Local government elections · 4 November 2026

Who is favoured in your ward?

Loading the forecast…

National vote share, ward ballot

Forecast against the poll average and past results (%). Polls under-count independents and local parties, so their share comes from history.

PartyForecast90% rangePolls (±)2021 local2024 nationalWards favoured

Metro councils

Expected seats from the council seat formula (ward seats plus proportional top-up), the chance each metro has a largest party, and the chance no party has a majority.

MetroLargestChanceHungANCDAMKEFFSeats

Coalition maths in the metros

Chance that each group of parties holds a majority of council seats between them. This is arithmetic, not a prediction of which deals will be made.

Metro

All wards

Click a row to find it on the map. Column headings sort.

All municipalities

KwaZulu-Natal: polls and by-elections disagree

Outside eThekwini there are no polls. National polls put MK close to its 2024 result, but in the by-elections held since 2024 MK ran well behind it and the IFP and ANC well ahead. The published forecast treats those by-elections as a regional poll; the alternative below lets national polls decide. The real result could fall anywhere between the two.

KwaZulu-NatalBy-election-led (published)Poll-led

Does the method work?

We re-ran the method on the eve of the 2016 and 2021 elections, using only data available then, and counted correct ward winners. Each province was held out while fitting.

How the forecast works

  1. New wards. The IEC lists the voting districts inside each 2026 ward (Municipal Demarcation Board boundaries). Past results for those voting districts are added up inside the new wards.
  2. History. For each party, a model links a ward's 2024 national-ballot share and its 2021 local ward-ballot share to the next local result. MK's share of the old ANC local vote follows each ward's 2024 ANC/MK split.
  3. By-elections. More than 200 ward by-elections since the 2024 election show where parties are running ahead of or behind their 2024 result, by province and community type. Wards get a light adjustment from these, at the strength that worked best in the 2016 and 2021 tests. In KwaZulu-Natal outside eThekwini, which has no polls, the by-elections also set the parties' overall level, corrected for how far KZN by-elections were from the next local result in 2016 and 2021.
  4. Polls. Each party's overall level is set to a poll average that favours recent and larger polls and corrects for each pollster's lean in 2024 (only pollsters whose final 2024 poll ended within three weeks of election day). Metros with their own polls are pulled towards them. MK and the IFP take their national totals from the model rather than the polls, because KZN by-elections contradict the polls.
  5. Candidates. Parties with no candidate in a ward (IEC certified lists, as supplemented on 23 October 2026) get no votes there; their share goes to the parties that are standing.
  6. Seats. Council seats follow the local-government formula: ward and party votes combined, quota and largest remainders, ward winners first.
  7. Uncertainty. 2,000 simulations add poll error and municipal and ward-level surprises, sized from the 2016 and 2021 tests. A ward's chance is how often each party won it. The national 90% ranges come from the same simulations.

What it doesn't know

  • Candidates' personal following. The model knows which parties stand in each ward, not who the candidates are; independents are counted with "Other".
  • Turnout. A turnout shock is simulated (parties whose voters turn out reliably gain when turnout falls), but its size is an assumption.
  • Census 2022 ward demographics are not released as data, so community types are built from voting patterns instead.
  • New parties without a 2021 or 2024 result are counted with "Other".