Of course I know – I’ve been following the 7-Eleven poll (blue cup – red cup purchase preference) and it is the best poll out there J. Predicting human behavior has long been the holy grail of prognosticators since the dawn of time and it takes on an almost mystical and magical hue during elections…..when you have polls being released almost every hour and then you have super polls and polls of polls and a composite of all the polls. Having said that, you always have people like David Brooks who say “ even experts with fancy computer models are terrible at predicting human behavior”. Shucks – I hope he’s wrong because I spend a lot of time studying and tracking polls during the election cycle(it’s the off season for golf so it keeps me occupied and out of trouble).
Let’s start with Nate Silver (considered to be one of the leading pollsters) and his model. In the 2008 election, he correctly predicted the winner of 49 states and is probably the most read and quoted pollster out there. He essentially combines a number of polls, weigh them according to a quality measure, adjusts based on other factors(employment numbers, stock market etc.), sprinkles some secret pixie dust and then runs thousands and thousands of simulations (Monte Carlo) and voila! We all know that Monte Carlo simulation has been around for a while and is the same methodology used by banks for their subprime strategies- and we know how that ended up. Brooks cites a study by Duke University involving 11,600 forecasts by CFOs about the S&P index and came up with zero correlation. I’m a big fan of Nate Silver who comes from the world of sabermetrics(do you remember Brad Pitt in Moneyball?).
You could look at the betting markets I suppose. And since the people participating in these markets have some personal skin in the game, they might tend to be far more accurate. There are sites like Betfair, Pinnacle etc. which will more than happy to take your money. And then there is Intrade – a site which is highly touted as a very good predictor of events. You could actually make some money by playing the price arbitrage between these markets. And there was a decent opportunity last week when the odds were quite different between those markets. And Intrade is actually cited by pundits and pollsters alike and a few of the poll of polls actually use Intrade as one of their inputs. Of course, if you wanted to, you could actually influence the odds at Intrade by making a huge purchase and therefore influence the polls – politics is not a nice game apparently.
I think generally speaking, polls probably represent a snap shot of where things are today and betting markets are probably more predictive in nature – if they are not susceptible to manipulation. Of course, if I knew exactly how they work, I would have retired by now. Polls do serve a critical function during these cycles but need to be looked at with many grains of salt. Understanding their methodology is critical before assigning any credibility – or betting any money. Obviously, I’m not advocating betting on the elections but if you were planning to – perhaps you can join me for a cup of coffee at 7 Eleven? Would you like a blue cup or a red cup?