Home > Oz Politics, Politics (general) > Polls and punters, yet again

Polls and punters, yet again

August 17th, 2015

I just read this piece on The Drum, taking the line that it’s better to rely on the betting markets, which have Labor and the government level-pegging, than on the polls, which have had Labor well ahead for a long time. Elections are only held every few years, so they don’t provide much data on which to test the relative performance of the two. But, if markets give better estimates than polls, we should expect to see movements in the poll results follow those in the market rather than vice versa (in econometrics, this is called Granger causality). Digging around, I located a study finding that, if anything, movements in polls Granger cause movements in betting markets.

Since a compelling observation beats an econometric analysis for most people, let’s look at the 18 months or so since I last posted on this topic. Labor started out with a small lead in the polls and stayed consistently in front, with the lead varying over time. Meanwhile, the betting markets favored the government until very recently, before moving to even money. It seems clear in this case, that the markets are following the polls and not vice versa.

Categories: Oz Politics, Politics (general) Tags:
  1. David Irving (no relation)
    August 17th, 2015 at 17:33 | #1

    Damn! I should’ve got a bet on Labor a while ago, before the price narrowed.

  2. Doug
    August 17th, 2015 at 18:03 | #2

    Evidence on this looks pretty clear – aside from evidence from internal party polls it is a bit difficult to know what other evidence would shift the the betting odds

  3. Collin Street
    August 17th, 2015 at 18:33 | #3

    Elections are only held every few years, so they don’t provide much data on which to test the relative performance of the two.

    We’ve got enough historical data to get a fair idea of the accuracy of the polls, at least. Since the historical gap between polls and election results is smaller than the gap between poll results and bookkeepers’ predictions, it’s hard to avoid the conclusion that it’s the gamblers who are most likely wrong here.

    [why don’t they do this analysis? Most people do, which is why most people don’t gamble: to a first approximation, the only people left in the gambling market are the ones with excess confidence in their predictions. As a self-selecting sample bias is to be expected, although exactly where and how is hard to predict in advance.]

  4. Ikonoclast
    August 17th, 2015 at 19:07 | #4

    I have no interest in gambling and no interest in which of the neoliberal duopoly parties wins so… yawn.

  5. Steve Hamilton
    August 18th, 2015 at 01:32 | #5

    I don’t understand this statement:

    But, if markets give better estimates than polls, we should expect to see movements in the poll results follow those in the market rather than vice versa (in econometrics, this is called Granger causality).

    Surely nobody would suggest that the polls contain no information, nor that the betting markets should not respond following the revelation of this information. But the betting markets reflect not only the information in the polls, but a raft of other information too. This is what makes them, iin theory, strictly superior to the polls.

    As such, putting aside whether you believe that the polls are inferior to the betting markets (or the EMH upon which this idea is based), why does the observation that the betting markets respond sequentially to the polls invalidate the superiority of the former? If anything, I think that the opposite is true – that is, that the betting markets responding to the polls suggests their superiority (in terms of information content).

  6. Uncle Milton
    August 18th, 2015 at 02:38 | #6

    The polls ask who would you vote for if the election was held today. The betting market predicts the election outcome knowing that it won’t be held today and guessing about when it will be held. It is not necessarily inconsistent for the polls to be showing Labor ahead in the polls and the betting markets to predict the coalition will win the election to be held in 12-13 month’s time. After all, no first time federal government has lost in over 80 years.

  7. Collin Street
    August 18th, 2015 at 06:40 | #7

    why does the observation that the betting markets respond sequentially to the polls invalidate the superiority of the former?

    Because it casts doubt on the existence of any mechanism wherewith the betting markets could acquire superiority.

  8. Ikonoclast
    August 18th, 2015 at 07:16 | #8

    Isn’t the betting market a self-selection of foolish people?

  9. Collin Street
    August 18th, 2015 at 07:20 | #9

    > Isn’t the betting market a self-selection of foolish people?

    Self-selecting, yes. Foolish? Certainly we can expect that they’re more-than-usually confident in their judgements… which is probably a mistake given that research shows that people are typically overconfident anyway. But if you were overconfident you’d express your overconfidence in multiple ways: most of the people who tout the value of the betting market seem to be people who engage in it. I have a comment in moderation that goes into a bit more depth.

