What our forecast - and the debate - tell us about the race

Trump is clearly the favorite, but it’s still early enough that things are fairly close.

This article was originally published on June 28th, 2024.

—

"White House" by Tom Lohdan is licensed under CC BY 2.0 .

A brief note: This is the first article of a "narrative" type using the forecast, which will blend data and opinion. I hope to publish the methodology article for the model tomorrow, when I can think clearly and describe the complex formulas coherently. I may or may not write a purely opinionated piece about what's happening with Biden's candidacy given what happened at the debate.

The topline

The only number that truly matters with these models is the overall odds of who will win the election on November 5th. According to our forecast, Trump is currently "slightly favored" with a 64% chance of winning, while Biden has a 35% chance. (both this and the electoral forecast is nearly identical to Nate Silver’s model, coincidentally) If you add those percentages together, you'll notice they don't equal 100. So, does RFK have a 1% chance of winning? No, that 1% chance actually represents a 1-in-100 outcome of an Electoral College tie.

Digging deeper into the numbers, the model predicts that Trump will win around 285 electoral votes on average. This is fewer than the 306 electoral votes both he and Biden won in their victories in 2016 and 2020, making it the closest race since 2000 (Bush won with 286 in '04). Biden is forecasted to win 253 electoral votes, giving him a fighting chance, but this is somewhat misleading.

Current state/electoral college forecasts. Each hexagon is one electoral vote.

Uncertainty, uncertainty, uncertainty

The reality is that if today were Election Day, Trump would have better odds. By changing the overall uncertainty/standard deviation in the model (basically the maximum expected polling error/swing) from its current 10.3 to 6, which it will be at the end of the campaign, Trump has a 70% chance. However, this doesn’t account for the decreased uncertainty in state forecasts, meaning polls are even more important and fundamentals matter less. Biden is being kept afloat by decent fundamentals right now (so-so consumer confidence, fundraising, and past electoral data) as it is still early. But as the election approaches, the model favors polls much more, and they can be up to 95% of the forecast on Election Day in swing states. If I had to guess, I’d say Trump would probably have around an 80-85% chance assuming nothing changes. This will be explained in more detail in the methodology post.

Could Trump win the Popular vote???

This is simply strange. A Republican has not won more votes nationally since 2004, and yet Donald Trump, seen as one of the most disliked and unpopular candidates in recent history, may very well be the one to break the party’s losing streak this time around. Our model gives him a 59% chance of winning the popular vote and currently expects him to win by about a point, 45.7% to Biden’s 44.8% nationally. But something else is happening here. If you had predicted a 1-point national win for Trump in 2016 or 2020, you would expect him to win with a significant electoral result, well over 300 electoral votes. However, that is not the case now, and it may be a sign that the GOP’s electoral college advantage is shrinking. In 2016, the difference between the tipping point state (WI) and the popular vote was about 3% in favor of Trump. In 2020, Pennsylvania was more like 3.5% to the right of the nation. But in our current forecast, the tipping point is Michigan, with Trump expected to win by about 1%. So the electoral college might actually be much closer to the popular vote this time. *The tipping point state will be displayed on the forecast later, I just need to add it.

Kennedy-Shana(ha)nigans

And then there are the third parties and independents. The model predicts that they will collectively (not just RFK Jr.) win around a whopping 10% of the popular vote, which is more than in 2016 and 2020. The average third-party vote share is 11% across all states, which is quite significant. Ultimately, I don’t think this is disproportionately helping or hurting either party, as they seem to draw about equal support in national and state polling, but it’s interesting nonetheless. There’s a pretty good chance that this number drops over time and ends up being closer to the typical 5-6%, but if it holds, it would be a major night for third parties and would clearly confirm the discontent with our current candidates.

The state(s) of the race

There aren't really any surprises here. Biden is performing much better in the Rust Belt (Wisconsin, Michigan, Pennsylvania), trailing by only about a point in the model, while in the Sun Belt (Arizona, Georgia, Nevada) it's more like five points. His most favorable path to 270 electoral votes is still through WI, MI, PA, and Nebraska’s 2nd Congressional District. However, he would need to win everything he did in 2020 except Arizona, Georgia, and Nevada, making it a close race that Democrats would prefer to avoid. Trump is performing quite well in the Sun Belt, especially in Georgia and Nevada, and it's uncertain if that will change in the near future. It's important to note that there is still a high level of uncertainty across all these states, with Georgia being the most confident state where Trump has a 61% chance of winning. Therefore, the outcome could still swing either way, and I want to emphasize that point.

The debate

So last night was rough for the President. Really, really rough. He had a terrible start, seeming disoriented, his voice raspy and not able to finish his thoughts. The Biden campaign says he had a cold, which I think is believable enough. In fact, the GOP is probably right that Biden was drugged for the debate, but probably on cold medicine, not cocaine or whatever they think he takes. I guess there’s some chance that that made him a bit drowsy, but it was still a weak showing regardless. Biden, for the most part, held a consistent narrative throughout the 90 minutes, but he was not combative enough. He needed to call out and directly refute his opponent more and couldn’t get his phrasing quite right. Had he been more aggressive, I think the media narrative would be a little different. Instead, we now see a “freakout” among some Democrats and the (small) chance of a major shakeup in which Biden is not the nominee. All in all, as someone who spends a good chunk of my Saturdays in the school year debating, this was extremely frustrating to watch. Both Trump and Biden made false claims (Trump’s were far worse), kept changing the subject constantly (again, Trump was worse here, completely ignoring questions from the moderators and refusing to move on), and ultimately just weren’t persuasive. It truly was just two old white guys having a weak argument with little actual debate. (If you don’t believe me, you should know that they flat out were bragging about their golf abilities because our candidates are obviously very mature).

But Trump “won” the debate because of his performance and confidence. Despite most of his substance being completely false, he showed up with energy, consistently on the attack, and barely stumbled. That comes off as “presidential” to voters, and Joe Biden lacked those qualities. Also, Trump, perhaps for the first time ever, kept himself in check and adhered to the rules, rarely interrupting or shouting down Biden, and staying relatively civil throughout the debate. I think Biden thought he would get under Trump’s skin more, and that just didn’t work out.

The takeaway

So what's clear is that despite the high uncertainty at this point, with four months to go, the advantage in the fight for the White House is with Trump. Biden (or whoever may replace him) has work to do in securing the Rust Belt and potentially making gains in the Sun Belt. He may even need to defend "blue" states like Virginia or New Jersey, depending on how polling ends up there. We still have a long campaign ahead of us, and I'll be updating the forecast twice a week, on Mondays and Fridays. The Senate forecast will hopefully be released soon, with more articles to come. Thanks for reading, and if you have any questions or comments on the model, feel free to reach out on Instagram @hdr_analytics.