How FuelRadar fuel price forecasts work

Which price are you looking at?

FuelRadar fuel price forecasts estimate how reported city averages may move. A station page can apply that movement to the station's own price. The resulting numbers can differ without contradicting each other: the city average describes a group of stations, while the station report describes one retailer and fuel grade.

Reported prices and estimates answer different questions

Figure	What it describes	What to check
Reported station price	The latest available report for one service station and fuel grade.	Station, fuel grade and report time. Confirm the bowser price before filling.
Reported city average	A measured average for the stations covered by that city report.	The city, fuel grade, sample and date. It is not an offer at every station.
City forecast	A dated statistical estimate of the city's future average.	The opening report, forecast date, midpoint, range and model evidence.
Station projection	The reference city's estimated movement applied to a local starting price.	Which city supplies the movement and whether the starting price is reported or estimated.
National average	A broader snapshot of reported prices for the selected fuel.	Coverage and update time. It is neither a station quote nor a future price.

The national forecast comparison lists separate capital-city outlooks. It does not turn those outlooks into one price that every Australian station is expected to charge. A national average can also differ from each capital's average because the samples and reporting times differ.

Why the station and city prices are different

For a current station projection, FuelRadar starts with a valid report for that station and the selected grade. It then adds each dated city estimate's change from the city's opening average. This is called anchoring: the starting station price changes, while the city model's movement stays the same.

Illustrative price example, not a forecast: suppose a city report opens at 233.5c/L and a station reports 239.9c/L. The next-day benchmark holds the city price steady, then the city model adds 2.0c/L on day two.

Illustration only: the same movement from different starting prices

Date relative to the report	City price or estimate	Change from city opening	Station price or estimate
Report date	233.5c/L reported average	0.0c/L	239.9c/L station report
Next day	233.5c/L no-change benchmark	0.0c/L	239.9c/L no-change benchmark
Day two	235.5c/L model estimate	+2.0c/L	241.9c/L model estimate

The day-two station calculation is 239.9 + (235.5 - 233.5) = 241.9c/L. The station and city estimates differ by 6.4c/L, but both rise by 2.0c/L from their own opening prices. This arithmetic does not prove that the retailer will make that change.

A station may change its price at a different time or by a different amount. Its previous weekday or price-change pattern is historical evidence, not a confirmed next change. Where a current valid report is missing, a local starting price learned from past reports must be labelled as an estimate; it must not be presented as a reported station price.

Tomorrow's no-change benchmark

The model's next-day benchmark carries the opening price forward. It means “use the last reported level as the comparison,” not “the retailer has promised to hold its price.” A benchmark issued yesterday is still based on yesterday's report. Check the report date before treating an overnight view as new information.

Compare that simple benchmark with the later statistical points and their ranges. Keeping tomorrow unchanged while showing a different day-two estimate is a deliberate model choice. It is not evidence that a station has scheduled a price change for day two.

Official WA quotes have a different source

Where supplied, FuelRadar marks FuelWatch WA's next-day price as an official notified quote for the matching station, grade and date. It replaces the statistical estimate for that date. It does not confirm the rest of the forecast.

FuelWatch explains its reporting rules: tomorrow's notified prices become available from 2:30pm, and retailers hold the notified price for 24 hours from 6am. These are Western Australian rules. Do not assume another state's next-day estimate has the same status.

Read the range before the midpoint

A midpoint is the model's single displayed estimate. The low and high are an estimated range, not guaranteed minimum and maximum bowser prices. If the range still includes the opening price, it does not establish a rise or fall. A range entirely above or below it is also an estimate, not a promise.

The model's signal score describes its internal signal strength. A score shown as a percentage is not the measured probability of getting a station price right. Longer forecasts involve more uncertainty, and a low point drawn on a chart does not confirm the cheapest day to fill.

A new high does not by itself establish a peak. The ACCC's petrol price-cycle guidance explains that retail cycles arise from retailer pricing decisions, vary in duration and do not include every retailer. Diesel and automotive LPG do not follow those petrol cycles.

What goes into a forecast

The plotted estimates start with reported daily city-price history. Eligible Australian trend forecasts can also use a dated terminal-gate wholesale-price change. A cycle forecast uses the reported history. The forecast's source explanation identifies when wholesale movement was used.

Oil prices, currency movements and cited news can appear in the separate market report. That commentary does not set the prices on the chart or confirm a cheaper day. News about rising costs and an estimated price path describe different evidence; neither guarantees a particular retailer's next move.

Why a forecast may have no fill or wait advice

FuelRadar withholds model-based fill or wait advice unless the current model meets the published replay checks against carrying the last price forward. A historical replay checks past dates using the model; it is not the same as proving future performance. Results for an older model, another forecast horizon or city averages do not automatically validate an individual station's next-change prediction.

Read the forecast accuracy and no-change comparison for dates, sample sizes, errors and limitations. The data sources and methodology explain the reporting context. An unavailable forecast means there is no usable current outlook to display, not that prices are certain to stay flat.

A useful check before your next fill

1. Choose the same fuel grade on the station and city pages.
2. Compare the report dates, reference city and source labels before comparing the numbers.
3. Check for an official WA quote where available; otherwise read future figures as estimates.
4. Use the fuel map to compare current nearby reports, distance and any membership conditions. Confirm the bowser price before filling.

Source guidance and FuelRadar's forecast behaviour reviewed 24 September 2026. The worked example is illustrative and does not use current station reports.