  10. August 18th, 2015 at 09:19 | #10

    I published an election forecasting model at the last election at the Financial Review. https://electionlab.wordpress.com/

    It performed only modestly well.

    In doing so though we faced the argument many times that if a betting market responds to a poll, the betting market is proving its own incompetence. I have to disagree.

    Betting markets are supposed to collect the wisdom of people who imagine they know what sort of effects event between now and election day might have on the result. Every poll gives those people a new baseline to work from. The poll establishes the state of play (if an election were held today…) to which the imagined insiders apply their prognostications (for example, the knowledge that the government will roll out a truly impressive array of national security scares before any poll.)

  11. derrida derider
    August 18th, 2015 at 09:41 | #11

    But the idea is that betting markets are better at predicting who will win than a poll asking who will win. Actual polls don’t ask who will win but who you would vote for – an entirely different question, so it’s no surprise that such polls influence the betting on who will win more than that market will influence who people want to win.

    The reason betting markets are better predictors than a poll asking who will win is simple of course – people putting hard money where their mouth think longer and harder about it, try and get inside information, and will be less prone to let their heart rule their head.

    On the last I’ve always thought a rationally risk averse person would bet against their heart to hedge their utility if their favoured side loses. Though of course if they were rationally risk averse they wouldn’t be taking part in the zero or negative sum game of a bet in the first place (unless they had that inside information).

  12. Ikonoclast
    August 18th, 2015 at 11:04 | #12

    @Collin Street

    Those who run betting markets (bookies, casinos etc.) construct the odds market so that over many trials (gambling events) they will win. They stack the odds.

    Anyone entering a rigged competition is ipso facto foolish unless;

    (a) they have rigged the competition themselves in favour of themselves;
    (b) they are indeed uncommonly knowledgeable and clever and can beat the odds (rare but some such people do indeed exist for some gambling events);
    (c) they have effective insider knowledge;
    (d) they can re-rig (cheat somehow) to sway odds back in their favour.

    Even then (c) and (d) are probably foolish as they will likely be caught and punished sooner or later. The gains from cheating usually don’t stack up too well against the punishment which could include disgrace, banning and prison as well as fines. What’s the opportunity cost of losing your livelihood and liberty?

    I can’t prove this and I don’t know if my view has any real meaning but my guess is that since most elections are close two horse races (essentially) and polls have some predictive ability then betting who might win is not all that hard. No matter who you bet for (of the duopoly) you have about 50% chance of being right. I mean does cogitating about this stuff really illuminate or elucidate any real point? I have my doubts.

  13. David Irving (no relation)
    August 18th, 2015 at 11:17 | #13

    @derrida derider
    [Gamblers] “will be less prone to let their heart rule their head.”

    Um, no. That doesn’t match up with the behaviour of most gamblers I’ve met.

  14. Newtownian
    August 18th, 2015 at 11:20 | #14

    Thanks for introducing Granger Causality – interesting rabbit hole which my Bayesian causality /inference silo hadnt yet linked to.

    On this matter of causality and polls I’m surprised both here and in the previous article there is no mention of Nate Silver.

    While he is a bit pretentious and egotistical and far from the only person who has looked at the maths of causality and Bayes his story is still very interesting and relevant to this discusssion.

  15. Newtownian
    August 18th, 2015 at 11:28 | #15

    Ikonoclast :
    I have no interest in gambling and no interest in which of the neoliberal duopoly parties wins so… yawn.

    Having read your contributions and finding most I agree with, I depart I think at this point.

    I used to be distinterested in gambling and still see those poor souls in the TAB at 9pm as rats in a trap, but the probability theory underlying gambling and inference is a another matter altogether – Linked as it is with free will, causality and determinism which controls whether we can actually change the economic system to an environmentally sustainable one by exercising out intelligence or whether we are effectively locked into a giant pin ball game where we have little control.

    I guarantee you would be fascinated to explore this rabbit hole if you have a mind.

  16. Ratee
    August 18th, 2015 at 12:09 | #16

    Its an interesting comparison – on the odds of Abbott being next PM vs Shorten, rather than the LNP vs the ALP.
    I still have a couple of bets – a Mars bar @ 50/50 on Abbott being PM after next election. No one will take wider odds, even though they confidently say he is done for when the buck stops ts a toss-up.

