Princeton Election Consortium

A first draft of electoral history. Since 2004

Predictions: Comparisons with InTrade

September 24th, 2012, 3:00pm by Sam Wang


Yesterday, I framed predictions as a tool for planning for the future. Today let’s look at InTrade, a famous electronic market that is said to have good predictive powers.

I have maintained that in political races, InTrade is not a very efficient market for making use of polling data. Here are two (seemingly) mostly played-out examples and one market that is still unfolding.

The Senate. On August 29th, I published a calculation showing that Democrats/Independents would probably retain control of the Senate in 2013, with 65% probability. This corresponds to a true market value of 100-65=35% for the InTrade contract* “Republicans to control the Senate after the 2012 Congressional Elections.” Then, on September 19th, I recalculated the Democratic/Independent-control probability as 88% (Republican contract value 12%). The question of control is right on a knife’s edge, and the probability of the future event has moved.

Here is the InTrade price history. I have indicated the Princeton Election Consortium probabilities with red asterisks. The second one is out of range and therefore plotted below the axis. As you can see, PEC probabilities for this question led changes in the InTrade market by a substantial period.

The Presidency. Starting on August 3rd, I have given the Obama re-elect probability in a very narrow range, between 87% and 91%. The same algorithm would have given a similar probability as far back as June/July. It is relatively unchanging because re-election races move so little during a campaign season. Here is the comparison.

Volume in this contract has been low until recently. The last few weeks have seen sharp movement toward the PEC probability, a trend that seems likely to continue.

An obvious comparison with FiveThirtyEight could also be made. However, FiveThirtyEight is not an electronic market, so there is no arbitrage opportunity.

The House. Here is the third graph, including the 74% prediction.

As previously noted, this is a softer prediction because of uncertainties in using the generic ballot. It is notable because the direction of the outcome implied by the price (that is, >50% vs. <50%) is clearly opposite to the PEC probability. I will be watching this contract with great interest over the coming weeks.

There are several likely reasons for why InTrade does not reflect true probabilities: (1) Bidders overestimate both current and future uncertainty, even in cases like a Presidential re-election race where movement is small. (2) A few of these markets trade at low volume until Election Day approaches. The net outcome is that they start to become more predictive late in the season, and appear to be a lagging indicator of well-aggregated polls, and of course other events like Rep. Akin’s “legitimate rape” statement. Look for markets to catch up with polls by mid-October.

*Note that I am not using the Democratic-control contract, which will not count the votes of Independents Sanders (I-Vermont) and King (I-Maine). The contract requires 5o actual Democrats plus Obama re-election, or 51+ Democrats. The trading price is 59%, whereas the poll-based probability is currently lower, 39%. Considering that these are both in the 20-80% midrange, this contract is relatively well-priced.

Tags: 2012 Election · House · President · Senate

24 Comments so far ↓

  • Eric Fisher

    Six days have gone by. Your metamargin fell from a historic high near 6.2 to a record low of 1.7. I rest my case as an economist: markets aggregate information accurately and in a timely manner.

  • Eric Fisher

    Sam,

    Obama’s price dropped 10% today, 5% during the debate alone. Watch your metamargin in the next six days. Please do not pooh pooh economists or expert markets any longer.

    Eric

    • Sam Wang

      If those markets move, it would be a reason in favor of a giant pooh pooh. Measuring crowd reaction is not prediction.

  • Eric Fisher

    Sam,

    I write this post at 0819 on 2 October 2012. Obama’s price on Intrade peaked yesterday. The minor current decline is perhaps in anticipation of the debate tomorrow. Your metamargin is near its historical high. It will begin to decline soon. Whose tail wags which dog?

    Still your strong admirer,
    Eric

  • Brad S

    Tim,

    Thanks for your post & clarifications.

    >There’s a reason that Intrade is based offshore. My understanding is that other futures exchanges in the US have applied to sell political-based futures contracts, and have been denied the permit to do so.

    Very interesting, I wasn’t aware of this.

    >The question with regard to Intrade is whether we want to commoditize our political process..

    >For instance, let’s say that you’re Karl Rove, and you have some $200 M to spend on elections. With this money, you could potentially use some to do the following: a) manipulate the prices of political futures to your side’s benefit..

