Slightly Off
How my 2026 forecasts held up
We’re still waiting on every vote to be counted and every race to be called, but as promised, here’s a preliminary view of what I got right and, more importantly, what I got wrong. Predictions were finalized on October 26th, 2026.
Senate
Very bad. I give myself a D- for these. The one saving grace for these predictions is that I had every Independent losing out west. I saw a lot of people with similar predictions to mine calling those races as much closer, or as wins for the Independents. Rather than being most surprised by any individual race, I’d say I’m most surprised by the margins of victory. As you can see on the map below, Dems strongly outperformed what I expected.
I’ll update this with Alaska and Mississippi once those races are all counted and called.
Loading map
iAbout this data
Margins and misses work the same way as for governors. In Montana, Nebraska and South Dakota the main challenger was an independent, so misses toward the challenger are labeled D/I and shaded on the blue side.
Figures are unofficial counts from the Associated Press, NBC News and Fox News as of November 6 to 8, 2026. Alaska and Mississippi are uncalled. For races I called Safe, only the winner is scored.
Governors
I give myself a C- on these. I think my misses weren’t terrible, and I had two exact calls, which is less than I’d really call good, but could be worse. My biggest miss and biggest surprise was Hinojosa, not just that she won, but that she won by so much.
Every miss was bluer than I thought it’d be. I’ll update this with Alaska and California once those races are fully counted.
Loading map
iAbout this data
Margins are the leading candidate’s share of the vote minus the runner-up’s, the same way my forecasts defined them. In Rhode Island and Wyoming the runner-up was not a Democrat or a Republican, so those margins are measured against the actual second-place finisher.
The miss is the distance from the result to the closest edge of the window I gave, in the direction of the party that did better than I expected. Results inside the window are dark green; races I called Safe that went the way I said are sage.
Figures are unofficial counts as of November 8, 2026, from NBC News and Fox News, and will move as the last ballots are counted. Alaska and California are shown as uncalled; Maryland and Oregon are also still counting.
House
These are a bit better. I rate myself a B on the House map. I still definitely underestimated the Democrats, but this seems to be a much more acceptable level of underestimation, especially compared to a lot of the other predictions that were out there. I’m most proud of my accuracy on ME-02, the only GOP flip that didn’t come from redistricting. I’m most surprised by IN-05.
Every miss but one, CO-03, was bluer than I thought it’d be. I’ll update this with the uncalled races once results are finalized.
Loading map
Scroll or pinch to zoom in, drag to move around.
iAbout this data
Each district is compared with the party that won it. Uncalled districts are the ones the Associated Press, whose race calls the New York Times uses, had not called as of November 8, 2026. In races between two candidates of the same party, a call only counts once the winner is called.
Winners are each seat’s current party except where the seat changed hands, using CBS News’s list of flipped districts. Races called after that list was published are included.
A theme for tonight, especially with the Senate elections, is that I underpredicted the blue wave. I was very much in the camp of expecting a reverse 2022 before election night, at least in the Senate, and that shows in these predictions.