2010 Congressional Elections - Polls & Prediction Markets

[MattusMaximus]

Okay, sooner or later someone had to start a thread on this, so here it is. Personally, I prefer prediction markets over polls any day (I'll explain why if you're curious), and one of the oldest & most respected prediction markets has recently opened up on the question of the 2010 Congressional elections.

As of today, here's what the Iowa Electronic Markets have to say...

DH_DS10 75.7%
DH_RS10 1.5%
RH_DS10 22.0%
RH_RS10 2.6%

DH_DS means "Democrats control both the House & Senate"
DH_RS means "Democrats control the House and Republicans control the Senate"
etc.

So, looking at this now (which is very very early), it seems the Dems are in good shape to hold both houses of Congress. Now whether or not that means they hold on strongly or just barely is an open question.

Btw, the specific breakdown on the House is...

DH.gain10 2.0%
DH.hold10 74.4%
DH.lose10 24.0%

Any other polls, prediction markets anyone wants to post?

 

[corplinx]

I predict some sociopaths might be replaced by other sociopaths. Many sociopaths will keep their current seats, because they are sociopaths.

 

[gtc]

DH.gain10 2.0%
DH.hold10 74.4%
DH.lose10 24.0%

That adds up to 100.4%. Does the 10 mean anything there?

 

[MattusMaximus]

That adds up to 100.4%. Does the 10 mean anything there?

Having numbers add up to more/less than 100% isn't uncommon in a prediction market (I'm not sure why). I've seen it before, and I've never figured it out.

The 10 stands for 2010.

 

[Brainster]

Having numbers add up to more/less than 100% isn't uncommon in a prediction market (I'm not sure why). I've seen it before, and I've never figured it out.

The 10 stands for 2010.

Because they are prices, not percentages. Effectively the same thing, of course, but all it reflects is the imperfection of the market.

 

[Dr Adequate]

Having numbers add up to more/less than 100% isn't uncommon in a prediction market (I'm not sure why). I've seen it before, and I've never figured it out.

I'm not really familiar with predictions markets, but my guess is that what should hold it at 100% is that if it doesn't add up to 100% there ought to be some way to make money out of the situation. But the collective wisdom of the market isn't going to get it right on the nose, rather it'll fluctuate tightly around 100%.

I guess I should look up how they work, eh?

 

[Brainster]

InTrade has the GOP takeover of the House at 35% right now, so if you want to make some dough, buy the Democrats at InTrade and the GOP at IEM.

 

[Dr Adequate]

I'm not really familiar with predictions markets, but my guess is that what should hold it at 100% is that if it doesn't add up to 100% there ought to be some way to make money out of the situation. But the collective wisdom of the market isn't going to get it right on the nose, rather it'll fluctuate tightly around 100%.

I guess I should look up how they work, eh?

Ah yes, here we are. From the Wikipedia article on the IEM:

It is possible that a given market is simply irrationally priced. If the price of all the different shares is greater than or less than $1, then there is at least one share which is not correctly priced and so can either be shorted or gone long on.​

Doing so brings the market closer to 100%.

 

[Francesca R]

It is possible that there's an anomaly in the prices but not very likely. That wikipedia quote is giving the least likely explanation.

You can't typically arbitrage the contracts because they have bid-offer spreads that remove the ability to get in and out with a certain profit. All the bids will usually sum to less than 100 but all the offers will exceed it. You deal on the "worst" of the two.

The market maker (who therefore gets the "best" side of the price) can make a profit with no position. That's their economic return for providing the ability to bet. If it is too generous (in their favour), somebody else will step in quoting narrower spreads and will win all the punters' flows.

 

[Dr Adequate]

It is possible that there's an anomaly in the prices but not very likely. That wikipedia quote is giving the least likely explanation.

You can't typically arbitrage the contracts because they have bid-offer spreads that remove the ability to get in and out with a certain profit. All the bids will usually sum to less than 100 but all the offers will exceed it. You deal on the "worst" of the two.