    With all due respect, do you realize that this not only sounds like a conspiracy theory, but a pretty foolish plan at that? Surely you must be familiar with the term “arbitrage”..? It’s actually a film now, starring Richard Gere. But let’s take your dark vision at face value & see where it leads us..

    Currently on Intrade, Romney to win is at 25%. For some reason I don’t quite understand, there are bids for ~10000 contracts, but only offers for a couple of thousand contracts. But anyway, assume Rove & Co. are determined to manipulate the market to their advantage. Going “Long Romney”, they could buy all 2000 contracts on immediate offer & push the price up to $2.74, at a guesstimated average cost of $12K. Let’s say they keep buying day after day, & push the price up to 50%, for let’s say $250K total.

    What are reasonable, predictable outcomes?

    First, isn’t it quite likely another player will come in & “short” Romney at such an artificially elevated price, after determining there was no fundamental reason for the “rally”, & drive the price right back down? There are plenty of rich speculators who aren’t die-hard Republicans. Rove & Co then “lose money” on their gambit.

    Rove is certainly smart enough to anticipate this, so we guess his calculation would be “What’s it worth to create a false & possibly short-lived sense of uncertainty in the race? Worth $250K? Who’s actually going to notice? Is that brief buzz worth more than $250K of real ad campaigns, candidate appearances, etc?”

    But secondly, wouldn’t Rove also have to be evaluating the risk of serious media blowback if & when it came out that Republicans were manipulating the InTrade market to make their candidate look more like a winner? Sounds pretty scandalous to me, anyway.

    >b) hedge in complicated ways against adverse political outcomes to recover money otherwise lost..

    This indeed sounds like classic commercial hedging, so arguably it should be allowable. Both parties could benefit, so why not? The key here is that hedgers have to be registered & report their trades & positions.

    >c) garner additional revenue from investing in races that are moving in your direction to fuel futher adverstizing spending and a further increase of the value of the future contract.

    I would refer you back to my cautionary “Investing 101″ comments. What you’re hypothesizing is someone thinking “We need X dollars to spend on advertising, so we’ll buy Y contracts on InTrade.” Let’s assume the Intrade market is big enough to make that a truly tempting option for the candidate & his backers. In this case, we’re talking about a vastly bigger InTrade contract market, with millions of shares traded daily on the candidate’s chances.

    It seems to me something’s missing from this picture. What’s going on in that future time that’s prompting 10s or 100s of millions of traded shares during the critical months of the campaign? Whatever that X-factor is, I think it’s likely to be as much if not more of a problem than the Intrade contract.

    IMHO, “Citizens United” is a far greater threat to what you call the integrity of US elections. Everyone truly concerned about undue influence should be pushing for a constitutional amendment clarifying that corporations are not people & do not enjoy free speech.

  • JohnJohn

    There is (almost certainly) no Romney lover intentionally skewing intrade.

    It’s certainly not insane to short the Obama contract at current prices. While most Intraders are risk – averse and tend to bet “all – or – nothing”, intending to hold till expiration, its not unreasonable to think a probabilistic thinker sees opportunity here.

    • Peter Principle

      LOL. Well, whomever it was — and whatever his/her/their motives — they appear to be gone now.

      And with Obama shares now above 77, and Romney under 23 (and poised to go quite a bit lower, judging by the order book) , I’d say their version of the Big Short didn’t work out too well for them — either financially or politically.

  • Amitabh Lath

    Peter, I believe you are correct about the “guardian angel”. The sums involved are really small for the attention it buys: “Intrade showing Romney keeping it close…”

    In the last week or so, this guardian angel(s?) got his clock cleaned. Intrade is still about 7 pts behind, so the crowd is probably going to punish this guy some more.

  • Brad S

    Tim,

    >gambling on political outcomes is a crime

    Just to share a little a bit more of where I was coming from in responding to you, I did sort of take your comments as a personal attack, b/c I’m a professional investor / speculator. So I internalized your words as “Brad, you are a criminal” & obviously that didn’t make me feel very good.

