Overview

We forecast the 35 Senate races using the same mathematical-modeling approach that we use for our presidential forecasts. For more information and references, click here. Check back after the election for an analysis of our forecast performance!

State Groupings

For the senatorial elections, we forecast the most competitive ("swing") states individually and aggregate all the other states into safe red or safe blue superstates, in which we forecast the mean vote margins. Click below to expand the breakdown of our swing states and superstates.
 
State Categorizations
Swing states:
  • Alabama (AL)
  • Alaska (AK)
  • Arizona (AZ)*
  • Colorado (CO)
  • Georgia (GA)
  • Georgia Special (GAS)**
  • Iowa (IA)
  • Kansas (KS)
  • Maine (ME)
  • Michigan (MI)
  • Montana (MT)
  • North Carolina (NC)
  • South Carolina (SC)***
  • Texas (TX)
 
We forecast the vote margin in each of these states individually.
 
* Arizona has a special election in 2020.
 
** The Georgia special election has more than two main candidates. Our model handles races between two candidates, so we do not forecast this election.
 
*** We considered South Carolina to be part of our Red superstate prior to October 13.
Red superstate:
  • Arkansas (AR)
  • Idaho (ID)
  • Kentucky (KY)
  • Louisiana (LA)
  • Mississippi (MS)
  • Nebraska (NE)
  • Oklahoma (OK)
  • South Dakota (SD)
  • Tennessee (TN)
  • West Virginia (WV)
  • Wyoming (WY)
 
We forecast the mean vote margin across these states.
Blue superstate:
  • Delaware (DE)
  • Illinois (IL)
  • Massachusetts (MA)
  • Minnesota (MN)
  • New Hampshire (NH)
  • New Jersey (NJ)
  • New Mexico (NM)
  • Oregon (OR)
  • Rhode Island (RI)
  • Virginia (VT)
 
We forecast the mean vote margin across these states.
 

Final Forecast of the 2020 U.S. Senatorial Elections

In our final forecast of the 2020 U.S. senatorial elections, Democrats gain control of the Senate about 85% of the time and Republicans maintain control about 15% of the time. We forecast a mean of 52.7 Democratic seats and 47.3 Republican seats, with a standard deviation of 1.4 seats across our simulated elections. We project especially tight races in Georgia, Iowa, Montana, and South Carolina, where the vote margins that we forecast are below 2 percentage points.

 

For the case of a tie in the Senate (50 seats for each party), we use our presidential forecasts to determine which party wins control. Because we do not forecast the vote margin in the Georgia special election (since it has more than two main candidates), we assigned this race an equal chance of voting Republican or Democratic when we created our Senate-seat distribution.

 

Our forecast uses polling data that we downloaded from FiveThirtyEight on November 1 to determine our model parameters. We posted our final forecast on November 2 at 22:41 CST.


 

Our forecast vote margins for each state or superstate for the senatorial races

*Arizona has a special election in 2020.

 

Our forecast of the Senate composition based on 10,000 simulated elections

 
 

Prior Forecasts

You can access the forecasts that we made for the 2020 senatorial elections on earlier dates by clicking the links below to expand the page.
 
November 1, 2020, 15:13 CST

Our forecast uses polling data that we downloaded from FiveThirtyEight on October 30. In 10,000 simulated elections, Democrats gain control of the Senate about 85% of the time. We forecast 52.7 Democratic seats, with a standard deviation of about 1.4 seats across our simulated elections.

 

For the case of a tie in the Senate (50 seats for each party), we use our presidential forecasts to determine which party wins control. Because we do not forecast the vote margin in the Georgia special election (since it has more than two main candidates), we assigned this race an equal chance of voting Republican or Democratic when we created our Senate-seat distribution.


 

Our forecast vote margins for each state or superstate for the senatorial races

*Arizona has a special election in 2020.

 

Our forecast of the Senate composition based on 10,000 simulated elections

 
October 29, 2020, 18:14 CDT

Our forecast uses polling data that we downloaded from FiveThirtyEight on October 27. In 10,000 simulated elections, Democrats gain control of the Senate about 86% of the time. We forecast 52.6 Democratic seats, with a standard deviation of about 1.4 seats across our simulated elections.

 

For the case of a tie in the Senate (50 seats for each party), we use our presidential forecasts to determine which party wins control. Because we do not forecast the vote margin in the Georgia special election (since it has more than two main candidates), we assigned this race an equal chance of voting Republican or Democratic when we created our Senate-seat distribution.


 

Our forecast vote margins for each state or superstate for the senatorial races

*Arizona has a special election in 2020.

 

Our forecast of the Senate composition based on 10,000 simulated elections

 
October 27, 2020, 21:58 CDT

Our forecast uses polling data that we downloaded from FiveThirtyEight on October 25. In 10,000 simulated elections, Democrats gain control of the Senate about 87% of the time. We forecast 52.7 Democratic seats, with a standard deviation of about 1.4 seats across our simulated elections.

 

For the case of a tie in the Senate (50 seats for each party), we use our presidential forecasts to determine which party wins control. Because we do not forecast the vote margin in the Georgia special election (since it has more than two main candidates), we assigned this race an equal chance of voting Republican or Democratic when we created our Senate-seat distribution.


 

Our forecast vote margins for each state or superstate for the senatorial races

*Arizona has a special election in 2020.

 

Our forecast of the Senate composition based on 10,000 simulated elections

 
October 23, 2020, 23:24 CDT

Our forecast uses polling data that we downloaded from FiveThirtyEight on October 21. We simulate 10,000 stochastic elections, and Democrats gain control of the Senate about 86% of the time. For the case of a tie in the Senate (50 seats for each party), we use our presidential forecasts to determine which party wins control.

 

Because we do not forecast the vote margin in the Georgia special election (since it has more than two main candidates), we assigned this race an equal chance of voting Republican or Democratic when we created our Senate-seat distribution.


 

Our forecast vote margins for each state or superstate for the senatorial races

*Arizona has a special election in 2020.

 

Our forecast of the Senate composition based on 10,000 simulated elections

 
October 13, 2020, 22:43 CDT

Our forecast uses polling data from FiveThirtyEight that we gathered on October 9 at 18:13 CDT. Unlike our prior forecasts, we now treat South Carolina individually instead of including it as part of the Red superstate. Interestingly, the model forecasts South Carolina as leaning slightly Democratic. Otherwise, the vote margin that we project in each state has changed very little since our last forecast.

 

We simulate 10,000 stochastic elections, and Democrats gain control of the Senate about 86% of the time. For the case of a tie in the Senate (50 seats for each party), we use our presidential forecasts to determine which party wins control.

 

Because we do not forecast the vote margin in the Georgia special election (since it has more than two main candidates), we assigned this race an equal chance of voting Republican or Democratic when we created our Senate-seat distribution.


 

Our forecast vote margins for each state or superstate for the senatorial races

*Arizona has a special election in 2020.

 

Our forecast of the Senate composition based on 10,000 simulated elections

 
October 5, 2020 22:43 CDT

Using polling data from FiveThirtyEight that we gathered on September 28, 2020, at 22:21 CDT to determine our model parameters, we find that Montana, Kansas, and Georgia have the tightest senatorial races. In our forecasts below, we simulate 10,000 stochastic elections and show the mean vote-margin outcomes. The yellow lines indicate the range that encompasses the middle 80% of our simulated elections. Many of these lines cross zero, suggesting close elections.


 

Our forecast vote margins for each state or superstate for the senatorial races

*Arizona has a special election in 2020.

2020 U.S. Election Forecasts -- Samuel Chian, William L. He, Christopher M. Lee, and Alexandria Volkening