The market maker (who therefore gets the "best" side of the price) can make a profit with no position. That's their economic return for providing the ability to bet. If it is too generous (in their favour), somebody else will step in quoting narrower spreads and will win all the punters' flows.

I am having some difficulty in grasping your point. Please could you explain that again but less concisely.

Thanks.

 

[Francesca R]

I mean you can't do a series of trades that ends up at no bet, but nets you a gain (which may be intuitively Bleedin' Obvious)

If you could buy all four contracts for a total price of less than 100, you could be assured of a gain, because they are exhaustive*, and one of them will definitely finish in the money and pay out 100 (while the others pay out nothing). Unfortunately you buy at the "offer" prices, and if you add those up the sum is (should always be) more than 100, so it is a guaranteed risk-free loss. And if you want to sell all four, you'll get less than 100 for it, yet you know you will surely have to pay out 100 on one of them, so that's another guaranteed loss.

If there were identical contracts on two different exchanges with very different prices--say you could sell [DH-DS10] at 72 (the bid price) on IEM and buy it at 71 (the offer price) with InTrade at the same time--then that's also a riskless profit. Unfortunately you can't very often do that either because there are thousands of others looking for free money like that. (In fact the contracts on those two markets are not identical)

*well not quite, but I don't see any for "no control"

 

[MattusMaximus]

Here's the latest update from IEM...

DH_DS10 75.5%
DH_RS10 1.0%
RH_DS10 20.9%
RH_RS10 2.9%

Specific markets for the U.S. House and Senate...

DH.gain10 1.1%
DH.hold10 74.1%
DH.lose10 25.2%

DS.gain10 6.6%
DS.hold10 90.0%
DS.lose10 5.0%

Meanwhile, over at Intrade, it is projected at 66.6% that the Dems will hold the U.S. House in 2010.

At this stage, while I think the GOP will likely pick up some seats in both houses of Congress this coming November, all this talk of a "GOP takeover" is just puffed-up hyperbole. But that's politics for you

 

[MattusMaximus]

Today's IEM update...

DH_DS10 75.4%
DH_RS10 0.9%
RH_DS10 20.9%
RH_RS10 2.9%

Specific markets for the U.S. House and Senate...

DH.gain10 1.2%
DH.hold10 74.1%
DH.lose10 24.4%

DS.gain10 6.4%
DS.hold10 90.0%
DS.lose10 5.0%

Intrade projects a 66.5% chance that the Dems will hold the U.S. House in 2010.

Are there any other prediction markets and/or polls of interest?

 

[Francesca R]

Maybe people can self-satisfy their interest concerning what the prices are (and whether they have moved by a tick) by visiting the pages themselves.

 

[MattusMaximus]

Today's IEM update...

DH_DS10 74.6%
DH_RS10 0.7%
RH_DS10 22.0%
RH_RS10 8.0%

Specific markets for the U.S. House and Senate...

DH.gain10 1.1%
DH.hold10 74.5%
DH.lose10 32.5%

DS.gain10 4.0%
DS.hold10 90.1%
DS.lose10 5.9%

Intrade projects a 65.9% chance that the Dems will hold the U.S. House in 2010.

I find this to be very interesting, what with all of the developments in the past couple of weeks (Sen. Dodd and Dorgan's retirements, the close MA Senate race, retirements of many House incumbents, etc), the markets are surprisingly stable and unchanged. The only exception to this is a slight uptick in the odds the Republicans will take back the House.

 

[Brainster]

There are two main factors at work here; the underlying situation and the amount of time remaining. Right now the time remaining is the main thing that the Democrats have in their favor; well, that and the fact that they have to lose something like 40 seats in order to lose the House. As I said in the Ted Kennedy's seat thread, apparently the market is currently discounting the idea that Brown's apparent close race there reflects broader discontent. It could also be simply a reflection that the Democrats have time to recover; a few solid employment results would do wonders.

