US Presidential Election 2016 betting round-up
Tuesday, 15th November 2016
Last week we saw 2016's second major case of a shock result for the betting market. As the first states were about to close on the East Coast of the USA, betting markets were approximately 82% confident that Hillary Clinton was going to be the 45th and first female President of the United States.
On the day action
All the significant on on-the-day action happened between 00:00 and 05:00 GMT. As seems to be the case with presidential elections, most of the story in that five-hour period was a story of one or two crucial swing states – Florida and Pennsylvania.
Interestingly the early results, despite indicating a very close election, were taken by the markets to be positive signs for Hillary Clinton. In the context of a result where lots of states bucked overall trends, it could well be that these early results genuinely were ahead of the market consensus. It's also quite possible though, that we were seeing a case of confirmation bias, where the market saw in the results only the signs that confirmed the pre-existing consensus.
At her peak, buoyed by an early Florida exit poll giving her a 2% lead in that crucial state, Hillary hit 91% chance of winning in the betting markets. In an obvious-in-hindsight lesson for all bettors in this volatile political era, in an obviously close election this was probably the moment to sell any long-Hillary positions.
From that moment onwards, as real results began to come in in Florida and Pennsylvania, the market pretty much only moved in one direction. Crossover happened at 02:40 GMT, with Donald Trump replacing Hillary Clinton for the first time in four years as favourite for the presidency. By 04:45 GMT, shortly before Florida was called, it was very clear there was no path back for Hillary and Donald Trump was at 98% chance.
How we got there - a nine month view
We tracked and aggregated betting-market-implied probabilities for the outcome all the way from 2016-03-09 to the close of the market.
Regardless of how accurate the betting market was in predicting the ultimate result, the movements over the run-up always make for interesting dissection.
Our data shows an interesting cylical pattern over the course of this nine-month period. We'd love to see further research on what might drive such a pattern. We presume it is largely driven by the polls, which in turn are largely driven by the media cycle. But what would explain quite why the media cycle is so cylical?
As ever, we look forward to seeing what our customers do with the historical data.