst would have given Labor even a 25 per cent chance by election night.

To be fair, Wolfers doesn’t put much weight on the election-night market. He says

data from Centrebet, Australia’s largest bookmaker, demonstrated the impact of current events on the betting odds throughout the nine months leading up to the election, reflecting immediately the electorate’s seesawing response to such events as leaders’ televised debates and the Sept. 11 attacks in America.

In fact, however, the betting markets reacted more slowly than the polls. In this piece written in September 2001, Wolfers and his co-author Andrew Leigh rated Labor a 55 per cent chancebased on the Centrebet data.

But enough of this ad hoc discussion. What test should we be applying here? It’s not appropriate, as nearly everyone in this field has done to treat polls and betting markets as separate predictions. Punters in the betting markets have access to the polls. So they should always do at least as well as any mechanical rule based on poll data. The test “have the markets done better than the polls” implicitly compares the actual betting strategies to the rule “at even money, bet on whichever candidate is ahead in the polls”. Even compared to this simple-minded rule, the improvement shown by the markets is marginal at best.

The real issue we should consider, before rushing to embrace terrorism futures and the like, is how betting markets would perform in the absence of information from polls. You’d have to go back before World War II for this, but it’s my impression that predictions of election outcomes in this period were often way off the mark.

Election betting is highly unusual in that the “race” or “game” happens only once, so there is no opportunity to improve one’s position with subsequent bets and with just two “runners” not much opportunity to lay off on other runners, unless the odds are fluctuating wildly over time and you take advantage of riding the markets. But this hasn’t happened and is unlikely to happen in Rudd v Howard.

The fact that the bettors are self-selecting is not a flaw of methodology because betting markets are not polls. Betting markets, as you have correctly pointed out, take in raw data such as polling but other indicators and guesswork as well. We have traversed some of them in this thread. What comes up is the result of analysis by a collective wisdom, a hivemind of bettors, who make an unsentimental, gimlet-eyed decision (I am hoping) because it is their money that is at stake. And from that I have been trying to draw some conclusions. Indeed, what started this off is my interest in why the betting markets were fluctuating at any given point in time.

I don’t want to buy into your notion that betting markets behaviour is made up of self maximising rational decisionmaking because it would set my argument, and myself by extension, as some sort of vulgar Hayekian, which it isn’t and I am not. But again, we can’t discount a theory that has some similarities with ideas held by people we don’t like. That would be irrational.

But most market behaviour stuff is a bit too theological for me.

Polls aren’t supposed to predict the result of an election. To the degree that they have some predictive value, it’s probably highest just before an election, but all they really measure is the sentiments of those sampled at the time they’re taken. But those people are sampled scientifically. With betting markets, it appears to me that all you can say is that the movements in price reflect the opinions and calculations of those who bet! And we don’t, frankly, know who they are, and they’re self selected rather than sampled.

The whole argument generally seems to me to rest on all sorts of flawed liberal premises about self-maximising rational individuals.

I can take you down to the Treasury casino and demonstrate to you quite easily that most people who wager on things do so quite irrationally.

And of course we are dealing here with trends and nuances, not absolutes.

Just one or two people, having a lash on the back of some dynamite news, or rather, its intelligent projection into voting intention is enough to alter the price substantially.

Yes, absolutely, it distorts the price just as in economic markets. But it all comes down to voting intention being translated into actual voting in the end so the speculation is always tempered by a moment of truth, whereas economic markets, like in Tokyo real estate or dot com hysteria of 2001, or derivates, do not have to have any relationship with reality.

And I doubt a lot of these insiders are betting anyway.

When I say a restricted circle has access to party polling, Ken, I mean really restricted. Probably less than ten people. The people who conduct the polling aren’t apparatchiks but commercial pollsters whose ethics, and reputation, would preclude any misuse of the information.

Anyway, if the analogy is with economic markets, it’s usually considered that insider trading distorts price.

Regardless of their ideological stance, they will calculate what sort of impact such information may have on the electorate once it is released.

Because they have got hold of this info early, before the markets or for that matter opinions polls are aware, they can place a bet at an advantageous rate of odds.

If the information is of high quality and likely to be of some import - say a major scandal involving a minister which hadnt yet broken, or economic good or bad news, or access to private polling not available to the public - their bet will soon affect the odds if the bet is substantial enough. But it is cause and effect. The bet comes first, the market reacts and adjusts. The bettor can be one individual betting the house on it or a number of individuals making more moderate bets.

Hence this is different to polls because it is an educated assessment of information on how it will affect voting after the information becomes known widely.

Once the bet is on, the odds will alter to reflect this and keen market watchers will move in concert, suspecting something to be afoot. In betting parlance the first is called “educated money” going on, and the latter are people who “follow educated money” as a betting strategy.

If party-polling results are compelling enough, say, as you have suggested, tracking undecided voters in marginal seats - and now I am just speculating here - what would stop an individual or two betting on that basis? Their bets would be untraceable. I hope you are not suggesting that Liberal Party apparatchiks and their polling mavens would be above common greed?

Nor do I understand what “inside info” means - someone might know that the government plans to release some whizzbang policy - but if they know that they’re probably a government staffer or someone who is ideologically biased anyway. Often either scandals or policy announcements fail to make the impression “insiders” think they will. The only “inside info” that I’d be interested in seeing is party polling which usually tracks marginals and undecided voters. But it’s held very close to the chest - only a very restricted circle see it.

If there were no published polls, a market would still be framed on the basis of any indicators available to that punter. Betting markets take into account published polls, sure, but on quite a few occasions the betting markets have been at odds to the published polls.

I may add that betting markets are a form of opinion poll but where mere talk is cheap, especially when it is given anonymously on the phone, betting requires a greater analytical input from the punter because there is a tangible monetary penalty for being wrong.

There are different kinds of punters. Lazy ones who merely follow opinion polls and bet accordingly just like the people who follow footy and horse tipsters. Then there are “educated money” punters who may have, or think they have, inside information, such as private party polling, or impending interest rate rise, or a scandal not yet broken. And then there are those who analyse raw data.

Also worth remembering is the fact that one punter, or a handful of punters (a cabal) feeling strongly enough, can affect the odds by betting big enough. This could mean inside info, or it could mean mischief, as I speculated above.

If there were no published polls, what would betting markets be driven by?

I still think they’re a useless measure. They measure sentiment and perception.

However, my friend, the Reverend Bro Geoff, reckons that Lib supporters are deliberately poulticing online betting shops as a means of skew public perceptions. (Bit like buying your own shares, a la Rodney to maintain the price). It doesn’t take much to do that at centrebet, just a few grand would do it. The others will then follow as they klay off. Bro Geoff is a seriosuly committed horse punter, so his advice re bet markets practice is worth taking seriously.

http://www.youtube.com/watch?v=4MzkW2_CXBo

Mind you they surveyed 417 people in each electorate totalling 1251 voters.

More info at Mumble. http://www.mumble.com.au/