Tang Shiping predicts Trump "highly likely" to win - probability more than 60%

Fudan professor said his Agent-Based Modeling (ABM) computer simulations show Trump to win Michigan, Ohio, Pennsylvania, Florida, & Wisconsin and Harris to win Georgia, Arizona, and North Carolina.

Nov 03, 2024

Below is a press release from the Center for Complex Decision Analysis (CCDA), Fudan University in Shanghai. CCDA was founded and led by Shiping Tang, the most unique and one of the most internationally influential social scientists based in the Chinese mainland.

CCDA underlines that the “forecasting exercise is purely academic, and we have no intention of influencing the actual U.S. election process in any way.”

2024 U.S. Presidential Election: Forecasting Results Based on ABM Computer Simulations

November 3, 2024

Center for Complex Decision Analysis (CCDA), Fudan University

On November 3, 2024, at 10:00 a.m. Beijing time (two days before the 2024 U.S. presidential election), the research team from the Center for Complex Decision Analysis at Fudan University, led by Professor Tang Shiping, officially released the forecasting results for the upcoming 60th U.S. presidential election scheduled for November 5. The team predicts that Donald Trump is highly likely to win the U.S. presidency (with a probability more than 60%).

To note, our team arrived at the above forecasting results through computer simulations on October 6, 2024. The screen shot below shows this:

Screenshot 1：We have obtained our final forecasting results on Oct. 06, 2024. They were based on data finalized on Oct. 04, 2024 (mostly with updated data from the third quarter).

Our forecasting is primarily based on Agent-Based Modeling (ABM) simulations. The distinctive feature of our methodology is its reliance on real data and agent-based models, rather than polling data or individual expert judgments. Previously, our research team has successfully and accurately predicted six election results in China’s Taiwan and the United States by developing the new forecasting platform. The published results show relative vote shares, not just simple win/loss outcomes.

The team’s forecasting process consists of three key steps: (1) Using ABM-based simulation to predict vote shares and Electoral College votes for eight key swing states (states where the vote share gap between Democrats and Republicans was less than 5% in the 2020 presidential election); (2) Using 1 million rounds of Monte Carlo simulation to predict tracking poll results and Electoral College votes for the remaining 42 states and Washington D.C.; (3) Combining both results to produce the final forecast for the 2024 U.S. presidential election.

Our ABM forecasting covers 8 key states (Note: the 8 states are the states in which the two parties had less than 5% difference in their relative vote share in the 2020 election). Our ABM forecasting of the 8 states is the following: Trump will win Michigan, Ohio, Pennsylvania, Florida, and Wisconsin; Harris will win Georgia, Arizona, and North Carolina.

Note: Since we rely on public polls on the other 42 states, our forecasting of the overall election results also depends on the reliability and accuracy of the public polls. By comparison, our forecasting of the results of the 8 states depends on ABM only. Also, due to the enormous attention for this year’s US presidential election, we shall release technical details after the election.

In fact, the team simulated three times this year in April, June, and October, using data from different periods. The team deliberately refrained from publishing these predictions to avoid possible accusations of influencing the actual voting process. [Based on our earlier and unpublished forecasting results, after the Biden-Trump debate (June 27, 2024), on July 13, 2024, Prof. Tang, publicly predicted that the Democratic Party would ask President Biden to step aside, on Weibo. See the screen shot below (in Chinese).]

Again, we stress our forecasting exercise is purely academic, and we have no intention of influencing the actual U.S. election process in any way.

This forecast is the result of a year-long research project conducted by the team led by Professor Tang Shiping at Fudan University. It follows the track record of accurate predictions for the 2016 Taiwan leadership election, 2018 US midterm elections, 2018 Taiwan local elections, 2020 Taiwan leadership election, 2020 US presidential election, and 2024 Taiwan leadership election. Our prediction system, which began development in 2015, has been continuously improved over nearly 10 years and has become increasingly sophisticated and advanced. For more information, please visit our official website: www.ccda.fudan.edu.cn.