  17. John Quiggin
    August 18th, 2015 at 13:31 | #17

    @Steve Hamilton

    On the standard theory, betting markets should immediately incorporate poll data (to the extent it contains new information). What we see here is that they are catching up to the polls with a lag of many months. That’s what the Granger causality tests find also.

  18. Historyintime
    August 18th, 2015 at 14:11 | #18

    Anecdotal but I have been following this very carefully since election betting began, both in Australia and OS. Betting markets are way more accurate than polls, 18 months out. By 7 days out polls are better than markets – ie there can be a market inefficiency there at the death knock.
    But to exploit it you need to know something about electoral politics – ie not arbitrage but judgement (there is alpha, so to speak). I expect this inefficiency will wither away over time

    And betting markets definitely follow polls rather than the other way around.

    And internal polling is way overrated. Professional public polling is way more accurate than internal polling State and national wide. But internal polling can provide good qualitative information and seat by seat analysis.

  19. Ivor
    August 18th, 2015 at 15:39 | #19

    Bob Ellis has this on his blog:

    The Ipsos actually shows Labor on 56 and Abbott needing to win back a million votes to survive wth a 2 majority. This is the worst result for any party since Federation, and would lose Abbott his own seat.

    As with Newspoll, the Ipsos headline is a cheat. 46 is what the Coalition would get if those who vote Independent dispersed their preferences as they did in 2013 when they thought Abbott was telling the truth on his policies. Asked how they would disperse them now they gave him 44. Why the 2013 figure is used by anyone is a good question. The only answer is ‘to have a headline better for Abbott’.

    Morgan has Labor on 57.5 and is in agreement, pretty much, with Ipsos. It is a result that would lose Abbott 50 or 55 seats, including his own.

    Will he be overthrown this week? Well, it’s possible. Robb or Smith may replace him, or maybe Turnbull’s hour has come.

    Long story short – bye, bye Abbott?

  20. Newtownian
    August 18th, 2015 at 16:11 | #20


    A political junkie friend loaded this on to me also on sunday night after the ‘Insiders’ program. A side element in this story was about some lunch meeting between just the Murdochs (father and son) and the beloved Devine and Sheridan in some posh restaurant in Melbourne which they booked out for just this four – was this a thinly veilled message of intent given they have already turned on Bronny?

    The alternative he is now pushing to me is plausibly (are all these narratives?) is post the WA by-election.

    Not that such gossip interests me at all.

  21. Collin Street
    August 18th, 2015 at 16:19 | #21

    > Long story short – bye, bye Abbott?

    No, because in the process of turning from an ugly Young Liberal caterpiler to a beautiful beautiful backbench butterfly doesn’t make you any smarter and doesn’t make you any more self-aware.

  22. derrida derider
    August 18th, 2015 at 16:26 | #22

    EMH and betting markets aside, if I was Bill Shorten I’d be hoping Abbott hangs on as leader. Far better to be facing a crippled Abbott than a settled-in Turnbull next year.

    Were I Shorten I might just run dead for a little while to maximise his chances of survival.

  23. rog
    August 18th, 2015 at 17:27 | #23

    @derrida derider Well I think that has been Shortens play all along.

  24. rog
    August 18th, 2015 at 17:30 | #24

    OTOH Turnbull is hated by most Libs and Morrison isn’t exactly Mr Nice Guy.

  25. Ivor
    August 18th, 2015 at 17:45 | #25


    I think you read my mind.

    I want to predict that Abbott is rolled if there is a swing after Canning by-election but I also want Abbott to stay so they will be wiped-out across Australia (say March 2016).

    I always look to the long game, and a Turnbull will be much harder to defeat.

    I suppose Devine and Sheridan have been given their instructions.

  26. Megan
    August 18th, 2015 at 23:27 | #26

    According to “sportsbet”, Abbott is favourite by a mile to take the LNP into the next election.

    This is an easy bet (but not valuable at $1.40).

    The LNP has two recent lessons to follow here:

    1. Howard stayed around far too long and his party room was too timid to tell him to hand over to someone else;

    2. Rudd was determined by faceless ALP backroom stooges to be a danger to the neo-liberal world order and took him out mid-first-term.

    The Rudd lesson is the most important one here. The Australian people sack PMs. And they get very angry if party operatives try to take that one small joy away from them.

