Home

How the NZ election 2026 seat forecast works

The forecast plays out the 7 November election thousands of times in your browser, from the poll you choose. Every percentage on it is how often something happened across those simulated elections.

1. Party vote

It starts from the weighted average of the 14 polls since 1 September: each poll counts less the older it is (its weight halves every 10 days before the newest poll), and a pollster's polls share their weight, so one that polls every week doesn't outweigh the rest. You can start from any single published poll instead. Every poll is listed on the polls page. You can also set any party's vote yourself, or let a share of Opportunity's voters switch late.

2. Polling error

Polls miss. In each simulated election every party's vote moves by a random amount, about as far as New Zealand polls usually end up from the result. Bigger parties move more in points. Optionally, the whole National-led side (National, ACT and NZ First) can do a bit better or worse together, and the Labour-led side (Labour, the Greens and Te Pāti Māori) the opposite, as happens when polls miss in one direction. You can switch to a smaller error, a big polling miss, or turn the simulations off for one exact result.

3. Electorates

All 71 electorates (64 general and 7 Māori) are decided in every simulated election. Each candidate starts with a chance built from the 2023 result (on the new boundaries where they changed), retirements and any published electorate poll, all listed with sources on each electorate's page. That chance moves with their party's vote: a party polling above its average does a little better in every seat. Calling a seat fixes it for one candidate.

4. Seats

Seats are shared out with the Sainte-Laguë formula over 120 seats, among parties with at least 5% of the party vote or an electorate. A party that wins more electorates than its party vote earns keeps them, and Parliament grows by the difference (the overhang). A party can't fill more list seats than it has list candidates: Te Pāti Māori has one, so list seats it earns beyond that are left empty.

5. Who governs

Each simulated Parliament is checked against its own majority line. A National-led majority means National, ACT and NZ First together hold it; a Labour-led majority means Labour, the Greens and Te Pāti Māori do. Opportunity and independents count as kingmakers unless you move them. The diagram shows one simulated Parliament with the most likely outcome, closest to the middle result, so you can see actual MPs. When the top two outcomes are within 10 points the forecast calls it too close to call.

6. Repeatable results

The random numbers are fixed, so the same settings always give the same result, and any change you see comes only from what you changed. You can share any set of settings as a link.

How this differs from other forecasts

ChoiceThis forecastOthers
Starting voteOne poll you choose, or the averageKiwi Polls, the Herald–Motu Poll of Polls and election2026.nz use averages
Pollster leanPolls taken as publishedHerald–Motu and election2026.nz adjust for each pollster's past lean
OverhangCounted, so the majority line movesHerald–Motu assumes a 120-seat House
OpportunityA kingmaker unless you move itCoalition forecasts often put it on one side
ElectoratesAll 71 simulated candidate by candidatePoliwave uses a uniform swing from 2023
Polling errorFixed at a typical size unless you change itelection2026.nz widens it the further out the election is

The forecast itself has a section that reruns the numbers changing one of these choices at a time, so you can see how much each matters. See also insights from the current data.

Sources

Polls as published by each pollster (see the polls page). Candidates and party lists from the Electoral Commission. 2023 results from the Electoral Commission's official count. Electorate boundaries from Stats NZ (CC BY 4.0) and coastline from LINZ (CC BY 3.0 NZ). The forecast is an illustration of uncertainty, not a prediction.

Updated 2026-10-28.