Welcome to Decision Desk HQ’s 2024 Election Forecast Model, a powerful tool for in-depth analysis and predictions for the upcoming presidential, Senate and House races.
In 2020 and 2022, our models correctly forecasted more than 90 percent of outcomes across more than 1,000 contested elections. This model builds on the framework of those models, incorporating methodological improvements to deliver even more accurate and comprehensive insights.
The model considers hundreds of data points including polling, economic indicators, national and local political trends, candidate characteristics and campaign finance information, all processed through advanced machine-learning algorithms. More details on our methodology can be found here.
In addition to our quantitative forecasts, Decision Desk HQ provides qualitative ratings for each race, categorizing them as “Solid,” “Likely,” “Leans” or “Tossup” for each party. These ratings allow readers to track significant shifts in the electoral landscape easily, as changes in these categories can signal important developments in key races. Our qualitative ratings also enable direct comparisons between our forecast and the expert analyses provided by respected outlets such as the Cook Political Report, Split Ticket and Sabato’s Crystal Ball.
The presidency
Less than six months before Election Day, President Joe Biden’s reelection prospects are in worse shape than many Democrats would like them to be. Many fundamentals of the race, including the incumbency advantage, campaign fundraising levels and many macroeconomic factors, generally favor Biden. Yet both national and swing state polls give former President Donald Trump a clear edge.
According to the latest national polling average from Decision Desk HQ/The Hill, Trump leads Joe Biden by 1.0 percentage points across 712 national polls — a significant pro-Trump shift from Biden’s 4.5-point popular vote victory margin in 2020. And if the election were held today, swing state polling suggests Trump would achieve a decisive victory in the Electoral College, flipping all six of the states that Biden won by less than 3 points in 2020: Arizona, Georgia, Michigan, Nevada, Pennsylvania and Wisconsin.
As the 2016 and 2020 elections demonstrated, polling can be subject to systematic errors that might underestimate any candidate by several percentage points. And since most of the campaigning will commence in full following each party’s nominating convention, there’s still plenty of time for movement before Nov. 5.
The 2024 Decision Desk HQ forecast currently predicts that Trump has a 58 percent chance of winning the presidency — meaning that if the 2024 election were held 100 times, Trump would win 58 times and Biden would win 41. It is important to note that a chance of winning above 50 percent does not mean that victory is guaranteed, nor does it represent the candidates’ predicted vote shares. And improbable events can occur. For example, before the Super Bowl in February, the betting lines gave the San Francisco 49ers a 54.1 percent chance of winning, yet the Kansas City Chiefs won, 25 to 22.
America is currently in an unprecedentedly competitive electoral era, marked by nine consecutive presidential elections (1988 to 2020) in which neither the Democratic nor Republican nominee secured a double-digit victory margin in the popular vote. In each of the last four elections, 40 of 50 states (and the District of Columbia) have voted consistently for the same party — a level of consistency not seen since the early 20th century.
As a result, our Electoral College forecast shows that both Biden and Trump have better than an 80 percent chance of winning in all but twelve states. The current predictions give Trump an edge in enough states to win 235 electoral votes compared to Biden’s 226. Since neither candidate’s core coalition puts him over 270 electoral votes, the campaign will thus focus intensely on seven key states: Arizona, Georgia, Michigan, Nevada, North Carolina, Pennsylvania and Wisconsin.
Five of those states — Arizona, Georgia, Michigan, Pennsylvania and Wisconsin — switched from Trump in 2016 to Biden in 2020 and are currently rated as “Toss-Ups.” Nevada is also rated as a “Toss-Up,” having voted for Hillary Clinton and Biden by nearly identical 2.4-point margins in the past two elections. North Carolina, having favored Trump by 3 points in 2016 and by 1 point in 2020, is currently rated as “Lean Republican.”
Our mean Electoral College projections show Trump averaging 282 electoral votes to Biden’s 256. However, the single most likely outcome at this point — occurring in 24 percent of the 14 million simulations we ran for our model — has Trump winning all seven key battlegrounds and a 312 to 226 victory in the Electoral College.
