Victoria88-26

The 98th Percentile forecast of the 2026 Victorian state election. All 88 Legislative Assembly electorates, voting on 28 November 2026.

Author

John Ward (The 98th Percentile)

Statewide first preferences

Per cent of the formal vote. The range under each figure is the plausible range: the model puts a nine-in-ten chance on the result falling inside it. Figures may not sum to 100.

First preferences over time

Estimated first-preference vote, statewide
Model estimate at each poll, with house effects removed. Shaded bands are plausible ranges.
Pollsters did not prompt for One Nation in Victorian state polls until early 2026. Before 1 February 2026 its vote sits inside Others, which is why Others falls at the break. The points at 28 November are the election-day forecast; their ranges are wider because opinion can still move and the polls as a group can miss.

First-preference voting estimates across all electorates

These are estimates of the first-preference vote, not predictions of who wins each seat. Victorian elections are decided on preferences. Each cell shows the median estimate, with the plausible range beneath it. Click a column heading to sort.

The party with the highest median in an electorate is not always the party most likely to lead it. The probability of leading comes from the full set of simulated elections, where each party’s figures move together.

Win probabilities by party across electorates

Victorian elections are decided on preferences, not first preferences. A candidate who leads the first count can still lose the seat once the other candidates are excluded and their ballots transferred. Win probabilities therefore need an estimate of how preferences will flow in every electorate.

Those flows depend on two things that do not yet exist. The first is the candidate field, which is fixed when nominations close on 9 November. The second is the parties’ how-to-vote cards, which show where each party directs its preferences. Publishing win probabilities before then would mean inventing both.

Win probabilities: published after the how-to-vote cards are registered

This section will show, for every party in every electorate, the probability of winning the seat, with seat totals for the chamber. Expected mid-November 2026, after nominations close on 9 November.

Every published figure is then scored against the result on the Track Record page, misses included.

What changed in the latest release

This is the first release. From the next release on, this section records what moved and why.

How it works

Three stages. The first pools every published Victorian state poll into a statewide estimate for each party, correcting for each pollster’s lean and for how far the polls as a group have missed at past elections. The second distributes each statewide estimate across the 88 electorates. The third, still to come this cycle, simulates the flow of preferences to decide who wins each seat. How the model works → · Victoria88-26: the Victorian application →

Data

The full first-preference estimates for every party in every electorate, with the plausible range on each figure, are published as data with each release:

Free to use with attribution to The 98th Percentile. The data files name the model run they came from.

Use of AI

I used large language models, Anthropic’s Claude Opus 5.5 and Claude Fable 5, to test this work for narrative flow, logical consistency, bias and errors. They also assisted with drafting and with the site’s code. Responsibility for the content is mine alone.

John Ward (The 98th Percentile)