ELECTORATE MODEL
Party-vote swing
Each seat starts with The Tally Room’s estimated 2023 candidate-vote shares on the new boundaries. For each of its two leading parties, we add the change in national party vote since 2023, multiplied by the swing strength. At 100%, a national gain of 3 percentage points adds 3 points to that contender’s local score. At 0%, the historical shares stay fixed.
local score = 2023 local share + strength × (your party vote − 2023 national share)
Scores are clipped to 0–100, then rescaled with the fixed residual for other candidates to total 100%. The leading modelled contender wins the seat. Ties retain the historical winner with a review flag. Rounded source shares can differ slightly from the published historical margin.
Hutt South: the baseline is National 44.1%, Labour 41.6%. National’s 2023 nationwide lead was 11.16 points. At full strength, the seat changes hands when that national lead falls below about 8.66 points.
This two-contender model is not backtested; its margins are not win probabilities. Only those two parties can win automatically; other party-vote changes affect Parliament but cannot introduce a third local contender. It does not estimate new candidates, tactical voting, incumbency changes, Māori-specific swings or electorate polling. If a historical contender is not standing, as with Opportunity in Ilam, the seat retains its historical call and is flagged for manual review. Review races with new challengers, especially Te Tai Tokerau. Port Waikato uses the source’s party-vote proxy.
The map colours reflect your final picks, including overrides. “Changing hands” compares their sitting parties with the boundary-adjusted 2023 winners; gold outlines mark those seats. General and Māori electorates use separate map layers because they overlap geographically. Auckland and Wellington have close-up views. The Chatham Islands are moved west into a labelled inset at the same scale. Boundaries are adapted from the final August 2025 Representation Commission / Stats NZ data (CC BY 4.0), clipped to Stats NZ’s coastline and simplified for display. Manual selections and “Use poll leader” pin a seat against future model updates. “Use model” releases one pin; “Clear overrides” releases all pins. Switching to Manual restores your manual calls. JSON exports save mode, strength and pins; old scenarios open in Manual mode. CSV electorate rows pin overrides when the model is active.
Official 2023 national baseline ↗CALCULATION
From two votes to Parliament
Kerrin Leoni: a Tāmaki Makaurau win is allocated under Labour, then displayed as independent with no replacement Labour seat. This follows her declared intention; formal recognition happens after the election. List-seat acceptance is not modelled.
Enter party-vote shares totalling 100% and choose one winner for each of the 71 electorates. Parties qualify for proportional seats with at least 5% of valid party votes or at least one electorate win.
The calculator uses the standard Sainte-Laguë divisors (1, 3, 5, …) to allocate 120 seats. Independent and non-party-vote winners are first deducted from that pool. Electorate MPs fill each party’s entitlement before list MPs.
If electorate wins exceed the entitlement, that party keeps the extra seats as an overhang. Available list candidates are capped using the published 2026 list sizes, subtracting any elected constituency candidates also on the list. Unfilled list entitlements remain vacant.
An exact tie for the last allocated seat is flagged. The real allocation resolves it by lot; this tool uses alphabetical party code order as a reproducible illustration. It does not model postponed electorate elections, recounts, withdrawals after nomination, or component-party arrangements.
Electoral Act, sections 191–193 ↗EVIDENCE & ASSUMPTIONS
Know what is in the model
“Historical margin” is the winning candidate’s lead over the runner-up in votes. On changed boundaries it describes the old electorate only. The displayed notional percentage-point lead is an estimate on the new boundaries.
Candidate choices follow the final nomination tables compiled on Wikipedia, with list ranks checked against official lists (469 electorate candidates, 443 list candidates). Independent and unregistered candidates are grouped for seat allocation.
The coalition counter is a simple majority of filled seats, assuming all MPs vote. It does not predict confidence-and-supply arrangements, abstentions or negotiations.
YOUR WORKSPACE
Bring your own assumptions
Changes save automatically in this browser. Export JSON to save or transfer a complete scenario, then import it later. Exporting includes the votes, electorate calls and selected coalition.
You can also import a CSV with columns type,name,value. Use party,NAT,30 for a vote share and electorate,Epsom,ACT for an electorate call. Party codes appear beside the inputs. CSV imports merge into the current scenario; JSON imports require a complete scenario.
Missing parties in a published poll are unreported, not observed zeros. Applying a poll sets unreported named parties to a zero assumption and puts any remainder in Other. If rounding pushes the reported total above 100%, it explicitly scales the shares and labels the change.
Data is a fixed snapshot as at 28 October 2026; no automatic updates are scheduled. Imported data stays in your browser. Poll and electorate source links open the original publishers.
SOURCE LIBRARY
Follow the evidence