This year, we began our daily poll-based analysis on August 22nd. Since then, Clinton had led Trump every day... until election eve. On Monday November 7th, 24 new state polls were listed on RealClearPolitics. Of these, 10 had more recent polling periods and therefore replaced polls we had in those 10 states. 4 polls were from The Trafalgar Group , a Republican pollster, who found Trump ahead in Florida by 3.59%, in Georgia by 6.5%, and for the first time, slightly ahead in Michigan by 1.7% and in Pennsylvania by 1.92%.

Our overall poll-based odds dropped from a Clinton win probability of 90.7% without these 4 polls, to a 67.8% win probability for Trump! Tuesday afternoon, when it became apparent that no other polls would be released, we posted our final poll-based analysis on our new blog page. Additionally, we posted our analysis without Trafalgar's polls, to show how just a few polls could tip the odds dramatically. The election was actually much closer than most had realized. We're unaware of any other analyst using poll-based analysis to determine the Electoral College winner, who pointed to a Trump win. We also projected that Clinton would win the popular vote.

From our poll-based analysis of the Senate, our overall odds indicated that the Republicans would control the Senate, regardless of which candidate became president. This also differed from most other analysts projections. Like other analysts, our market-based odds indicated: a Clinton victory (incorrect,) that the Democrats would control the Senate (incorrect,) and that the Republicans would control the House (correct.) When actual votes began being counted, our one-hour live chart of Clinton's overall odds, updated every 40 seconds, showed Clinton's odds beginning to drop. At 9:26pm PST, Clinton's market-based odds dropped below 1%. We projected a Trump win on our blog page. ... more than 2 hours before any network or news agency would call the race.