# US Election

These graphs will be updated once a week, every Monday Afternoon/Tuesday Morning for week ending Sunday data . [Last Update - 8 September]. All data is derived from the** Intrade** US political markets.

*All charts are thumbnails – click on them to expand.*

**Projected Electoral College Votes by State + Expected Electoral College Votes**

This graphic charts the number of Electoral College votes given to the Democrats by the Intrade State markets as well as the Expected Value of the ECVs. The Expected Value is derived by multiplying the Intrade probability of the Democrats winning each state by the Electoral College Votes of each state, then summing those totals up.

**States holding Democrat: **

CA NY IL PA OH MI NJ MA WA MD MN WI CO CT IA OR NM HI ME NH RI DE DC VT NV VA

**Current Democrat win probabilities by State**

**Win Probability Change Over Past Week – By State**

This shows how the Democrat win probability for each state changed over the past week – an increase in the probability of the Democrats taking a State is shown as a positive on the left hand axis, a reduced chance as a negative and where States are sorted left to right on the bottom axis according to electoral votes (so the small population States start from the left and go through to the large population States on the right)

**Intrade Election Map**

Intrade probabilities for Democrats winning the Electoral College votes of each State as a map. Above 50% is shades of blue (representing a Democrat lead) , below 50% is shades of red (representing a Republican lead) – the darker the shade the higher the probability of victory. Each State is also given with it’s number of Electoral College Votes.

**Simulation of State by State Intrade probability data – 100,000 trials.**

This simulation adjusts for the non-independence of State markets as well as the dodgy behaviour of the probability tails and the fact that given Intrade probabilities are only approximately their true value at any given time. To see the methodology of the simulation – it’s described at the very bottom of the page.

This simulation gives us an estimate of how the electoral college vote is looking for the Democrats according to the Intrade State markets.

**Intrade Electoral College Simulations Over Time**

If we look at how the mean, median and mode of those weekly simulations has changed over time for the projected number of Democrat Electoral College votes we get:

**Simulated State by State Democrat win probability over time.**

If we use those simulations to look at how the probability of the Democrats getting at least 270 Electoral College votes has changed through time compared to the headline market of “Democrat as President”, we get:

The reason we are running State market based simulations is because of my suspicion that the collective information held by the State markets is superior to that held by the national market. The underlying reasoning is simple – the US election is effectively the combined result of 50 separate but interdependent, smaller, local elections.

The information that market participants in those state markets know about a given state election (as a proportion of the total amount of information in existence about that State election) is greater than the information that market participants in the national market know as a proportion of the total amount of information in existence about the entire country. The hypothesis is effectively that the knowledge gap in each state is less than the knowledge gap of all states combined into a national headline market.

Although it’s early days yet and we don’t have enough observations to empirically test the hypothesis at the moment, the early indications are good and the State by State market simulation appears to be a leading indicator of the national headline market.

**Simulation Methodology**

Simulations are easier than they sound.For instance, let’s take some hypothetical US State – we’ll call it Canada – and let’s say that Canada had an Intrade probability of a Democrat victory of 70%. If we generate a random number between 0 and 100 and then compare that random number to the 70% probability for Canada, if the random number is less than or equal to 70, we give the State’s Electoral College votes to the Democrats. If the random number is greater than 70 we give that State’s Electoral College Votes to the Republicans. If we did this 100 times, we’d expect to get results showing that around 70% of the time, Canada goes Democrat.

But we don’t just do that for any single State, we do it for every State, and then sum the total of the Electoral College Votes that the Dems could be expected to win. We then do it another 99,999 times and we get those histogram and CDF charts which tell us the probabilities of the Democrats getting any number of Electoral College Votes.

However, there are three key problems with Intrade probabilities.The first is that the further away from 50% a probability gets, the less likely is the chance that the implied Intrade probability is true. For instance, if we had 5 states all with a 20% probability of a Democrat victory – if an election were held tomorrow we’d expect one of those States to show up as a Democrat win. If we had 50 states with a 20% probability of a Democrat victory we could be highly certain that one of those States would show up as a Democrat victory. If we had 200 such States (yes, I know – there’s not 200 States in the Union but stick with me here), we could not only be virtually guaranteed that at least one of those States would be a Democrat victory, but we would expect that somewhere around 40 of those States would go Democrat – simply on the implied probabilities.

