Presidential Forecast

As of September 15th, Kamala Harris and Tim Walz are projected to win with 273 Electoral Votes. Notably, Harris is projected to win the key battleground states of Georgia, Michigan, and North Carolina. However, Trump is projected to win Arizona, Nevada, Pennsylvania, and Wisconsin. See below for all projected state outcomes. Read further for a discussion on modeling philosophy and a brief commentary on the state of the race.

Forecast Philosophy

By design, I will never publish a “probability” or “margin of error” with this model. Of course, there will be some factors I did not adequately account for and unknown factors that will influence the results of this election. Such is life. But I will not hedge my bet nor hide under some probability when I am inevitably wrong on some level big or small. This model will predict the results and I will evaluate my success by the accuracy of those predictions.

I don’t intend to extensively disclose how the model works. However, I will provide generalized information on key inputs. The bulk of the model is based on recent national and battleground state polling using both head-to-head and multi-candidate vote choice questions. Additionally, I have included race “fundamentals” (in this case, an all-encompassing term for non-candidate choice polling results) as I see them:

1. Weather forecasts (increasingly weighted as we approach Election Day)
2. Economic factors (using various stock market and Bureau of Labor Statistics indicators)
3. Media and keyword search trend data
4. National unfavorability polling data

Note: my intention is to update weekly with a final update on November 3rd.

Forecast Commentary

I often say electoral politics can be boiled down to one metric: unfavorability. Most elections are won by simply being the least unpopular person on the ballot. This is why my model is intentionally blind to favorability and significantly weighted to absolute and relative unfavorability.

Through this over-simplified theory of politics, Kamala Harris has some good and bad news. The good news: Harris is currently 6 points less unpopular than Trump (according to my averages). The bad news: Trump is marginally less unpopular than he was in 2020 (-1pp) and Harris is significantly more unpopular than Biden was (+4pp). Furthermore, while Trump’s unfavorable’s are generally stagnant (with both a high floor and low ceiling), Harris has seen massive improvement in recent weeks. To me, this indicates perceptions of her are still very malleable with the real possibility to fall back out of favor with the electorate but also possible growth yet to be realized between now and November 5th.