Victorian state election · 28 November 2026 · figures as at 29 October

How the forecast works

The forecast doesn't predict one result. It runs the election 40,000 times with slightly different assumptions each time and counts how often each outcome happens. Here is what happens inside one of those runs.

The whole machine

37 pollssince 2023 Fundamentals12 years in office 2022 resultsall 88 seats Statewide voteeach party's share average blend 88 seat votesfirst preferences share out starting point repeated 40,000 times Random shockspolls can be wrong Count each seatwith preferences One parliament88 winners again new votes winners
Everything left of the dashed box happens once. Everything inside it happens 40,000 times, each with different random shocks, and the forecast is the tally of those 40,000 parliaments.
Step 1

Average the polls

Every published poll goes into a weighted average. A poll counts for more when it is recent (its weight halves every 21 days), has a bigger sample, and comes from a pollster with a good record. Each pollster's lean is measured against the others and removed. For example, Roy Morgan tends to show the Coalition about 2 points lower than other pollsters do.

CoalitionLaborOne NationGreensOther 29.5%24.3%22.5%14.3%9.4%
The current average of first-preference votes. Coalition, Labor and One Nation are within seven points of each other, so the race is three-way rather than the usual two-party contest.
Step 2

Check it against history

Polls have missed before. In 2018 they understated Labor by almost 4 points. So the poll average is blended with a "fundamentals" estimate. That estimate starts from Labor's 2022 result and subtracts for a government seeking a fourth term, for Labor also governing in Canberra, and for recent by-election swings. Four weeks out, the polls get 74% of the weight. Here the two agree, so the blend barely moves anything.

44 45 46 47 48 49 50 Labor two-party-preferred vote (%) vs Coalition Polls 46.4weight 74% Fundamentals 47.5weight 26% Blended 46.7
The further from election day, the more weight the fundamentals get. A year out, the polls would get only about 42%.
Step 3

Turn the statewide vote into 88 seats

Each seat starts from its own 2022 result. Each party's statewide change since then is shared out across seats. One Nation, which barely ran in 2022, gets more of its vote in rural and outer-suburban seats. That pattern follows where it polled best at the 2025 federal election. The parties it takes votes from lose most in those same seats.

Werribee, first preferences (%) Labor 45.429.8 Liberal 25.323.4 One Nation did not run26.2 Greens 6.89.3 Other 22.511.2
2022 actual2026 projection (average across runs)
Werribee in Melbourne's west was a safe Labor seat in 2022. In the projection, Labor still tops the first-preference count, but One Nation is less than four points behind.
Step 4

Count the preferences

Victoria uses preferential voting, so a seat is decided by a full count, just as the Electoral Commission would run it. The candidate with the fewest votes drops out. Their votes then pass on according to how that party's voters usually rank the others. This repeats until two candidates remain. The flow that matters most this year is where Liberal votes go once a Liberal candidate is out.

First preferencesGreens outOthers outLiberals out 29.826.223.411.29.3 35.527.624.512.3 38.933.028.0 Labor 50.6One Nation 49.3 50% Greens votes mostly go to Labor Other votes split several ways Liberal votes: 58% One Nation, 42% Labor
LaborOne NationLiberalOtherGreens
Werribee on the average projection. The Liberals finish third, and their preferences almost erase Labor's lead. Across all 40,000 runs Labor wins Werribee 42% of the time, One Nation 36% and the Liberals 21%.
Step 5

Shake it before every count

Before each run, the model adds random shocks at four levels. They are sized so the spread of results matches how far real seats strayed from forecasts in 2018 and 2022. The biggest is the first: when the polls are wrong, they tend to be wrong everywhere at once.

Statewide polling errormoves all 88 seats together · about ±2.6 two-party points, with occasional bigger misses Regional shockshared by similar seats, e.g. all outer-suburban seats Seat shocklocal candidates and campaigns · about ±4 points per seat Preference flowshow strongly each party's voters favour the others
Each layer sits inside the one above. A single run might have the polls overstating One Nation statewide, the outer suburbs swinging harder to Labor, and a strong local campaign in one seat, all at once.
Step 6

Repeat 40,000 times and count

Each run produces one parliament. Stacking all 40,000 gives the range of outcomes. The Coalition's median is 39 seats, short of the 45 needed for a majority. That is why a hung parliament is the most likely result.

Labor majority 6% Labor largest, hung 20% Coalition largest, hung 45% Coalition majority 28%
Top: how often each Coalition seat total came up. Bottom: what share of the 40,000 parliaments fell into each outcome. "Largest" means the bigger of the two major parties in a hung parliament. Who governs then depends on negotiations the model doesn't simulate.
Why one poll moved the Coalition's odds from 21% to 28%

The 22 October forecast had no poll newer than about two weeks. Newspoll (15–22 October) then put the Coalition on 33, its highest figure in any poll this year. As the freshest poll it carries about a quarter of the weight in the average, and it lifted the Coalition by 0.7 points and cut One Nation by 0.5.

That small shift matters because the Coalition's median sits just below the 45-seat line, where the odds change fastest. Many seats are Liberal v One Nation contests that flip together. Each point the Coalition takes from One Nation is worth about 11 points of majority chance. With an average that reacts more slowly to new polls, the odds would still have risen, to 26% instead of 28%. Most of the jump comes from that sensitivity rather than from overweighting one poll.

The two settings that matter most

Two assumptions explain most of the gap between this forecast and others. Each is shown below at the setting used and at plausible alternatives (12,000 runs each).

Where Liberal preferences go

When a Liberal is excluded in a Labor v One Nation seat, what share of those votes goes to One Nation?

Share to One NationLabor seatsOne Nation seats
48% (what RedBridge's respondents say)3210
58% (used)2813
65% (South Australia, March 2026)2616
How bunched-up One Nation's vote is

The same statewide vote wins more seats if it is concentrated in fewer places.

ConcentrationOne Nation seatsCoalition majority
Half as concentrated1136%
As modelled1328%
1.5 times as concentrated1622%

Rerun on the 2018 and 2022 elections using only the polls available beforehand, the method picked the right party in about 87% of seats. Explore the full forecast, seat by seat, on the forecast page.