However, in all the Intrade political results I’ve looked at, literally hundreds including all of the Intrade Congressional, Senate and State Presidential winners, I’ve never seen a single example of a political party winning anything with a probability of less than 20%. So the Intrade tails aren’t a true representation of the real probability.

We can adjust for this by saying that any probability lower than 10% is actually 0%, and any probability higher than 90% is actually 100%. Hence, when we draw our random numbers, we don’t draw them between 0 and 100, we draw them between 10 and 90. As we approach the election, we expect the Intrade markets to gain more certainty, so we will reduce that draw to 20/80, then 30/70 as we approach E-Day.

Another problem is that the Intrade probabilities arent true (even at 50%), but are only approximately true at any given point because markets like Intrade are always in a state of flux, attempting to reach a hypothetical equilibrium that can never be achieved (because political events that drive these political markets are constantly occurring). This relates to another problem, a much larger problem, in that State results aren’t independent from each other.

If the Democrats lose New York for instance, they aren’t going to be winning Virginia or Colorado any time soon. In fact, if they lose New York they will probably have lost nearly every State in the process- so we need to adjust our simulation to accommodate the fact that nearly all States are dependent on the broader result happening across the entire country.

To solve these final two problems, and thanks to our readers here, and Caf particularly, for their exceptional assistance with this, we do a number of things.

For each iteration of the simulation we first generate a random number between 10 and 90 (to solve the first problem). Let’s call this number M.

Then we generate another random number for every State, but a random number from a normal probability distribution that has a mean of M and standard deviation of Q -we’ll call this number S.

If S is less than or equal to the State Intrade probability of a Democrat victory, we give that State and its Electoral College votes to the Dems, if the number S is greater than the Intrade probability we give the State’s Electoral College votes to the Republicans.

This solves the second problem on the one hand (as it simulates intrade probabilities being only approximately true) and allows us to solve the third problem on the other hand by making the random numbers that are drawn to call each State dependent on each other by way of their sharing a common mean value and standard deviation of the probability distributions they are drawn from.

Using the Intrade election data from 2004 to calibrate our system, having a standard deviation of 1 (which is a lovely and convenient round number by any yardstick) gave us the most accurate results – so we’ll use 1 as the value of Q for our simulation this year.

After each State has been called by their own probability distributions based on one value of S (with mean M and standard deviation Q), we sum up the total number of Electoral College votes for the Democrats and that gives us one simulated election result. We then run another iteration, where another value of M is drawn and where each State then draws another random number from their own probability distributions (of new mean value M and standard distribution Q), to determine whether each State goes Democrat or Republican. We add the ECV’s up and this gives us another simulated election result.

We do this 100 000 times and arrive at our final weekly simulation results.

## blindoptimist said

I love a bit of data with my cereal. Top job Possum. Long may you calculate!

## Kirribilli Removals said

But is Obama wearing a flag pin?

## Enemy Combatant said

Once The Kid flicks Brutusina from his coat-tails he’ll kick up over 300 on the Dem EV Probability Sum. Charted, it will be a work of psephological beauty.

Fantastic analysis on the Bludger the other day,PC. Glad we’ll have a place to hang if BillBo shuts up shop on the gravy stroke of the Sep campaign, as he’s threatened to do if the W.A. election is called in October:(

So I guess we won’t be entirely cast aside like orphans in cyberspace, if we can get some serious action here:)

## Catrina said

After going though this material I have a request (a.k.a. user requirement). While the graphs give us an indication of the general trend and probable Democrat victory, what is missing is the degree of sensitivity of the the position. For example, if we were to impose a +-3% variation in the numbers, and if the Democrats or Republicans are only just getting some of the states by the skin of their teeth, the number of electoral votes can jump about in a non-linear fashion. This brought me to the idea that the display of a band with an upper +3 and a lower -3 would provide a more in-depth picture in which we could get a sense of the potential volatility.

## Possum Comitatus said

Good idea Cat – I’ll get onto building something that shows EV changes with certain price movements.

## Possum Comitatus said

Is that the sort of thing you were after Cat?

## Kirribilli Removals said

Poss, you could fix the web link to The Randies, if you’d like more people to hear them.

