Our Thoughts on the 2016 Election

Cheri Franklin |

Regarding this election, the first thing that comes to mind is “Thank God it’s over!” Without question, this was the worst election in our lifetimes on so many different levels. As with Brexit, the question being universally asked is how did the polls, pundits, prediction models, and even the prediction markets get it so wrong? We may never know the answer to that, but there are some important lessons we can take from this historical upset.

First, no matter how much thought and effort we put into predicting the behavior of a large population, whether through polling it or modeling its behavior, we are bound to be upended by factors that escaped our consideration. In the case of the election, this appears to be the unwillingness of certain voters to truthfully admit whom they were voting for. We are reminded of another instance where complex predictive modeling failed spectacularly which was the Global Financial Crisis of 2007-9 when subprime mortgages and their related derivatives imploded, nearly bringing down the financial system.

Our second lesson relates to the prediction markets (or gambling websites) which had Clinton as an overwhelming favorite (about 80/20). How could they have made such an inaccurate forecast? The key to understanding this is that for the prediction markets, there truly is no right or wrong unless something is predicted with zero or 100% probability, but then there would be no market for the prediction! To say that the prediction markets are flawed, we need a large sample of predictions that we can use to assess how accurately calibrated they are. For example, if we had 100 elections in which the winner was given a 75% probability and say 77 of those elections were actually won by the favored candidate, then we could say that the prediction markets are working quite well. If, however, 98 of those elections were won by the predicted winner, then we could say with very high certainty that the odds-setting process is flawed. In other words, the fact that sometimes the less-favored candidate wins is exactly what assures us that the prediction markets are functioning properly. The same can be said for predictive models such as the one employed by Nate Silver that gave Trump a 35% chance, which some observers criticized as being far too high.

Regarding the financial markets, we saw that as the results became favorable for Trump, the overnight futures trading showed a drop by as much as 800 points in the Dow Jones Industrial average, but when the market opened Wednesday morning, the Dow itself climbed up by about 1.4% (a little over 250 points). The lesson for us is to avoid the temptation to trade (whether it’s with or against the herd) as a response to political events. As long-term buy-and-hold investors, we step back and let the speculators frantically trade with each other.

The final lesson is one that you may have heard from us before yet bears repeating: Ignore the pundits and stick with your plan.