The day before election-day, our final forecast is for Trump to win the general election by a margin slightly less than Obama’s victory over Romney in 2012, or by about 3 percentage points.

This forecast is heavily weighted towards social media, on-line activity, consumption data, interest levels, anonymous polls, and an expected record-setting turnout.  In contrast, other forecasters tend to rely on live interviewer polls, poll averages, and the assumption that relative turnout should remain more or less in-line with previous elections.

The most quoted forecasters have Clinton running away with the election.  The probability of a Clinton victory are at 99% at the Princeton Election Consortium, 98.1% at the Huffington Post, 89% at PredictWise, 88% at Daily Kos,  84% at the New York Times, and 68.5% at 538.com.  In other words, the ‘market’ has essentially declared Clinton the winner.

Table 1:  Forecast of Popular Vote for 2016 US Presidential Race

Source:  FiveThirtyEight, RealClearPolitics (RCP), and Coogan     Note:  The Stein forecast from FiveThirtyEight is estimated.

In terms of Electoral Votes, the forecasts are equally as different.

Table 2:  Forecast for Electoral Votes for 2016 US Presidential Race

Source:  FiveThirtyEight, Daily Kos, Princeton Election Consortium, and Coogan

In our forecast, no states flip from Republican to Democrat, while NH, PA, OH, IA, MI, FL, NV, and ME (district 2) went from Democrat in 2012 to Republican in 2016.

CO and VA are extremely difficult to call and ended up in the Democrat column mostly due to inertia of having voted Democrat in 2012.  These two states could easily flip to Republican hands by changing just a few minor assumptions.

Electoral surprises could occur in many states as this election is just so different than previous ones.  Spikes in turnout from the very pro-Trump ‘whites without college degrees’ demographic could remake the electoral map, shocking many poll followers.  WI and MN have been Democrat for so long it is hard to imagine them voting for Trump.  But, in the case of aggressive turnout from this demographic, these states could flip.

In the case that Social Desirability Bias plays a larger role than expected, NJ and RI could flip Republican.  These states, though polling strongly Democrat, have provided enough indications of an underlying bias that a pro-Trump vote is possible, though not likely.

Lastly, NY will shock most political observers by the small margin of Clinton’s victory, which we forecast to be in the single digits.  Current polls have Clinton up by about 20 percentage points while actual results from 2012 had Obama winning by 28 points.

The 2016 US Presidential Election has been extraordinary with much of the political environment dramatically changing.  As described in the many posts connected to this site as well as in ZettaCap research, new data and tools are needed in such a modified environment.  Social media, on-line activity, and a slew of other approaches have proven themselves extremely insightful in 2016.  The old model of relying blindly on live interviewer poll averages will likely hit the end of its useful life in 2016.