Product
More
Legal
© 2026Daily24.All rights reserved.
From Alaska to Texas, the battle for control of the U.S. Senate is unfolding across 35 races. And a new Northwestern University election model is tracking them all—every 10 minutes.
Called NowCasting, the new system uses data from political prediction markets to calculate the probability of a Republican victory in each race as well as the likelihood that the party will retain control of the Senate.
Every 10 minutes, the model runs 1 million hypothetical midterm elections—simulating 35 million individual Senate race outcomes in the process. Its algorithm places particular emphasis on market activity from the preceding hour. Using this method, NowCasting can see immediate responses to singular news events.
As of noon on Tuesday (Oct. 27), NowCasting projects the Republican Party has a 23.2% chance of maintaining control of the Senate. Users can explore state-by-state predictions and overall trends on an interactive, color-coded map.
"Politics can change quickly, and I wanted to develop a system that can keep pace," said Northwestern's Thomas Miller, who developed the forecast platform. "By combining the collective expectations of thousands of traders with sophisticated statistical modeling, we can capture how the political landscape shifts in real time. We're not just looking at individual races. We're connecting the dots across the country to understand the bigger picture in the battle for Senate control."
Miller is the faculty director of the master's in data science program in Northwestern's School of Professional Studies.
NowCasting is the latest model in a string of election forecasts for Miller. During the 2020 U.S. presidential election, he developed The Virtual Tout, a prediction system that used political betting markets to simulate 1 million hypothetical elections every hour. His final forecast predicted 294 Electoral College votes for Joe Biden, just 12 votes shy of Biden's actual total of 306.
Miller introduced prediction surveys in January 2021 to track Georgia's two Senate runoff elections. His model correctly predicted wins for Democratic Sens. Jon Ossoff and Raphael Warnock.
Returning to the presidential race in 2024, Miller's Virtual Tout correctly predicted President Donald Trump's victory. His model favored Trump even as several other prominent forecasts—such as FiveThirtyEight, The Economist and Nate Silver's Silver Bulletin—narrowly favored Vice President Kamala Harris.
Now, Miller has taken his model a step further.
Although Miller's previous models relied on PredictIt, NowCasting draws on two major platforms: Kalshi and Polymarket. It also incorporates statistical models used in finance and economics to identify underlying trends and relationships among the 35 Senate races. The new model updates every 10 minutes—six times more frequently than his original hourly system.
But the biggest change is how Miller interprets the data. While his earlier models focused on forecasting election outcomes, NowCasting aims to capture what prediction markets signal right now. Instead of treating prediction market prices as probabilities, the system analyzes and calibrates the data to generate its own estimates of each candidate's likelihood of winning.
"Prediction markets provide event contract prices, not probabilities," Miller said. "NowCasting takes that information and uses statistical modeling to generate probabilities that reflect what the markets are telling us right now."
Who's behind this story?
Gaby ClarkMA in English, copy editor since 2021 with experience in higher education and health content. Dedicated to trustworthy science news. Full profile →
Andrew ZininMaster's in physics with research experience. Long-time science news enthusiast. Plays key role in Science X's editorial success. Full profile →
Citation: New election forecast updates predictions every 10 minutes (2026, October 27) retrieved 28 October 2026 from https://phys.org/news/2026-10-election-minutes.html
This document is subject to copyright. Apart from any fair dealing for the purpose of private study or research, no part may be reproduced without the written permission. The content is provided for information purposes only.
