Final weekly Intrade roundup
23 weeks ago we started tracking Intrade on a weekly basis as our measure of how the US Election was playing out where at the time the probability for a Democrat victory was running at 62% with 293 projected Electoral College votes for Obama. Today we come to our final Intrade Monday where the headline market gives an 88.5% probability of victory, our State market simulations have now climbed to a 99.977% win probability and the projected Electoral College votes are sitting around the 353 mark.
To start with will have a squiz at the daily trackers:
Obama has started to come of his high 1 to 2 weeks ago, consolidating around the 350/360 EV mark. The simulation probabilities have continued to run toward the Dems as the state probabilities have polarised out of the 30/70% threshold of contestability, while the Intrade headline Party markets have started to ease off as the probability approaches 90% (although over the next few days that will certainly change as the headline Party markets start closing in on 100% come Election day).
If we look at the Democrat win probabilities by State and how they’ve changed over the past week we get:
Montana, West Virginia, Indiana and Missouri moved strongly Republican over the last 7 days as the strong Democrat polling results of a few weeks ago have failed to be repeated. Interestingly North Dakota moved towards the Republicans and currently sits on 29% for the Dems – even though the Pollster.com LOESS regression polling trend line has Obama leading McCain by a few points in the State.
Over the next 2 days, massive amounts of polling will be released – so expect to see some movement on Intrade if ND, MO, WV, IN or GA start looking good for the Dems.
Our simulations have tightened up so far that there isn’t much potential movement left barring some massive change in market sentiment:
If we focus on just the EV allocations of these sims we see the mean, median and mode all coalescing around the 350 mark.
To sum it all up, even though some states have moved back quite strongly toward the Republicans over the last week, dragging down the projected number of Democrat Electoral College votes in the process – the probability of the Democrats taking the Whitehouse has slightly increased as the time left for McCain to find a game changer has all but run out.
Please login below to comment, OR simply register here :
Thank you for registering, we have just sent you a confirmation email, which includes your new password to be entered below.
“the time left for McCain to find a game changer has all but run out”
It’s a shame really. You would’ve thought they could’ve come up with some last minute desperate smear. Instead they had to embarass themselves with that SNL sketch!
Why is the simulation predicting 353 when several of these states (I’m thinking specifically MO and IN here) haven’t dropped by big margins? These two weren’t over the 70% cutoff anyway, so why should a change in win probability that’s still in the central range make such a big difference to the model? If it’s within say 45-55 surely they’re too close to call. I’m wondering whether we were counting these states as DEM in the sim when we shouldn’t have.
Also, where did requestathon go? Seems to have done a nelson and sank without a trace…
Dan, MO was sitting above 50% for a fair while and Indiana was closer than it is now. So in the simulations, an extra 11 votes turned up over 60% of the time (caused by either MO falling Dem a majority of times in the sim and a few times when MO was called for the Repubs but Indiana went Dem).
Now that MO and IN have pulled back, 11 votes (from either MO or IN) get called for the Dems less than 50% of the time, hence the change from 364 where Intrade has been sitting for a while down to 553. Last week IN and MO were both called for the Dems a majority of the time making last weeks EV 375 – but that was a bit of a blip with MO and IN just happening to peak at above 50% for a day when the data was collected at the close of trade Sunday Intrade time.
Requestathon is coming – but you bastards asked such tricky questions I’m gallivanting around the place trying to hunt down the data to answer them!
Surely one of you could have asked something I already had sitting in my database!
But no – you lot had to go and get all creative and tricky and obscure!
Ltep – I wanted my October Surprise too!
That’s pretty close to our forecast. We’ve got Obama at 364 (http://www.hubdub.com/election_map) with Indiana too close to call.
It is interesting how the markets seem to have factored in the potential for an October surprise. Each day that goes past without a surprise the markets move another half a percent to Obama. Maybe the surprise this year was that there is no surprise?!
This shows what intrade is saying today but what about tomorrow? I mean what I want to know is, in 2004, how accurate intrade was over the lifetime of the contract? And where was it a day or two from the election? I have been unable to find this data anywhere.
Captain, try
http://geekmedia.org/tradesports/
The site seems to be having a temporary conniption at the moment, but should be back up soon.
It has a database of 2004 intrade prices for the pres election.
3 Trackbacks
Please login below to comment, OR simply register here :
Thank you for registering, we have just sent you a confirmation email, which includes your new password to be entered below.