## Possum Comitatus said

Thanks KR – all fixed, and changed as well.

## Catrina said

I think we are talking about the same thing (maybe). The thing I do know is the states vote, and sometimes when a state votes it can be very close (e.g. Florida back in 2000). Now – if according to some numbers coming out of Intrade, we has lets say 4 states perilously on the edge – then a scenario onfolds in which a large number of electoral votes disappear due top a small change resulting in a state switching to the other side. It’s this sensitivity that I’m talking about.

I figure if we played with some offsets from the price, that should quickly identify and to some extent qualify those states that are the swing states, the the overall swing volatility at any given time.

That make sense?

## Possum Comitatus said

Cat, Sure does.

I’ve added that graph that shows each state, where the state sits on the Democrat win probability line and the EV votes involved as a really primitive way to do to it for the moment.

I’ll have to think about how to build some given level of sensitivity into a graphical display that doesn’t make ones head explode when they view it. I tried a few earlier today and ended up going WTF!?! about the lot of them – way to complicated. Coloured confidence interval areas overlaying cumulative EV lines running along the probability axis at the bottom and stuff – by the end of it I had know idea what I was talking about! :mrgreen:

But I’ll definitely give it a lot more thought

## The Battle Is On « InsideOut said

[...] http://possumcomitatus.wordpress.com/intrade-data/ http://www.openleft.com/tag.do?tag=Presidential%20Forecast http://www.270towin.com/ [...]

## Andos said

Those probability simulations are pretty stark, for a target of 270 ECVs.

A 90% probability of getting 270 or over on the June 15 results (from the reverse cumulative frequency)… I am reaing that correctly, aren’t I?

## josh lyman said

Andos’ point raises a question – why the huge discrepancy between the second graph (probability of Obama victory = 64% on June 15) and the last one (90%)?

The former looks a lot more realistic. This may be Obama’s to lose, but there’s no way McCain has only a 10% chance this far out.

## Possum Comitatus said

It’s just because of where the State probabilities sitting. A lot of Democrat States with relatively large numbers of electoral college votes are above 70% probability. With only 270 to get, there’s around 330 or so in play for the Dems, but for the Reps it’s really only 280 thereabouts in play – hence the huge probability for the Democrats.

The question is why the huge difference? Headline figure traders listen to Fox News? Better info in the smaller State by State markets? The markets are out of whack – so it will be interesting to see which one moves over time to match the other.

## Possum Comitatus said

Actually Josh, thinking about it a bit more – it’s hard to see any of the State markets that have odds which might be considered unreasonable at the moment when it comes to those States that the Democrats are favourites in.

If we were to assume that the headline market for a Democrat Presidency is on the mark, then some of those Democrat States odds would need to be seriously reduced (and some of the closer Rep States would need to firm). That get’s us on to the questions of which ones and by how much?

The problem I’m having is actually identifying those States.

Which makes me wonder if it’s actually the headline market that’s out of whack – but probably not by much as that particular market is really just treating uncertainty differently to the State by State markets?

## Andos said

It is certainly going to be interesting to watch the graphs’ progression. Keep up the excellent work, Possum; it is greatly appreciated.

## josh lyman said

I think I follow what you are saying Poss – your state-by-state column graph has Ohio at 62% Dem (and thus theoretically well out of reach, along with the White House).

Interesting by contrast that Florida – won/lost by 500 votes 8 years ago (after various acts of theft and a Supreme Court version of a coup, etc etc) is now down to 29% – are all the Dem retirees the ones who will switch to McCain now that Obama is the Dem?

Still, 90% is clearly an inaccurate forecast, no matter how it was generated. No sane pundit would put their name to that.

## Possum Comitatus said

It makes you wonder.What if Obama wins – can we then look back and say that it was inaccurate, or even over egged?

That’s the tricky bit with these sorts of things.

The thing about the 90% mark coming from Intrade is that they might not be sane, but thousands of people have put their name to it by betting on the election. That is, after all, where the probabilities are coming from.

Normally with betting markets on politics, it’s usually the headline market – the pick the overall winner market – that leaps ahead, and then the provincial markets like the State by State (or seat by seat) slowly play catch up to the headline act. But in the US at the moment it’s working in reverse. The Dems have been ahead for a long time in the State by State markets and it’s the headline market that is slowly catching up.

