Historical Model Results

A look at how often Solid, Likely, and Lean favorites won—and where Toss-Ups landed—in retrospective Election Day tests of the Model. These are backtests, not the forecasts originally published before those elections.

Results updated August 26, 2026.

The big picture

How often did the favorites win?

These percentages are observed win rates within each rating, not the probabilities assigned to future races. A 100% result in this sample does not guarantee future wins; some groups contain only a handful of races.

U.S. House

2016–2024 elections

1,953 races

Solid1,367 of 1,376 favorites won

99.3%

Likely234 of 247 favorites won

94.7%

Lean184 of 230 favorites won

80.0%

Toss-Up100 races; no official favorite

65 R · 35 D

U.S. Senate

2016–2024 elections

165 races

Solid124 of 124 favorites won

100.0%

Likely9 of 9 favorites won

100.0%

Lean11 of 17 favorites won

64.7%

Toss-Up15 races; no official favorite

7 R · 8 D

Governor

2017–2025 elections

104 races

Solid70 of 70 favorites won

100.0%

Likely12 of 13 favorites won

92.3%

Lean12 of 15 favorites won

80.0%

Toss-Up6 races; no official favorite

3 R · 3 D

Toss-Ups are not counted as right or wrong. A Toss-Up means neither party reached a 60% chance. We simply show which party won those races.

How these results were tested

These are walk-forward tests. The Model was tested one election at a time using only results available before that election.

For House and Senate races, 2014 was the baseline: 2016 could use 2014, 2018 could use 2014 and 2016, and so on. Governor races used the same approach. Below are the definitions for each rating.

SolidVery strong favorite: 92% chance or better.

LikelyClear favorite: 80% to under 92%.

LeanNarrow favorite: 60% to under 80%.

Toss-UpNeither party reached 60%, so no favorite was named.