    I don’t want to get into a lengthy self-justification here, but I do feel confident that I’m not harming anyone by making wagers on political races. On the contrary, accepted mainstream economics posits that society benefits from the various “signals” that market-based prices give to people. So I do think it’s fair to put the burden of proof on you to convincingly articulate the criminal threat that my actions pose to the general good.

    >I began to contemplate how I could invest money in both Intrade and the Congressional races, using the expected Intrade gains to cover the political donations..

    For what it’s worth, this is a VERY risky idea & absolutely not recommended! I’m sure you can find agreement in any number of Trading 101 guides. NEVER let yourself take action based on the thinking “I want to spend X dollars on whatever so I’ll earn that via such-and-such trade / investment.”

    Also, more specifically, while Sam has an amazing methodology, he’s clarified on more than one occasion that his predictions aren’t 100% certain. I’ve never based my wagers solely on Sam’s calculations, but have always done my own independent research so that I could truly take responsibility for my choices. For this reason, I’ve chosen not to pursue the “Control of US House” contract, b/c I can’t handicap that on my own. I would be completely relying on Sam.

    I would urge all readers to likewise, “do your own thinking & be very sure of what you’re doing before risking real money on what’s going to happen in the future”

    • Tim in CA

      Brad,

      Thanks for the clarification. I should have been more clear in my post as well. I didn’t mean to imply that you or other individual investors are doing something criminal in using Intrade. Furthermore, it seems that I misspoke when I said that using Intrade to wager on political outcomes is a crime. I should have just said that the legality is unclear, and left it at that. There’s a reason that Intrade is based offshore. My understanding is that other futures exchanges in the US have applied to sell political-based futures contracts, and have been denied the permit to do so.

      My concern has less to do small-scale individual investors using Intrade than major SuperPACs and the like. My fear is that turning our political elections into a market commodity opens the door to a lot of potential funny business from major political actors. For instance, let’s say that you’re Karl Rove, and you have some $200 M to spend on elections. With this money, you could potentially use some to do the following: a) manipulate the prices of political futures to your side’s benefit, b) hedge in complicated ways against adverse political outcomes to recover money otherwise lost through adverstizing on lost races, c) garner additional revenue from investing in races that are moving in your direction to fuel futher adverstizing spending and a further increase of the value of the future contract. To these possibilities, we can add the prospect that politicians themselves could benefit economically from manipulating the election in various ways.

      I agree that, in general, investing in stocks and other public ally traded commodities can offer useful information and help make economic makets more efficient. The question with regard to Intrade is whether we want to commoditize our political process. Of course, 527s, Citizens United, and the like have already pushed our elections in this direction. Opening our elections to futures-based investing accelerates this trend. I would argue that our elections should be free of such influence and from the prospects for manipulation outlined above. In theory, elections are the process that allows citizens to step back, assess the the trajectory of our society and make a change if need be. The influence of Intrade seems to me to diminish this essential function of political elections, and to render them more like horse races. In my opinion, this not a positive developement for our political process.

  • Amitabh Lath

    Too many pundit-types talk about Intrade for me to feel comfortable with it. It is thinly traded enough that a small sum dropped at the right time could move the index and buy you a day or two of news releases. (Okay, not a small sum for academics….)

    But if I was some GOP moneybags, I wouldn’t think twice about dropping the cost of dinner-for-12 on the Romney side just to keep things from running away.

    Maybe the recent run on Intrade from 60/40 to the current 70/30 is due to some of these moneybags losing interest. It doesn’t exactly correlate with any particular news item.

    If you look at the Iowa Electronic Markets run by the University of Iowa Henry B. Tippie College of Business, you see (in their winner-take-all market)
    an 80/20 split. I have not nosed around their site a whole lot, I don’t know for instance how thinly traded these contracts are.

    But I trust it more because
    a) they don’t seem to garner much attention from cable news so why bother skewing them
    b) they’re academics, so being gamed would be more humiliating for them than being wrong ,
    (one would much rather a single result came out wrong than the entire experimental technique was shown to be non-viable) and
    c) I was born in Iowa.

    • Peter Principle

      If you’ve been watching the Romney contract recently, you would have seen that it appears to have a guardian angel — a player who appears, right on cue, whenever the news is bad and the price is falling, and places huge orders (huge for Intrade, anyway) at just below the most recent bid.