On any given day, without looking at the data, were I asked which market would I have more faith in predicting the winner, the headline or the State aggregate? I’d chose the state aggregate every time.

It’s just now that the State aggregate is saying such extraordinary things that makes me get nervous about it.

There’s no statistically rational reason to doubt the probabilities – just that it’s saying something we aren’t used to.

Yet, on the basis of the numbers alone, and ignoring my “nah, it cant be true but , well, it just CANT be true” feeling – the 90% probability of a Dem victory actually looks pretty realistic.

## josh lyman said

Then I suppose we should all put money on Obama at 63%, because from what you’re saying it will keep going up to match the state results (barring a calamity, of course), so it’s actually a very god price.

## Possum Comitatus said

I might build a separate series for next week that measures the weekly mean probability of the simulations starting April 1st and chart it against the weekly “dems to win presidency” market probability. Then we can see which series will move to meet the other between now and November.

## Possum Comitatus said

That didnt really come out right – I’ll measure the probability of the Dems getting 270 electoral college votes or greater from the sims going back to April and then chart THAT against the headline market probability

## feral sparrowhawk said

Possum, this seems a too obvious point for you to have missed, but it looks to me like the discrepancy between the two graphs is because you’re treating the individual states as independent events.

Clearly they’re not. If someone thinks that there is a 32% chance McCain will win Ohio probably half of that is the possibility of some meltdown in the Obama campaign (eg a new Jeremiah Wright debacle). This would ripple nationally and knock him out in most of the borderline states.

I’ve a very shaky understanding of the maths you’re using, but if you’ve allowed for this I can’t see where.

## Possum Comitatus said

Sparrow,

Independence is a tricky one in these multiple seat prediction markets. On the one hand, we need to ask “are the individual markets actually independent?”, yet on the other, if we agree with the very basic assumption of prediction markets that they exhibit somewhere around a semi-strong efficiency in the whole efficient market hypothesis spectrum, then considering the enormous amounts of polling information and publically available analysis being done, it isn’t a very big leap to suggest that any serial correlation has already been adjusted for in the individual state market prices.

I’m a little stuck on this question regarding independence. At the moment I’m willing to give the individual markets the benefit of the doubt – but it may not actually be correct. But we dont really have any way to either measure independence properly in the context of political support in the individual states, nor adjust for it if we were to find it.

I’d be interested in anyones thoughts one this?

## David Gould said

I think the difference is simply a function of how people feel. People think that the election as a whole will be close, but that such-and-such a state is in the bag. If they do not work through each and every state, they cannot see the discrepancy in their own thinking.

I think that it is going to be a closer election than people think. If you examine the polling, there are five or six states on margins of less than five per cent that, if they all went to McCain, would see him slip across the line. Now, on raw numbers, the chances of McCain winning all of those states is pretty small. That is where the 90 per cent comes in. But, it nonetheless still gives him a ‘realistic’ – ie, possible – shot at winning.

## David Gould said

I guess what I am saying is that there is an identifiable path to the presidency for McCain on a state basis – holding Florida, and winning Ohio, Michigan and Virgina, plus one of either Nevada, Colorado, New Mexico or Missouri. (this of course assumes no other changes.) Such an outcome is plausible, and thus while the state polling is certainly reflected in the now 95 per cent confidence in an Obama victory, the path to victory is reflected in the closer outcome when people are thinking about the outcome as a whole.

## Catrina said

Possum

I love the new graph!

But given that I’ve just been thrown under the bus, I’m obliged to redirect my energies here. This is an early warning – I hope you don’t mind.

:-)

p.s. I’m assuming you going to update this page with the crunchy stuff …

http://possumcomitatus.wordpress.com/files/2008/06/forcatrinajune23.jpg

:-;

## Jimbob said

What I would like to see done is a validation of this against previous elections. This may not be possible, of course, the data might not be there, but it would be great to see what that simulation graph looked like 6 months out from the 2004 or 2000 election.

## Possum Comitatus said

I’m trying to hunt down a complete “week ending” intrade data set for the 2004 election. I’m slowly getting there so maybe soonish we might be able to do stuff like that. The big problem is that the further away from the election, the harder the data is to get.