Barring that or terrible unemployment numbers, don't expect big movements in this market until September or so. And even on the day of the election, you can have monstrous swings both ways; check out the IEM Presidential Markets for election day in 2000 and 2004 for examples of how wild it can get.

Part of it is that "all or nothing" aspect of the market. This isn't like buying stock in a stable company where you know your upside and your downside is limited, and you hope it will perform at the upper end of the band. The day after the election, the people who invested in one party will get nothing and the people who invested in the other party will get a tidy profit. That is part of the reason why these markets work fairly well provided they are not too narrow; greed and fear can work wonders.

 

[MattusMaximus]

There are two main factors at work here; the underlying situation and the amount of time remaining. Right now the time remaining is the main thing that the Democrats have in their favor; well, that and the fact that they have to lose something like 40 seats in order to lose the House. As I said in the Ted Kennedy's seat thread, apparently the market is currently discounting the idea that Brown's apparent close race there reflects broader discontent. It could also be simply a reflection that the Democrats have time to recover; a few solid employment results would do wonders.

Yes, any positive job numbers in the next few months will make the GOP's work much harder for them. In addition, if Obama and the Dems can get some popular legislation (re: jobs programs and Wall Street taxation, etc) passed, that could help shore up support for them.

The Dems are really eager to get health care reform over and done with - it's really weighing them down and preventing them from getting to other things. Not good.

Barring that or terrible unemployment numbers, don't expect big movements in this market until September or so. And even on the day of the election, you can have monstrous swings both ways; check out the IEM Presidential Markets for election day in 2000 and 2004 for examples of how wild it can get.

Part of it is that "all or nothing" aspect of the market. This isn't like buying stock in a stable company where you know your upside and your downside is limited, and you hope it will perform at the upper end of the band. The day after the election, the people who invested in one party will get nothing and the people who invested in the other party will get a tidy profit. That is part of the reason why these markets work fairly well provided they are not too narrow; greed and fear can work wonders.

Good insights. Nice comparing notes with you again

 

[casebro]

I guess the predictive markets are supposed to be better than polls because having a couple dollars bet means that the subjects are better informed than the average poll subject?

But predictive markets are still only polls, correct?

I predict that as the predictive markets gain members, their predictive value will decrease to the point that they are no better than polls.

 

[JoeTheJuggler]

I guess the predictive markets are supposed to be better than polls because having a couple dollars bet means that the subjects are better informed than the average poll subject?

But predictive markets are still only polls, correct?

Not really.

Here's the difference as I see it:

When you read a poll that says 55% would vote for a given candidate, that's exactly what it means. If that's a good solid number (that is with a low margin of error--say + or - 2%), it means the probability of that candidate winning is very near 100%.

If a predictive market says 55% in favor of a certain outcome (as, for example, a given candidate winning), that outcome is very nearly a coin toss.

You can read the predictive markets as a comment on the probability of a certain outcome rather than a poll on how many favor a given outcome.

ETA: Now it'd be different if these predictive markets used a point spread as they do in football betting. For example, just prior to the 2008 presidential elections, polls showed Obama's lead as being maybe 5% (and this is oversimplifying things, because we still do presidential elections by delegates per state, etc.), but the predictive markets showed Obama to win at maybe 80% or more (I don't remember exactly what). But if the question wasn't about whether or not Obama would win, but that he would win by a certain number of votes, you'd change that. If you did it just right, you'd change it so that the bet would always be very close to a 50% probability of winning. (And of course buying shares isn't quite the same as making a bet, but it's closer than asking who you'd vote for in an opinion poll.)

 

[MattusMaximus]

I guess the predictive markets are supposed to be better than polls because having a couple dollars bet means that the subjects are better informed than the average poll subject?

But predictive markets are still only polls, correct?

I predict that as the predictive markets gain members, their predictive value will decrease to the point that they are no better than polls.

No, the big difference as I see it is this... a poll will ask who you will vote for or who you want to win. But a prediction market asks you to lay down your own money on who you really think is going to win. That little shift in psychology can make an interesting difference.