      These quotes typically disappear from the order book once the market cools down — although whether that’s because they’ve been filled or withdrawn, I don’t know, not being well versed on Intrade’s trading rules.

      In effect, this trader has been “stabilizing” the market, ala Goldman Sachs managing a new IPO. Which may explain why Romney/Obama contracts are trading about 7-8 points too high/low compared to the bookie odds and the Iowa Electronic Markets.

      What I can’t understand is why somebody isn’t doing to this guy was Soros did to the BoE in ’92 — i.e. milking the spread for all it’s worth. Then again maybe someone is but I can’t see it.

  • Olav Grinde

    I would think that the safest bet would be to buy shares (stock) in InTrade itself.

    If you add up the probability and payoff for the election resulting in a Republican controlled House with that of a Democratic controlled House, you do not reach 100 %.

    What is the payout percentage relative to the total money taken in? What is InTrade’s cut here? My guess is that it’s hardly insignificant.

    That’s what I would bet on if I was a gamblin’ man.

  • Terry

    Sam, I’m curious to know if your calculations take into account InTrade’s position on the likely Independent win in Maine? According to a recent clarification by Intrade the Independent seat held by Sanders in VT counts as a Democratic seat because he has declared himself to caucus with the Democrats but the Maine seat likely to be won by King will not count toward Senate control for either party unless King declares that he will caucus with a particular party before the election results are known. Thanks.

    • Sam Wang

      Yes, I did. That’s exactly why I chose the Republican-Senate market – to avoid this question. There are no Independents who express interest in caucusing with Republicans.

      InTrade’s recent ruling is a bit head-spinning. My reading of SENATE.DEM is that it specifically excludes Independents. Now they say that their combined-outcome market will count Sanders…but as you say, maybe not King. I think this would (currently) define SENATE.DEM as 50 Democrats-but-not-Sanders + VP Biden, and PRESIDENT+SENATE markets as 50 Democrats-plus-Sanders + VP Biden. Oy gevalt.

  • Brad S

    >It seems that the gambling on political events via offshore futures markets like Intrade is illegal, or at best the legality is disputed.

    This statement is almost certainly untrue, if intended categorically. It sounds like you’re likely confusing InTrade with offshore gambling sites.

    >It’s a bit disturbing to me that nobody seems to pay attention to the fact that Intrade appears to be, and indeed should be, illegal. Gambling is a vice, yes, but gambling on political outcomes is a crime.

    A crime, really? I’m not unbiased, having taken moderately sizable InTrade positions in the current race. Your moral argument isn’t convincing at all, however. You don’t cite any laws or legal theory & you don’t distinguish among different types of wagers. Are you anti-gambling in general? If not, how are political wagers worse that sports bets, speculative commodity trading, short-selling of stocks, non-hedge CDS, lottery tickets, etc etc?

    Why are you so worked up about this? You seem to be feeling very strong emotions here. If you don’t feel good about InTrade, don’t pursue it. But please don’t condemn others who want to express their opinions economically.

    On an emotional level, I understand where you’re coming from: for example, I personally feel it’s a disgrace to our country & to us as a people that so much collective money is “wasted” on political campaigns. Senator McCain & others have tried to make the tremendous level of spending “illegal”, & obviously that effort has failed spectacularly.

    Whenever politics become frustrating & disappointing, I feel it’s always helpful to remind oneself that ultimately those events are just amplifying & reflecting the human nature of some great number of individual people, who if you spoke to them individually about their politics, would likely be even more frustrating & disappointing! Hope that helps..

  • Tim in CA

    Hi Sam,

    Kudos to your post for Pres. Tilghman yesterday. Class of ’05 here. My freshman year coincided with her first year as well (fall 2001).

    One note on your intrade references. I looked into the matter based on your posts about a month ago. It seems that the gambling on political events via offshore futures markets like Intrade is illegal, or at best the legality is disputed.

