Robinson Analytics Models Correctly Forecast 2016 Presidential Election
Presidential Race Analysis
Robinson Analytics has developed two probability models based on all of the national polls. Our traditional or classic model uses the national polls that are used to calculate the Real Clear Politics average of the polls. We made this change to the model to reflect the reset in the general election that we think happened on October 28th, when the FBI reopened their investigation into Secretary Hillary Clinton. So this latest update includes national polling from October 28, 2016 through AM of November 7, 2016. We ran 100,000 simulations on the polling data to forecast our point estimates for Hillary Clinton and Donald Trump on Election Day November 8, 2016.
Due to the unconventionality of this presidential cycle, we are publishing two probability models (a Classic or Traditional Model and a Brexit Model) for our elections. As we get closer to Election Day, this Brexit Model seems to show more strength for the outsider candidate. Brexit is when Britain voted this summer to exit the European Union and all of the polls and prediction models said that it wouldn’t happen.
Our Classic or Traditional Model Says…
We are highly confident in a two-way race, Hillary Clinton’s vote total on Election Day will be between 37% and 56.2%, and Donald Trump’s vote total on Election Day will be between 35.6% and 54.8%.
In our first probability model, in a two-way race, we estimate that Hillary Clinton will get 46.5% or more of the vote and that Donald Trump will get 45% or more of the popular vote. These vote totals carry a 45.3% probability of a Clinton win and a 54.7% probability of a Trump tie or win.
In a four-way race (including Johnson and Stein), we estimate Secretary Clinton will get 45.5% or more of the popular vote and Mr Trump will get 42.5% or more of the vote. This comes with a 66.8% probability for a Hillary win and a 33.2% probability of a Donald tie or win.
This race is essentially a tie in the popular vote, because it is within the margin of error. Our projection is a 1.5% difference in the vote between the two in a two-way race and a 3% difference in the two in a four-way race. They both are within the margin of error. However, we give the edge here to Secretary Clinton. So it is “likely” that Mrs Clinton will win the popular vote. This is not guaranteed, but the most “likely” outcome.
Popular Vote Actual: Clinton 48.2% versus Trump 46.1% | Our Forecast was correct
Other Predictors:
The prediction markets are giving Clinton a 73.8% probability of winning the election.
FiveThirtyEight (Nate Silver) is giving Hillary a 66.9% probability of winning.
The prediction markets are giving Secretary Clinton a 56% probability of winning the popular vote in the range of 2 to 2.9%.
The Iowa Election Market is giving Clinton a 71.7% probability of winning the presidency.
The RealClearPolitics polling average for Secretary Clinton in a two-way race is a 2% win over Mr. Trump.
The RealClearPolitics polling average for Mrs. Clinton in a four-way (including Johnson and Stein) race is a 2.2% win over Donald Trump.
Our Brexit type of Model says…
Our second probability model point estimate for Trump is 55.9% on election day. That would put Hillary with less than 44.1% on election day. If you notice I had Donald at 49.5% with this model on October 3rd. That is a 6.4 point increase since October 3rd. There is really something strange going on out there among the electorate. This point estimate goes along with a 77.5% probability of Trump winning and a 21.1% probability of Hillary winning. All I can say is that this election is a really a head scratcher.
Robinson Analytics Brexit Model points to a Trump win. Our Brexit model is viable because the polling is within the margin of error. There is still a possibility of a surpise upset here. There is a 33.2% probability of a surprise upset.
I think what this model is picking up is the fact, according to the Reuters/Ipsos poll released on October 3rd, that 64% of American’s are dissatisfied with the direction of the country. I think that is what is driving this model.
Electoral College Actual: Clinton 43.1% versus Trump 56.9% | Our Forecast was correct
Other Predictors:
The stock market performance 90 days before the election is 86% accurate for predicting which party will win the election. This has been so for over a period of 88 years. Since August 8, 2016 until today, the Standard and Poor’s 500 stock market index is down approximately 4%. This signals that the incumbent party will lose. Thus a Republican victory. Wall Street heavily favors a Hillary presidency.
Professor who’s predicted 30 years of presidential elections correctly is doubling down on a Trump win. Allan J. Lichtman, distinguished professor of history at American University, said Democrats would not be able to hold on to the White House. He actually doubled-down on his prediction last week October 28th. He doesn’t want Donald Trump to win, but his model is telling him that is what is going to happen.
Which model do you choose?
So if you think that a Brexit type of event is not possible in the United States, then our Classic or first model is for you. If you are of the persuasion that a Brexit type of event is possible, then our second or Brexit model is for you.
Electoral College Projections with Latest Polling: 270 is needed to win
Below projection is as of November 4, 2016 polling:
These models will be updated as relevant data become available.
Amos B Robinson is the Principal of Robinson Analytics (www.robinsonanalytics.com). In 2016, he is celebrating his 12th year in business. He specializes in helping small to medium-sized businesses grow their sales and profits. Call Amos to talk with him about growing your sales and profits. Amos can be reached on LinkedIn or on Twitter @RobinsonBC1