Donald Trump is aiming for a second term, Joe Biden is down, Kamala Harris has rejuvenated the base. Can the Democrats win? Or is Trump stronger than he seems? Welcome to the New Statesman's election forecast for the 2024 US election.

Our model used a variety of polls and demographic data to produce near-daily forecasts of vote share estimates for each U.S. state. It gives us a simple prediction: How likely is it really that Kamala Harris or Donald Trump will win?

For much of 2024, Joe Biden lagged behind Trump in the polls. Trump was unusually ahead of the incumbent – and the televised debate between the two in June only reinforced that dynamic. This put Trump within reach of states like Virginia, New Hampshire and Minnesota.

But Biden's decision to withdraw turned everything on its head. The Democratic base rallied behind Harris, although it was unclear how lasting this recent surge of energy was.

Her appearance in the debate against Trump served as an introduction to voters. And their numbers benefited from it. But with the withdrawal of RF Kennedy and the lack of further debates, the novelty wore off. And it was Trump who spent much of October grabbing headlines and setting the political weather.

The election will likely be decided in seven states: Pennsylvania, Wisconsin, Michigan, Nevada, Arizona, Georgia and wild-card contest North Carolina. Biden won six of those elections in 2020, but Trump has maintained long-term leads in all of them this year.

The New Statesman's model is a probabilistic model. Polls have margins of error, and in some of these key states pollsters have performed poorly in the past. But analysts have also learned from their mistakes: namely, from the catastrophic misjudgments during Hillary Clinton's 2016 election campaign.

A healthy model should take all of this into account. While pollsters claim they have improved their methodology, healthy skepticism would not be amiss. For example, after 2016, the industry claimed it would change its sampling and surveys. But in 2020, the Trump vote was again underestimated in key swing states.

In 2020, our model predicted that Trump had a one in 10 chance of remaining in the White House. If our model had run in 2016, the numbers would have shown that Trump would have had a one in three chance. Pollsters claim they have corrected because they have underestimated the Trump base since this election. But there is a risk that they have also overcorrected.

Here are the state-by-state odds for Kamala Harris to win.

(See also: Vladimir Putin's inner enemy)