If you didn’t know who Nate Silver was a few weeks ago, you’re likely familiar with him by now. The founder of fivethirtyeight.com, Silver is a statistician who has spent the past half decade making sense of political polling and using that polling data to make projections for political elections.
Using state polls, Silver correctly predicted the 2008 presidential election. He also nailed the 2010 mid-term elections and the Republican takeover of the House. Despite this impressive track record, a number of pundits question Silver’s projection for this year’s election. Over the past few months, Silver consistently had Obama’s chances of winning the election at 65+%. In mid-October, Silver had Obama had 75+% while, at the same time, Intrade had Obama at 63%. But talking heads in general, and Joe Scarborough in particular, said this election was a coin flip. A 50/50 proposition.
Silver responded by offering a bet to Scarborough. Scarborough didn’t bite. Of course, we now know that Silver would have won the bet handily as Obama rolled to a relatively easy electoral college victory.
But let’s look at Silver’s particular state by state projections. Now, keep ¬†in mind, Silver’s model relied on state by state polling. And as that polling changed, the model’s forecast would change. So, each week, Silver’s forecast for a particular state shifted based on new information. The day before the election, Silver gave Obama a 91% chance to win. And he provided a projection for each state.
The results: Silver’s day-before-the-election forecast correctly predicted the outcome of 50 out of 50 states (yes, we’re giving Florida to Obama even though the state has not been officially called.) That’s not too bad.
It’s worth noting, though, that only two states went for a different party this year than they did in the 2008 (so you would have gone 48 out of 50 if you just used the last electoral map). And if you had simply bet the betting favorite in every state, you would have won 49 of 50 states.
The only state that went to the underdog: Florida. Obama was +150 to win the state.
So did everyone miss a chance to make a million by simply betting on Silver’s predictions? Nope, not at all.
If you made 50 bets – one for each state – and laid $100 on the candidate Silver predicted would win, you would have won a whopping $394.50. Yep, you would have ‘risked’ $5,000 and netted $394.50 for a 7.8% return. Not bad, but not all that remarkable. And this goes to show that the betting market had largely priced in Silver’s projections.