## David Gould said

Possum,

I have been having a bit of a think on these numbers – the 95 per cent chance particularly. I played around on a site called http://www.270towin.com/ and looked at how they calculated percentage chance to win (using current polling data, I also came up with a 95 per cent change of Obama winning).

However, looking at how it is done it seems to me to be somehow inadequate. I am not sure if I can express this very well, but here goes:

Any particular opinion poll, assuming that it is done correctly, has a particular chance of having correctly gauged the mood of the public at the time the poll was taken. (and I know I am teaching you to suck eggs – I just want to put it in for my own thinking).

In the case of the US election, we have 50 races and we can use the probabilities for each state to work out the probabilities for the election as a whole.

However, as polls are not predictive, this ‘percentage chance’ of winning is the percentage chance that Obama would have won if the election had been held at a certain instant in time.

Now comes the hard part. Is there a way of modelling the chances of voters altering their opinions over time – for example, by looking at past data? Is this even theoretically possible or would it not be useful even it was? (I am not asking that you do this, by the way – it is just something that I have been thinking about).

## Andrew B said

Possum,

Do you take into account the , for want of a better word, “depth” of a price i.e. How much money is behind any particular price point on the market?

## jen said

Catrina –

there you are! (excuse me Poss for borrowing yr site)

Would like you to revisit PB, but gird your loins, girl.

It’s part of the fun.

Nose clips in the mail.

Cheers,

Jen.

## Possum Comitatus said

Andrew – not at the moment, but I’d be certainly open to suggestions on a workable way to do that. Over a the http://possumcomitatus.wordpress.com/2008/06/26/request-for-mental-assistance-on-us-election-intrade-stuff/

…post I’ve asked about that – so any insights or suggestions would be greatly appreciated.

DG – what you’re after is basically the holy grail of polling analysis! Never let it be said that you don’t set your sights high! :mrgreen:

We can take a whole series of polling snapshots and try to model them – and we can get as sophisticated as we want with that, but the problem always comes back that people will often change their minds because of political events, and those political events cant really be modelled in terms of the probability of their likelihood of occurring in the future.

So the best we can really do is model a whole heap of different things and look at how they are changing, and then try to bring that altogether in some way.

The best model though is one where there is incredible consistency in the polling – like Rudd had last year. For then we know that even though there’s a possibility of something happening that will change the vote, the incredible consistency of the polled vote level gave us greater certainty that the results would be less likely to be dramatically different from the long term polling trend than would be the case if the polls were actually all over the place in the lead up to the election.

## David Gould said

Difficult, I can see. I have often wondered about the impact of political events on voting patterns. We can see things like the budget at times altering voting intentions (although often simply within the margin of error of the two polls being compared). But over time, is there a tendency for a kind of regression to the mean?

I remember looking at some stuff that you ran last year (I cannot remember specifically what it was) examining changes in polling after events, and there seemed to be a delay in any effects, and then a gradual washing out of the effect of any particular event.

Couldn’t political events be modelled in the change probability in any case?

Anyway, this is just my crazy thinking at the moment – ignore it.

## feral sparrowhawk said

Sorry Poss, but I don’t think these things are remotely independent. I’m not sure what, if anything, that says about strong market theory, but they’re just not.

Barring something relevant on the VP front I’ll bet at almost any odds that Obama will not win Missouri while losing Ohio. Won’t happen. He might win both, lose both or win Ohio while losing Missouri, but the other is just not going to happen (unless McCain takes an Ohian VP or Obama takes one with a Missouri connection). But if you treat them as independent events there is an 18% chance of that happening and if I’ve understood correctly that’s factored into your probability of him winning.

Some other pairs of states are more independent because they are demographically more different. The level of dependence between Colorado and Ohio would be fairly low I’d say. It’s different voters that will decide those. In Ohio its all about the working class socially conservative voters who backed Clinton in the primaries. In Colorado a big factor (although not the only one) will be seriously rich people who are gay or environmentally concerned or just hate racism and might vote for Obama against their economic interests.

But Pennsylvania, Ohio and Missouri are really pretty similar, with Pa having a slightly more democrat lean, followed by Ohio. Obama’s vote in Ohio will probably be 2% less than Pa and 3% more than Missouri, plus or minus 3%.