    As the election draws nearer and everyone realizes that we really might end up with PM Shorten (faceless nothing man), the polls will probably become much narrower. The people don’t want to be forced to choose between “the lesser of two fascists”. We’re sick of it.

  27. Ivor
    August 19th, 2015 at 00:18 | #27

    So how much of this unadulterated defamatory crap are we expected to stomach?????

    faceless ALP backroom stooges … the lesser of two fascists

  28. Uncle Milton
    August 19th, 2015 at 03:26 | #28


    Everybody to the right of Mao is a fascist. 🙂

  29. Ikonoclast
    August 19th, 2015 at 07:04 | #29


    I don’t think you understand Australian society and politics at all. Robust debate and healthy disrespect for leaders has always been part of the Australian ethos. Australians understand that almost all “leaders” are hollow men (and women). We understand that they are domineering types who just like to push other people around. We treat them with the disrespect they deserve.

  30. Julie Thomas
    August 19th, 2015 at 07:44 | #30


    Ivor if you are not Mel are you Mark Latham?

  31. Ivor
    August 19th, 2015 at 08:15 | #31


    The examples I have posted – sometimes deleted by John Quiggin – are not “healthy”.

    There is such a thing as defamation law. Both author and publisher are liable. John Quiggin is the publisher. The comments I have identified spewing forth are in fact malicious.

    John Quiggin is responsible.

  32. Julie Thomas
    August 19th, 2015 at 08:54 | #32


    Your attitude is the reason that there are many people of the left who do not vote for the ALP. It is this authoritarian and irrational repression of people who do not toe the party line that damages Labor.

    You are a classic example of all the things that have been wrong about ‘the left’ and the reason so many otherwise sensible people do have a thing about the faceless men of the labor party and won’t vote for you and your kind.

    Your attitude does more damage to the cause than any amount of bad language directed at the political arm of the party.

  33. John Quiggin
    August 19th, 2015 at 09:15 | #33

    Ivor, as you’ve indicated, your style of argument attracts hostile responses, leading you to threaten defamation action. As publisher, I’m going to protect myself by asking you to take a month off commenting. When/if you return, please try for a more civil style.

    Megan, I think your posts are part of the problem. Please, no further use of the term “fascist”, except literally, in reference to Mussolini and similar. Also, I think we understand your view that the two major parties are equally bad, so you don’t need to keep restating it. Can you avoid further comments along these lines unless you have something genuinely new to say.

  34. August 19th, 2015 at 09:24 | #34

    Megan, well done, tenacity with continuing eloquence! though, respectfully, your “faceless alp backroom stooges” is pretty unremarkable, your characterization of the situation for voters as one of “the people forced to choose between ‘the lesser of two fascists’ ” is brilliant, though i don’t understand the need for quotation marks around ‘the lesser of two fascists’, is it a citation? a literary allusion? anyway, you succinctly & boldly as usual capture the present predicament & direct a spotlight at the heart of the problem – which is that both alp & lnp, (and murdoch and much of the establishment) uncritically support modern american led fascist imperialism. i enjoy your stuff in the morning. bracing! -alf.

  35. Newtownian
    August 19th, 2015 at 09:49 | #35

    While we are on the subject of betting of winners here is a nice piece form the Independent on Jeremy Corby – it looks like those of (left wing) faith are set to clean up.

  36. Steve Hamilton
    August 19th, 2015 at 10:12 | #36

    This would be consistent with the polls having no information content, though. @John Quiggin

  37. Steve Hamilton
    August 19th, 2015 at 10:14 | #37

    And do the betting markets have to be entirely frictionless to be superior?

  38. Ivor
    August 19th, 2015 at 10:22 | #38

    @John Quiggin


    Finally, if it takes the law, then so be it.

    I note however that another poster has immediately started spewing the same word at least three times in its following post.

    So I reserve my right to deal with it.

    It is perfectly legitimate to argue that there is little difference between parties – I agree with this.

    That was not the project being run through a long series of posts on this blog.

  39. wilful
    August 19th, 2015 at 10:31 | #39

    Ivor, get your hand off it mate.

  40. Jips
    August 19th, 2015 at 14:22 | #40

    QLD election the betting agency said far more people backed ALP but a few big votes backed LNP. So just means Liberals spend more trying to manipulate markets……hmmm polls I mean.