The core of our election forecasting system is based on Agent-Based Modeling (ABM) simulation technologies, and does not rely on opinion polls or individual expert judgments. During the simulation process, our team strives to: (1) conduct model specification under the guidance of political science theories; (2) combine individual-level data of voters and candidates with structural-level data of economic and social conditions; (3) integrate long-term stable effects and short-term shock effects, attempting to predict elections more accurately through standardized models, rigorous procedures, and algorithms. By introducing ABM simulation, this research represents a valuable attempt to advance computational political science and make election predictions more scientific.

Standardized and rigorous prediction research is a direction that political science needs to explore, and our effort aims to be a valuable attempt to advance computational political science and make election predictions more scientific.

Appendix:Previous Election Forecasting Records by the Center for Complex Decision Analysis (CCDA), Fudan University

(1) On January 5, 2016 (Chinese version) and January 10, 2016 (English version), before the Taiwan leadership election that year, our team released predictions for the election results. This was our first attempt at election forecasting using the new methodology of ABM simulations. Please click the following links for details.

Chinese version:

http://www.ccda.fudan.edu.cn/index.php?c=article&id=75

English version:

https://china.ucsd.edu/_files/01112016_Taiwan-election.pdf

https://chinafocus.ucsd.edu/2016/01/10/taiwan-election-results-predicted-in-computer-simulation/

(2) On November 4, 2018, before the US Senate election, our team released election predictions for two states (West Virginia and Missouri). This was our second attempt at election forecasting using the new methodology. Please click the following links for details.

Chinese version:

http://blog.sina.com.cn/s/blog_744a73490102y7bd.html

http://www.ccda.fudan.edu.cn/index.php?c=article&id=80

English version:

http://www.ccda.fudan.edu.cn/index.php?c=article&id=81

https://www.linkedin.com/pulse/united-states-senate-elections-west-virginia-missouri-tang-dr/?trackingId=shAzPfa2GP9KRu%2BbwvnIyQ%3D%3D

Comparison of Predicted Results and Actual Results:

http://www.sirpa.fudan.edu.cn/?p=12377

https://www.linkedin.com/pulse/prelimary-assessment-our-forecasting-excellent-perhaps-tangdr/?trackingId=vmFPQdRhrHHJhTrkX%2Bh%2F3Q%3D%3D

(3) On November 22, 2018, before the election day, our team released predictions for Taiwan's local elections (Taipei, New Taipei, Taoyuan). This was our third attempt at election forecasting using the new methodology. Please click the following links for details.

Chinese version:

http://www.ccda.fudan.edu.cn/index.php?c=article&id=83

English version:

http://www.ccda.fudan.edu.cn/index.php?c=article&id=84

https://www.linkedin.com/pulse/taiwan-local-elections-taipei-new-taoyuan-forecasting-tang-dr/?trackingId=VmpIAUnefnghFHhzA9ODeg%3D%3D

Comparison of Predicted Results and Actual Results:

http://www.ccda.fudan.edu.cn/index.php?c=article&id=85

(4) On January 9, 2020, before the election day, our team released predictions for the Taiwan leadership election. This was our fourth attempt at election forecasting using the new methodology. Please click the following links for details.

English version:

https://www.linkedin.com/pulse/2020-taiwan-elections-forecasted-computer-simulation-shiping-tang-dr/

Comparison of Predicted Results and Actual Results:

http://www.ccda.fudan.edu.cn/index.php?c=article&id=96

http://www.ccda.fudan.edu.cn/index.php?c=article&id=97

(5) On November 1, 2020, before the election day, our team released predictions for the 2020 U.S. Presidential Election. This was our fifth attempt at election forecasting using the new methodology. Please click the following links for details.

Chinese version:

https://ccda.fudan.edu.cn/news/2020-us-presidential-election-forecasting/

English version:

https://ccda.fudan.edu.cn/news/2020-us-presidential-election-forecasting-en-us/

Comparison of predicted results with actual results (in Chinese):

https://ccda.fudan.edu.cn/news/2020-us-presidential-election-abm-simulation/