New election forecast model predicts Harris' victory

Kamala Harris and Tim Walz. – © AFP SAUL LOEB

A new model updates the US election forecasts. This is done based on daily data determined by the betting market. The model comes from a data scientist at Northwestern University. Using this data, the new forecast model predicts how the Electoral College will vote.

Similar models correctly predicted the 2020 presidential election and the 2021 runoff elections for two Senate seats in Georgia.

The model updates the odds of former President Donald Trump or Vice President Kamala Harris winning every day. With this level of precision, new media and other interested parties can see how individual events – such as a debate, campaign activities or court rulings – could affect the potential outcome of the US presidential election.

Thomas Miller of Northwestern University, who developed the platform, believes the presidential debates will be crucial and “Trump's legal battles are also crucial. In recent months, there have been massive changes in the projected Electoral College votes related to current events and campaign activities.”

Miller is the faculty director of the Master's in Data Science at Northwestern University's School of Professional Studies. Viewers can follow his daily predictions and accompanying analysis on his website, The Virtual Tout.

Miller's system uses data from PredictIt, prediction markets where users bet real money on political elections. Miller then uses price data as input to his forecast of how the Electoral College will vote.

Using his daily forecasts, Miller can measure reactions to individual news or campaign events. For example, when Trump received a stay of sentencing for his New York conviction on “hush money” charges, Harris' campaign suffered a projected loss of 68 electoral votes, according to Miller's model.

Miller's system currently predicts that the Harris-Waltz ticket will win the November election with 289 electoral votes. The candidates need 270 votes to become presidential nominees.

Miller's models proved to be quite accurate in the 2020 presidential election—they only predicted one state incorrectly (Georgia). Miller updated the algorithm after that error, uncovering inherent biases that led to a forecast of the Democratic vote in the Electoral College that was 12 votes lower than the actual result.

Miller's 2020 model was still more accurate than Nate Silver's FiveThirtyEight. While FiveThirtyEight predicted 348 electoral votes for Biden-Harris, Miller predicted 294 electoral votes for the Democrats. Ultimately, the Electoral College gave 306 votes to Biden-Harris.

Miller says popular election forecasting systems, including FiveThirtyEight and the Economist's prediction project, are inherently flawed because they use data from opinion polls. According to Miller, that data is outdated compared to fast-moving news cycles.

Miller cites another key advantage of prediction markets over political polls: Large groups of investors stay in the markets until Election Day. These groups grow larger as Election Day approaches. Miller relies on prediction markets with tens of thousands of investors, with thousands of stocks traded every day. Typical opinion polls have between one and two thousand participants, with new participants recruited for each poll.