Nate Silver’s FiveThirtyEight general election model was released today. Depending on the model, he gives Hillary Clinton between 73% and 80% chance of winning in November.
I was curious to see if Silver had so much clout that his prediction would move betting markets. Despite all the attention, the answer is no:
The lack of movement could be because the odds were already closely aligned with Silver’s predictions. Electionbettingodds.com, where I drew the graphic from, puts Hillary Clinton at 74% chance to win to Trump’s 22%. That is between the probabilities of the two FiveThirtyEight models.
But even if Silver’s models didn’t add any new information for bettors, there’s still reason they could move the market. In a way, Silver’s prediction* is an event in and of itself. It’s extensively covered by the media, often with references to Silver’s near perfect record of forecasting the 2008 and 2012 elections. Rather than predicting who wins, the FiveThirtyEight model could actually change who people plan to vote for if it influences their perceptions of who could win.
Political Scientists have demonstrated in numerous experiments that voters’ perceptions of who they think will win influences thier choice. People like to belong to the winning team. Giving Clinton an 80% chance of winning in November could create a self fulfilling prophecy, unintentionally influencing undecided voters and tipping the election toward Hillary.
That being said, Silver’s predictions weren’t exactly a surprise. Bettors could be considering the “Silver effect,” but already priced it in.
At the very least it seems that Silver didn’t include this possibility in his model. Maybe I’ll email him.
*It’s not really a “prediction” which is binary, but a probability based model. But go tell that to all the third party news outlets covering it.