2026 Midterms · House & Senate
Only a few dozen races are polled at all, so Polls is counted without a target. FEC funding covers every race with a filed candidate.
Method 1 · Poisson-binomial, exact
Senate
51%
Democrats win the majority
49%
Republicans win the majority
House
79%
Democrats win the majority
21%
Republicans win the majority
Method: Poisson-binomial with a national swing. Every race has its own blended win probability. Treating the races as independent coin flips with different odds, the exact distribution of “how many seats do the Democrats win” is the Poisson-binomial distribution, computed by convolution rather than sampling — so there is no simulation noise. The odds shown are the mass of that distribution at or above the majority line (218 in the House; 51 in the Senate, including the 34 Democratic caucus seats not on the ballot, with 50–50 going to Republicans through the Vice President).
Races do not miss independently, though: when polls and markets are wrong they tend to be wrong in the same direction everywhere. So the calculation is repeated across a grid of national swings — a shift added to every seat on the log-odds scale, drawn from a normal distribution with σ = 0.35 (a coin flip becomes roughly 59/41 at one sigma) — and averaged with normal weights. Without that layer, 470 near-independent contests would add up to false certainty.
Method 2 · 10,000 simulated elections
Senate
50%
Democrats win the majorityMethod 1: 51%
50%
Republicans win the majorityMethod 1: 49%
Share of runs ending at each number of dem caucus seats. Blue bars are Democratic control, red bars Republican control.
House
79%
Democrats win the majorityMethod 1: 79%
21%
Republicans win the majorityMethod 1: 21%
Share of runs ending at each number of democratic seats. Blue bars are Democratic control, red bars Republican control.
Both chambers together
From the same runs, so a bad night in one chamber is a bad night in the other. A 50–50 Senate counts as Republican control because the Vice President breaks ties.
How it works. Each of the 10,000 runs plays out all 470 races once. First a national swing is drawn — normal, σ = 0.35 on the log-odds scale, the same as the projection above — and added to every seat. Then each state gets its own smaller swing (σ = 0.15) shared by its House and Senate races, so neighbouring districts miss together. Finally every race is decided by one coin flip at its shifted probability, and the Democratic seats are counted. The bars show how often each seat total occurred; the headline odds are the share of runs on each side of the majority line.
Why the two methods agree so closely. Both start from the same seat odds and the same national swing, which carries most of the uncertainty. Method 1 is exact; this one adds the state layer and about half a point of sampling noise, so the two usually land within a point of each other and often round to the same figure. What the simulation adds is the full seat-count distribution and the joint outcome of both chambers.
35 races on the ballot (33 Class II plus the FL and OH specials) · Democrats need 51 because the GOP holds the vice presidency
51
Dem caucus
48
Republicans
Today: 47 – 53
Marker sits at 51 of 100 seats — the line for majority control.
Expected seats · blend of both methods
50.6 Dem caucus – 49.4 Republicans
Method 1 (Poisson-binomial mean) 50.6 · Method 2 (Monte Carlo mean) 50.5 Dem caucus
What it takes
Projected swing
Race call, not a probability. The large numbers give every seat to whoever leads it, tossups to neither side — a headcount if every favourite won. Expected seats above is the probability-weighted total: the mean of Method 1 and Method 2 from the two cards above, which agree to within a fraction of a seat. Includes holdovers not on the ballot: 34 D caucus, 31 R. Tossups are split 0 D-held and 1 R-held.
All 435 seats on the ballot · 218 for a majority
230
Democrats
202
Republicans
Today: 215 – 219 · 1 independent
Marker sits at 218 of 435 seats — the line for majority control.
Expected seats · blend of both methods
232.8 Democrats – 202.2 Republicans
Method 1 (Poisson-binomial mean) 232.8 · Method 2 (Monte Carlo mean) 232.8 Democrats
What it takes
Projected swing
Race call, not a probability. The large numbers give every seat to whoever leads it, tossups to neither side — a headcount if every favourite won. Expected seats above is the probability-weighted total: the mean of Method 1 and Method 2 from the two cards above, which agree to within a fraction of a seat. Gains and losses are flips versus the current holder of each district. Vacant seats count toward the party that last won them, so today's split reads 215–219 rather than the 214–218 of sitting members. Tossups are split 0 D-held and 3 R-held.
Races on the ballot
AL
1–6 · Sen Rep
AK
0–1 · Sen Dem
AZ
5–4
CO
5–3 · Sen Dem
FL
6–22 · Sen Rep
GA
5–9 · Sen Dem
IN
2–7
KS
1–3 · Sen Rep
ME
1–1 · Sen Dem
MA
9–0 · Sen Dem
MN
4–4 · Sen Dem
NJ
10–2 · Sen Dem
NC
5–9 · Sen Dem
ND
0–1
OK
0–5 · Sen Rep
PA
10–7
SD
0–1 · Sen Rep
TX
12–26 · Sen Dem
WY
0–1 · Sen Rep
CT
5–0
MO
2–6
WV
0–2 · Sen Rep
IL
14–3 · Sen Dem
NM
3–0 · Sen Dem
AR
0–4 · Sen Rep
CA
48–4
DE
1–0 · Sen Dem
HI
2–0
IA
2–1 · Sen Rep
KY
1–5 · Sen Rep
MD
7–1
MI
7–4 · Sen Dem
MS
1–3 · Sen Rep
MT
0–2 · Sen Rep
NH
2–0 · Sen Dem
NY
20–6
OH
6–9 · Sen Toss
OR
5–1 · Sen Dem
TN
0–9 · Sen Rep
UT
1–3
VA
7–4 · Sen Dem
WA
8–2
WI
3–5
NE
1–2 · Sen Rep
SC
1–6 · Sen Rep
ID
0–2 · Sen Rep
NV
3–1
VT
1–0
LA
1–5 · Sen Rep
RI
2–0 · Sen Dem
Projected seat flips
Democrats gain 26
Republicans gain 8
Click a state to pin it, then expand a race for source detail.
US
Hover a state to see its races. Click to keep it pinned, then expand a race to see what each source predicts.
No race in this state under the current filter.