  41. Tim Macknay
    August 19th, 2015 at 17:25 | #41

    However annoying you may find it, the idea that hurling insulting epithets at political parties and their supporters in general is defamatory is completely risible.

    If people on blog threads keep saying things that you find annoying, threatening to sue for defamation is probably the most counterproductive possible way to respond, other than making a death threat. People say annoying things on blog threads all the time. Why not just ignore them?

  42. Megan
    August 19th, 2015 at 21:59 | #42

    I’m placed in an interesting dilemma.

    Back in April on a “Monday message board”, I clearly explained the reasoning behind the use of the term as follows:

    There is a wide range of differing views of the definition of the term. I use it in a broad way that fits one definition:

    “Fascism should more properly be called corporatism because it is the merger of state and corporate power.” – Benito Mussolini.

    I agree generally with the following sentiment:

    “The early twentieth century Italians, who invented the word fascism, also had a more descriptive term for the concept — estato corporativo: the corporatist state. Unfortunately for Americans, we have come to equate fascism with its symptoms, not with its structure. The structure of fascism is corporatism, or the corporate state. The structure of fascism is the union, marriage, merger or fusion of corporate economic power with governmental power. Failing to understand fascism, as the consolidation of corporate economic and governmental power in the hands of a few, is to completely misunderstand what fascism is.”

    I understand that a lot of people quibble about whether there is necessarily a “nationalistic” element before the term can be used. In this globalised corporatised 21st century, I think that aspect (nationalism) is not essential to the definition.

    I’m happy for other people to call it other things, but I’m going to continue to call it fascist.

    But now I’m not allowed to, unless I’m referring to certain fascists – such as the source of my original explanation of the modern applicability of the term.

    Maybe I can refer to “corporatist” in future, but it loses its punch somewhat.

  43. O6
    August 20th, 2015 at 06:10 | #43

    I’d like to support Megan. We should be able to name it when we see it.

  44. Ikonoclast
    August 20th, 2015 at 08:02 | #44


    Perhaps we need another term. If we follow the evolution of corporatism we see basically;

    We are now in a period of advancing post neo-corporatism. Big unions and big government are being progressively weakened, corrupted or even destroyed in some cases. This leaves big business (as corporations) as the only genuine ruling force.

    It’s hard to find a neat term for it. I would suggest Corporate Dictatorship. The functional facts are these. The corporations tell the duopoly Parties what to do. We fill out slips of paper every three years to pick a Party. Then the Party tells us what to do according to corporate dictate.

  45. Collin Street
    August 20th, 2015 at 09:27 | #45

    > I’d like to support Megan. We should be able to name it when we see it.

    It’s probably worth noting the historical connections of fascism to the reactionary wing of the catholic church, and also the connections between the reactionary wing of the catholic church and the [current] liberal party and [historical] labour right, to be fair.

    Megan’s language obscures a lot of the distinctions I think are important / useful, but what’s “useful” depends on the context: in a different context than here I wouldn’t have any problems.

    [the nazis weren’t catholics, but they were anomalous among fascists in a number of other ways also [lot more internationally aggressive, stronger focus on genes less on culture] and may possibly be best-regarded as a kindred group rather than fascism-properly-so-called…]

  46. John Quiggin
    August 20th, 2015 at 10:31 | #46

    Megan, feel free to use “corporatist” if you think it’s appropriate. I have to say, I don’t perceive any significant element of corporatism in Australian politics. Hawke aspired to a kind of corporatism but business didn’t buy it, and the Accord rapidly boiled down to a series of deals between government and a powerful interest group.

    But, to be absolutely clear, I’m going to put the term “fascist” in the automoderation list for the moment. I’ll rescue any comments about actual fascists, and delete any use as a term of abuse, analogy etc.

    I think the term you and others are looking for is “oligarchy”. Rule by a wealthy elite was a well-known form of government millennia before Mussolini added in elements like mass mobilization, uniforms, a corporatist ideology and so on (none of which appear to me to be present in Australia today).

  47. Megan
    August 20th, 2015 at 11:05 | #47

    OK, I’ll settle for one of those in the circumstances.