Biden’s position is particularly weak in Arizona, Georgia, Nevada and North Carolina, where he has consistently trailed in the polls by 4 points or more in all of these states, even as he mostly polls within the margin of error in Michigan, Pennsylvania and Wisconsin.
The Senate
Of the 34 Senate seats on the ballot in 2024, 23 are held by Democratic caucus members. Among the 11 presumed battleground states, Democrats are defending seats in eight of them, including three states — Montana, Ohio and West Virginia — that Trump won in both 2016 and 2020. Even if Biden holds the White House, securing Vice President Kamala Harris’s tie-breaking vote, Democrats must win eight of these 11 states (likely all six states Biden won in 2020, plus two more) to maintain their majority.
The 2024 Decision Desk HQ Senate Forecast gives Republicans an 80 percent chance of flipping control of the upper chamber. The GOP is currently favored to win 51 seats to Democrats’ 46, with the contests in Nevada, Arizona, Michigan and Ohio rated as “Toss-Ups.” Republicans hold a more commanding lead in “Safe” states, with 48 compared to Democrats’ 40. This includes a Republican flip in West Virginia, where incumbent Sen. Joe Manchin (D) is retiring and Gov. Jim Justice (R) has secured the Republican nomination, setting up one of the easiest Senate flips in modern electoral history in this Trump+39 state.
The House
House Republicans and Democrats are gearing up for a highly competitive race. Although House Republicans fell short of their lofty expectations in 2022, they still managed to flip nine seats, securing a majority and the Speaker’s gavel. They must now defend this narrow majority against the Democrats, who hold 213 seats — just five seats shy of the 218 needed to regain control.
Our model currently predicts that Republicans have a 64 percent chance of holding the House of Representatives. The GOP is currently favored in 213 congressional districts to Democrats’ 206, with 16 races rated as “Toss-Up.” Our mean projection shows a 223 to 212 Republican advantage, which would leave the House GOP in a situation similar to the result of the 2022 midterms.
A likely Trump victory in 2024 would mean Republicans have a near certain chance to flip control of the Senate, making the House Democrats’ best chance to prevent a GOP trifecta. Given the high volume of competitive House races, specific developments in individual races (such as retirements, single polls and fundraising advantages) have less impact on the overall balance of power compared to the Senate. As a result, shifts in generic ballot polling are the most important factor in determining changes to our forecast for House control.
What is this forecast, anyway?
Forecasting, whether for basketball games or elections, is inherently fraught with uncertainties. Just as underdogs often win in March Madness, our election forecasting model is not immune to error. This reality doesn’t detract from the utility of our forecasts, but it does underscore the importance of interpreting the results correctly.
In the aftermath of the 2016 election, many critiques were levied at forecasters for their apparent failure, exemplified by numerous articles asking, “How did everyone get it so wrong?” Such criticisms often overlook the essence of probabilistic models. For instance, Nate Silver’s FiveThirtyEight model gave Trump a 28.2 percent chance of winning — better than a one-in-four chance. To put this in perspective, consider drawing a card from a standard 52-card deck: The chance of pulling a heart is one in four. You would not be shocked if you were to pick a heart, even though the odds are against it, yet the odds of Trump winning in 2016 were actually better than that.
Forecasts are not certainties. They are designed to help understand the range of potential outcomes. In a presidential election, there are millions of combinations of state results that could lead to a candidate achieving the necessary 270 electoral votes. At DDHQ, we harness extensive data and run thousands of simulations to demonstrate the most likely paths to victory.
Our role is to provide comprehensive and accurate analysis. The public’s role is to interpret this analysis thoughtfully. We work diligently to offer insightful data in hopes that it will be met with critical engagement. In elections, where our pre-existing biases can heavily influence the information we seek out and how we receive it, critical thinking and analysis of information is crucial.
All modelers strive to be right, but accuracy is only part of the equation. Ultimately, the greater goal is to foster an environment where citizens not only consume data but use it actively to refine — not define — their own viewpoints.
Scott Tranter is the director of data science at Decision Desk HQ, where Ryan Gest is a data science and digital content fellow and Zachary Donnini is a data science fellow.