New Election Models Forecast US Senate and NZ Races
Thomas Miller and Stuart Donovan launched new forecasting tools to predict U.S. Senate outcomes and New Zealand parliamentary seat distributions.
Two new political forecasting tools have launched to predict upcoming elections in the United States and New Zealand. Thomas Miller, a professor and data scientist at Northwestern University, introduced NowCasting via Virtualtout.io to track 35 U.S. Senate races. The system updates every 10 minutes by simulating one million hypothetical elections using betting data from prediction markets such as Kalshi and Polymarket.
Miller's model adjusts for time series analysis and the influence of Donald Trump, whom the system identifies as the primary force driving market trends. As of October 27, the model projects a 23.2% to 23.3% chance that the Republican Party will maintain Senate control. The data suggests Democrats are positioned to flip five seats, potentially shifting the current 53-47 Republican majority to a 52-48 Democratic majority. While Republican odds hit a low of 21.3% on October 24, they have seen modest gains in Texas, Maine, and Michigan, while remaining low in North Carolina, Ohio, and Alaska.
In New Zealand, Stuart Donovan of Motu developed the Herald-Motu Poll of polls model for the November 7 election. The electorate predictor suggests the National Party may see a drop in party vote support but could still return to power through a coalition with Act and NZ First, aided by a predicted seven-seat overhang in Parliament.