  48. Tim Macknay
    August 20th, 2015 at 11:24 | #48

    To my mind, the trouble with equating corporatism with the ‘f’ word is that corporatism is a much broader term that encompasses political forms that are not ‘f’, like Scandinavian social democracy, and FDR’s ‘New Deal’ arrangements. As Prof Q says, Hawke’s ‘Accord’ with the trade unions was something of a failed attempt at corporatism (of the social democratic kind), and the contemporary market-liberal thinking that dominates the agenda of the major parties is not corporatism at all.

  49. Ikonoclast
    August 20th, 2015 at 13:15 | #49

    In this broad debate, a key mistake made by at least one pundit has been to conflate party politics with politics. There seemed to be an assumption that if one did not participate in party politics then one had no right (and no way) to participate in politics at all. To the extent that this might reflect a part of our current reality it only illustrates what it wrong with our system.

    There was also the assumption that if one did not support Party A then one had to support Party B. Furthermore, if one had big reservations about Party B the only thing to do was join it and reform it from the inside. No other action was seen as politically valid. However, it seems to me that it might well be a valid judgement to conclude that Party A and Party B are now well “beyond the pale” as the saying goes. That is to say they are both so corrupt and degenerated that the idea of reforming them from the inside is an absurdity.

    I think it is perfectly valid for a person to stand outside of both party politics and bourgeois representative politics and say that these are becoming failed systems. They cannot be reformed from within their own grave limits and biases. I think part of the problem now is “gaming”. These systems were better than their predecessors and also better until their formal characteristics were more understood. It has now become clearer how to corrupt and game them. The creation of parties and disciplined voting on party lines is in essence a corruption of the representative intent.

    Representative democracy itself is seriously limited in comparison to full community and workplace democracy. The evolution of full democracy (if it can ever be managed) is only in its infancy. We have a long way to go. If we get there, today’s politics will seem as primitive, corrupt and arbitrary as medievalism seems to us. One sign that we have got there is that political parties and all the egregious corruption and partisan interests that go with them will be consigned to the dustbin of history. This can occur when broad participatory democracy and workplace democracy are established.

  50. Megan
    August 21st, 2015 at 01:39 | #50

    I may have arrived at a solution to my dilemma.

    The issue is the melding of the corporatised/globalized corporate sphere of power with the supposed repository of citizens’ power (‘democratically’ elected government within a rigged party system – in Australia, a duopoly therefore by definition) to deliver a force of control over society – an “oligarchy” – which is effectively a single source based on a shared ideology of those operating in each sphere.

    We call this two things (pretending, or believing, they are distinct), “democracy” and “free-market capitalism”. But we mostly go along with it regardless of its dishonesty and failings.

    This is clearly ridiculous.

    One definition of “farce” is: “n. a comic dramatic work using buffoonery and horseplay and typically including crude characterization and ludicrously improbable situations.”

    Which is one way of portraying the rule under which we live.

    So, in order to express the concepts I previously bundled under the problematic and now banned word, I will use new terms: “Farcist” and “farcism”.

    Speaking of which, I strongly recommend putting the kids to bed before reading the submissions to the Senate Committee on what these farcists have been doing to real human people on Nauru over the last few years.

    I particularly recommend submission #95 and the supplementaries.

  51. Tim Macknay
    August 21st, 2015 at 12:16 | #51

    So, in order to express the concepts I previously bundled under the problematic and now banned word, I will use new terms: “Farcist” and “farcism”.

    Heh heh. Perfection.

  52. Megan
    August 23rd, 2015 at 12:08 | #52

    In the UK Labour leadership contest the “machine” has taken to telling members they won’t be allowed to vote. It is all part of the establishment freaking out over the possibility of a Corbyn win.

    The hashtag #LabourPurge has been trending.

    The film-maker Ken Loach has been told he can’t vote (he once left Labour to join “Left Unity”).

    One tweet from a Labour member said she received the email and when she rang she was told it was because she had “retweeted something from ken loach at some point”.

    In the Independent, comedian Mark Steel says: “It’s a standard thing that clearly goes out to everyone. It says there are two reasons [for rejection]. One is that you don’t support the ideals and values of the Labour party. Or you are a member of a rival organisation”.

    “Rival organisation” is understandable, but this is obviously aimed at ‘lefties’. Weird.

Comments are closed.