If we all close our eyes, focus our minds and pretend really hard – all my moving boxes might miraculously unpack themselves and it might also seem like Monday.
The belated penultimate weekly roundup of the Intrade figures is pretty much more of the same, with once contested states continuing to push out in their probabilities to make them close to Democrat certainties, and a few new Republican states moving into the contested battleground. The key feature this week is how the probabilities have changed over the last 7 days:
The ordinarily Republican states of Montana, Arizona, Indiana and Ohio all experienced double digit probability movements to Obama, while the 5 to 10% improvement range were filled with more Republican states like Nevada, Georgia, North Carolina and North Dakota with Iowa thrown in for good measure.
The State simulations have reached the 99% plus probability estimate for a Democrat victory as the State probabilities continue to consolidate for Obama. If we separate Electoral votes into bins 10% wide and look at the Democrat probabilities using today’s data we get:
The Dems have 234 EV’s in the 90%+ range, 57 in the 80-90% range, 20 with a 7 in front, 27 with a 6 in front and 26 with a 5 in front. At the moment, Pennsylvania is the State that delivers the 270th Electoral College vote (giving Obama 273) and it sits on an 86% probability of a Democrat victory. With such consolidated probabilities for the Democrats in so many states, we end up with extremely tight simulation results.
If we take today’s sim (which is only 0.004% different from week ending Sunday’s data) and turn the simulation result into a cumulative density function, we can see the Intrade probabilities of the size of the Democrat victory for any number of Electoral College votes:
Intrade doesn’t see much of a chance of breaching the 400 EV mark – about a 3% chance. Similarly, the 376 EV Obama ceiling we were looking at a few weeks ago (which in this sim is actually 375 because there isn’t an Intrade market on splitting the Nebraska votes) has the Dems sitting on an approximate 22% chance of hitting that ceiling let alone breaking it.
If we look at the daily tracking charts, while the probabilities of victory have continues to increase, there’s been a fair bit of consistency over the last week or two in terms of projected EV’s for Obama
As always, the rest of the results from the weekly Intrade roundup, including the simulation methodology, can be seen over at the dedicated US Election page.
5 Comments
Possum, the Electoral Vote Probabilities for Democrat win -by State image on the US Election page has the state labels mixed up. For example, New York is listed as the Democrats most likely win with 4 EVs, while MD (Maryland) sits on New York’s bar with 31 EVs.
Also, since your model has more than 270 EVs at over a 74% chance of victory for the Democrats (with 74% being the high end of the draw width on Monday), shouldn’t your model give a 100% chance of victory for Obama?
Ta Ghost – that was my ‘add label’ macro not working well with updated data.. all fixed.
On the sims, when we draw our first random number (currently between 26 and 74), that number becomes the mean draw number for each state with a standard deviation of 1. It’s that second draw that we use to call against the Intrade probabilities each iteration (which let’s us approximate the issue of state non-independence).
So we’ll still regularly get 73’s and 72’s and even 71’s drawn at the moment to call the individual states with, which is how Obama is staying barely below 100%.
Interesting analysis. We currently are forecasting Obama to pick up 375 EV’s with Missouri and Indiana recently walling into his camp. Georgia and even McCain’s home state, Arizona, are now trading at about 20% likelihood to go Democrat. If Georgia falls to the Democrats then you will be able to walk from Florida to Canada without leaving the Blue states!
Here’s the map: http://www.hubdub.com/election_map
Nigel, are you guys running anything on the individual Nebraska districts?
Obama is looking good in NE-2 (the second congressional district to give him 1 EV there), but no prediction market is running the the individual districts of Maine or Nebraska that I can find.
Hi Poss. Long time reader, first time writer (as the saying goes)….
There seems to be an extraordinary amount of handwringing about polling data on the US election despite significant leads for Obama in poll after poll. It takes me back to your coverage of the AU Federal election – with your analysis and the polls being pretty much dead on the money. Is there something remarkably odd in the US approach to polling and the voluntary voting concept, or is it just the media trying to keep the game in motion and build the drama? Apologies if this has been covered before.