Obama 91% Romney 9%
I published yesterday about my modeling showing Ricky Ponting’s last innings helps predicts the chance of Australia winning. Today I came across the New York Times “538″ blog which uses a prediction model to predict Obama currently has a 91% chance of winning the Presidential Elections. The reason it is so different from the polls which just report the percentage of people who say they will vote for one or other candidate is because in the US electoral system is all about winning the electoral college votes State by State. The “Chance of winning” model is put together by a bloke (Nate Silver) who rose to fame with his models to predict athlete’s performance. His predictions use polling data, plus a regression analysis, projections and multiple simulations (see here). Fascinating.
Tagged: Election, Modeling, New York Times, Obama, Prediction, Romney 
6 Responses to “Obama 91% Romney 9%”
Nothing wrong with Silver’s models, but I put more stock in the betting markets. Not that they’re perfect either. See here:
https://sciblogs.co.nz/thedismalscience/2012/11/05/documenting-the-arbitrage/
Eric Crampton has posted on the betting markets in this context: See https://sciblogs.co.nz/thedismalscience/2012/11/05/documenting-the-arbitrage/
Currently:
Intrade: Obama 68.3%
At Sports bet $1 on Obama will reap 1.22 if he wins, and $1 on Romney 5.4.
IPredict’s last trade was 92% for Obama
URLs
https://www.intrade.com/v4/markets/contract/?contractId=743474
http://sports.betfair.com/politics/market?id=1.21311313
https://www.ipredict.co.nz/app.php?do=contract_detail&contract=US.DEMOCRATIC12
Thanks Eric for the URLs
I know politics has a lot to do with science, but wouldn’t this be better on a maths site or political site. Doesn’t seem like science to me.
electickiwi – Kidney-punch is syndicated from $100Dialysis where I do not necessarily speak only of science. Having said that, I put this up because it uses prediction models similarly to what I use in my own research into Acute Kidney Injury. I posted on those yesterday using an example from cricket (because I can).
Exactly, John – statistics, & predictive models, have an awful lot to do with science.
Sry I take back what said, science appears in many forms.