God know how you could model all this – but that’s your skill area not mine.

## Catrina said

I know – it’s not relevant to the scientific view on all of this .. but what the heck – here are a few talking heads talking about this subject.

http://www.msnbc.msn.com/id/22425001/vp/25402771#25402771

## Updates here and there. « Possums Pollytics said

[...] US Election [...]

## Kirribilli Removals said

Well, let me be the first to get the chairs and refreshments set up here, as BillBowe Bagins is kicking us out at the Bludgertorium and like a desperate bunch of addicts we’ll be heading to the nearest warm cozy and accommodating space to bitch and bicker and chuck around bits of information generally involving this year’s mega-spectacular US presidential election.

Possum, if we get too rowdy let us know, and we’ll find other digs, but in the meantime, get ready for the chattering horde! LOL

## Enemy Combatant said

“Well, let me be the first to get the chairs and refreshments….”

Hola Segundo, care to click the link at 36 and pass the freakin’ popcorn while you’re at it:)

## Diogenes said

KR and ecky

I’ll pop by to make sure you are behaving yourselves.

## Hari Thiru said

This is a fantastic site that provides a state by state analysis of the lection based on demographic and up-to date polling data.

Highly recommended

http://fivethirtyeight.com

## Enemy Combatant said

Poss, broke my duck over at Poll Bludger 2008(US Edition) which as you know is a “splinter blog” from Billbo’s legendary site. “Dad” has been very supportive allowing Mumma Cat and her poll kittens to carry his name therby sparing us the ignominy of being derided as bastards in cyberspace. William has also given us a spiffy coloured Mt. Rushmore link. Perhaps a few of the Sep tragics who mosey on by thes parts here might enjoy a quick squiz.

And thanks for the plug space, PC:)

http://pollbludger.wordpress.com/2008/07/15/electoral-fraud-can-diebold-do-it-again/#comments

## Enemy Combatant said

Poss, remembering your “rollercoaster” after our last federal election here, check this out for Word’s Best Practice Geekism.

Linked today by ChrisB on PB-US(Election2008).

http://www.boston.com/news/politics/2008/specials/demographic_map/

## Possum Comitatus said

Ecky, that is just exceptionally spiffy!

## Kirribilli Removals said

That play map is the sort of thing Antony G does here, and they are such fun, eh?

I’ve just been looking at Intrade state prices on their site and noticed how steeply the odds have moved over this year on various states, and you have to say the punters are well ahead of the national polls in these markets. In fact it’s quite funny reading the national press, who seem to think it’s all ‘quite close’ and are somewhat perplexed by that.

Not too well informed, are they?

But we, on the other hand, have Possum! LOL

## Understanding polling, US-style « Not a Hedgehog said

[...] in mind when trying to get a feel for the Obama vs McCain match-up. And don’t forget that Possum Comitatus is keeping track of things as well. [...]

## A Few Charts. « Possums Pollytics said

[...] US Election [...]

## Catrina said

With 3 days to go before the Democratic National Convention, it’s going to be interesting to see what changes in the next two week cycles. Leading into the current numbers we have had Obama on vacation while the McCain Campaign has been pushing negative content but in the last couple of days things have been going the other way – McCain’s how-many-houses drama combined with a media perspective that he’s pushing his POW card way too many times.

## Enemy Combatant said

Losing your “house effect”.

“How Pollsters affect Poll Results”

http://www.pollster.com/blogs/how_pollsters_affect_poll_resu.php

## kazak said

Do your figures allow for inaccuracies in voting machine

http://www.bradblog.com/?p=6309

Diebold has admitted that their tabulator software, known as GEMS, and used all across the country, in at least 34 states, does not count votes correctly.

## josh lyman said

Poss, I recall you talking a while back about the parallel realities in coverage of the US election – in one, Obama is going to win in a landslide, in the other it’s 50/50.

I guess we just found out which reality McCain thinks he’s in…

## US Election Update « Possums Pollytics said

[...] US Election [...]

## Jacqueline said

Thanks a ton for spending free time in order to publish Blinds To

Go “US Election Possums Pollytics”. Thank you yet again

-Bryce