    If you think about it for a moment, there are VERY GOOD reasons for gambling on political events to illegal. For instance, based on your other bit of advice to invest $ in close Senate & House races in this election cycle, I began to contemplate how I could invest money in both Intrade and the Congressional races, using the expected Intrade gains to cover the political donations. Then I stopped to think about it for a moment, and realized how corrosive such actions would be for the political system writ large. I am just a bit political player, moving negligible sums. What if some if these large Super PACs started playing such games? It would be extremely corrosive for the political process, and should ABSOLUTELY be illegal. Fortunately, Intrade doesn’t take investments from American banks, at least for the political contracts, for this reason.

    Intrade prices have become common currency in the world of political commentary. It’s a bit disturbing to me that nobody seems to pay attention to the fact that Intrade appears to be, and indeed should be, illegal. Gambling is a vice, yes, but gambling on political outcomes is a crime. You might want to at least mention this, when you discuss the possibility of profiting on your forecasts.

    Other than that, this posts clearly demonstrates the advantages of your meta-analysis.

  • Eric Fisher

    Sam, your readers ought to know that it is very difficult for an American to bet on Intrade. One opens an account using your passport, and then deposits money by a cashier’s check. An experts’ market is not meant to aggregate polls. Markets reflect information based upon the many decisions of those who transact in them. The classic reference is Hayek’s “The Use of Knowledge in Society”: http://www.econlib.org/library/Essays/hykKnw1.html

    The polls may contain systematic bias, perhaps because the median pollster is using an incorrect model of likely voter turnout in the sample stratification. Your own predictions were quite wrong in the 2010 Congressional races, not because you are a bad modeler, but because your models were fed wrong data. (Au contraire, you are a great modeler!) There may be a herding effect in how pollsters predict likely voter turnout, and that would wreak havoc with your scientifically model based prediction, Someone who has accurate (inside) information about likely voter turnout would start to transact on Intrade. That’s when its price might move, even when your model’s prediction does not budge.

    • Terry

      It is actually not difficult for Americans to use InTrade. All one needs to do is make a wire transfer or, as you mentioned, send a money order. As anti money laundering checks, InTrade now also requires one to upload to them, via their website as JPG or PDF files, a copy of either a drivers license or passport, a copy of a utility bill and a copy of a bank statement. The IRS has refused to answer a letter submitted directly to them by InTrade requesting an official position on the legality of using InTrade for US citizens so that remains unclear. For the record I am not affiliated with InTrade although I do use the site.

  • Peter D

    I jumped on the train a while back. There’s still room…

    So… latest MM is 4.66. tcdf(4.66/2.2,3) = 0.9378.

    Using 71.2%P/82.4%S/25%H for the Intrade prices and 93.8%P/88%S/84%H for the true probabilities give us Kelly Fractions of 78.4%P/31.8%S/65.3%H. Go ahead and use half of that – it still suggests you should skip your 401k contribution this month.

    Kelly has its problems (http://www.amazon.com/Fortunes-Formula-Scientific-Betting-Casinos/dp/0809046377), but bets this chunky don’t come around often…

  • Olav Grinde

    This is fascinating! I can’t see a single light-blue or pink state on your map anymore. And you have the highest spike at 347.

    New polls showing Obama is widening his leads in Colorado, Nevada, Iowa and Pennsylvania. And now more likely to win than lose North Carolina.

    Moreover, it seems the President is becoming competitive even in Missouri and Arizona.

    I don’t gamble, but I’m betting that election night will be a delightful relief. Unless Karl Rove has a surprise in store for all of us…

  • steve in Colorado

    I would love to make money on this election using your data, however, I am squeamish about any type of gambling. I was definitely tempted when you were showing 88% chance of Obama and Intrade was around 58% or so… But now that gap has narrowed and Intrade is at 71.6%. A lot less room for profit.

    • Sam Wang

      (1) Gambling is a vice. (2) You had two chances a month ago. No credit for admiring the train after it leaves the station.

    • The Political Omnivore

      I’m not a bettor–but PredictWise (betting market aggregator) is going to be unveiling a sort of “Fantasy politics” game where you can place non-real-money bets on various election outcomes, key states, etc.

      They showed it to me as a sneak-peak and, of course, I “bet” everywhere the polls were contrarian to the betting.

      When it goes live you might enjoy it.