Early Voting Adjusted Forecast: 2025 Federal Election

Last updated: 3 May Apr 2025 10:20 am because of final Prepoll & Postal votes from the AEC, inclusion of randomised polling error variable and forecasts now produced from a Monte Carlo Simulation. 

This forecast estimates what 2PP% the LNP would need to achieve on election day.

The forecast is based on the prepoll & postal voting to date (as reported by the AEC) and the state of the polls during the prepoll voting period. The forecast also uses trends on prepoll and postal voting between the ALP and LNP at the last election.

Updated model

This model now incorporates a randomised polling error variable. The settings for this polling error variable are based on the polling error from the 2019 election.

Forecasts  are now produced from a Monte Carlo simulation where we run this scenario through a simulator 10,000 times. A polling error is randomly generated for each simulation. Average values for each of the forecasts are then calculated from the results produced by the simulation.

This makes the forecasts more realistic as they comprehend how these scenarios could change in the event where the polls were not accurate.

For Election Day

If the election was a one-day cricket match

For those having trouble getting your head around this forecast, I've produced the cricket equivalent of the current state of play above.

Using One Day International statistics from the last 10 years and LNP 2PP% from all elections since 1940, I've converted the above scenario to what it would be like in a One Day International match.

From early votes to date

Based on Prepoll votes received from the AEC up to 1 May 2025 and current 2PP% polling averages.

Commentary

The LNP vote chase has fallen well short of the required rate and they now face an almost impossible 57.02% 2PP required of election day votes to pull off a highly unlikely victory. Our model estimates, based on current polling and allowing for potential polling errors, that the LNP will probably score 45.65% 2PP instead.

About the forecast

Why this forecast?

Currently, pollsters (and forecasters) attempt to estimate the 2PP% of the major parties as a way to determine who is likely to win the election. The assumption is that if the 2PP% estimates hold for election day then this will enable a reliable prediction to be made.

But this approach ignores the fact that in the weeks leading up to election day, millions of votes have already been cast, where the 2PP polling performance of the major parties can be different to what they ultimately achieve on polling day. This inevitably undermines polling forecasts for election day.

What's required is an updated forecast that takes into account how many people have voted early and the likely way they would have voted.

It's a bit like limited-overs cricket

This forecast aims to estimate the LNP 2PP% required for victory, to be achieved on election day based on their current estimated 2PP%, prepoll & postal votes so far, and the estimated number of votes remaining to be cast in the election.

The logic here is depending on how many votes have been cast so far and how each party is performing in the polls at the time, this can substantially affect the required 2PP% of votes that need to be achieved for the remainder of the election campaign.

A good analogy to this forecast is limited-overs cricket where (for the team batting 2nd) the required run rate from the remaining overs changes as the game progresses. If the chasing team is ahead of the initial required run rate, they are more likely to gain victory. Conversely, if the chasing team is behind the initial required run rate, then they are less likely to win and more likely to lose the match.

To continue our limited overs cricket analogy, for pollsters ignoring prepoll and postal numbers, this is like continuing to base your match prediction on the required run rate for the team batting at the start of their innings, whilst the innings progresses and their scoring rate changes. 

Granted, in elections, unlike cricket where we can see the runs scored each over, we cannot see the votes cast for each party during prepoll until after the election.

However, given that polls close to polling day can usually predict the voting intention within their margin of error, therefore we should be able to make reasonable estimates of 2PP votes during prepoll if we use 2PP% estimates (derived from poll aggregators) applied to the number of prepoll votes cast so far.

Now, if we can set a required 2PP% for victory (akin to the required run rate in limited overs cricket) then this enables us to calculate a 2PP% required for the remainder of the campaign and on election day, similar to the run rate required from the remaining overs for the team batting second in limited overs cricket.

Calculations

All voting statistics have been sourced from the Australian Electoral Commission.

Pre-election values

Current state of play

Please note; these values do not have a polling error applied to them.

Projection for the remainder of the campaign

Please note; these values do not have a polling error applied to them.