How accurate is our fuel price forecast?
23 Aug 2026 to 14 Nov 2026 · 113 cities · 49,002 next-day checks · tested 14 Nov 2026
Next-day city estimates missed by 0.9 c/L on average over the 30 days to 14 Nov, closer than assuming prices stay the same (1.0 c/L).
The record in one minute
Replay of 23 Aug 2026 to 14 Nov 2026
All next-day checks, 23 Aug 2026 to 14 Nov 2026: our estimate missed by 0.9 c/L on average against 1.0 c/L for “no change”, 0.1 c/L closer; 85% of next-day city averages fell inside the range we showed.
Our next-day, 3-day and 7-day estimates beat the simplest guess, that prices stay the same; 14-day and 30-day estimates were only about as close as that guess. Waiting when we said wait saved money out to a week. Past a week, waiting cost money, so 14-day and 30-day estimates are only a rough guide. Our weakest point is direction a month out: when prices moved, we called it right 38% of the time.
Did waiting pay off?
Wait calls · average result a litre
Yes, out to a week. Past a week, waiting when we said wait cost money.
| Forecast point | Average result a litre | Wait calls |
|---|---|---|
| Next day | 1.8 c/L savedPaid 82%Paid off | 2,121 |
| 3 days | 2.2 c/L savedPaid 75%Paid off | 7,652 |
| 7 days | 2.0 c/L savedPaid 64%Paid off | 9,968 |
| 14 days | 0.1 c/L lostPaid 53%Cost money | 8,667 |
| 30 days | 3.4 c/L lostPaid 45%Cost money | 6,098 |
How a wait call is counted
A wait call is a day when the estimate was at least 1 c/L below that day’s city average. The result is that day’s average less the average on the forecast date, averaged across the calls: “saved” means waiting saved money, “lost” means it cost money. “Paid off” is 1.0 c/L or more saved.
How close were the estimates?
Average miss in c/L · lower is closer
Closer than “no change” the next day, 3 days ahead and 7 days ahead; not 14 days ahead or 30 days ahead.
- Our estimate
- Assume no change
| Forecast point | Average miss | In range |
|---|---|---|
| Next day | Our estimate: 0.9 c/LAssume no change: 1.0 c/LRight direction 62% | 85% |
| 3 days | Our estimate: 2.0 c/LAssume no change: 2.4 c/LRight direction 64% | 89% |
| 7 days | Our estimate: 4.6 c/LAssume no change: 4.9 c/LRight direction 51% | 87% |
| 14 days | Our estimate: 8.0 c/LAssume no change: 8.1 c/LRight direction 45% | 81% |
| 30 days | Our estimate: 16.5 c/LAssume no change: 16.6 c/LRight direction 38% | 61% |
What these figures mean
Average miss is the gap between the estimate and the city average reported later. In range means the reported average fell inside the low-to-high range we showed. Right direction counts only days when prices moved, by 1c or more.
Next-day accuracy in each capital
Next-day average miss · best first
Brisbane was closest for U91, missing by 0.6 c/L on average; Perth missed most, by 1.4 c/L.
| City | Average miss |
|---|---|
| Brisbane | 0.6 c/L |
| Darwin | 0.6 c/L |
| Adelaide | 0.7 c/L |
| Canberra | 0.8 c/L |
| Hobart | 0.8 c/L |
| Sydney | 0.8 c/L |
| Melbourne | 1.1 c/L |
| Perth | 1.4 c/L |
Regional cities (105)
| City | Average miss |
|---|---|
| Alice Springs | 0.5 c/L |
| Bunbury | 0.5 c/L |
| Busselton | 0.5 c/L |
| Murray Bridge | 0.5 c/L |
| Port Hedland | 0.5 c/L |
| Whyalla | 0.5 c/L |
| Albany | 0.6 c/L |
| Cairns | 0.6 c/L |
| Esperance | 0.6 c/L |
| Geraldton | 0.6 c/L |
| Port Augusta | 0.6 c/L |
| Port Pirie | 0.6 c/L |
| Sunshine Coast | 0.6 c/L |
| Toowoomba | 0.6 c/L |
| Townsville | 0.6 c/L |
| Warwick | 0.6 c/L |
| Yeppoon | 0.6 c/L |
| Ballina | 0.7 c/L |
| Broome | 0.7 c/L |
| Central Coast | 0.7 c/L |
| Coffs Harbour | 0.7 c/L |
| Forster–Tuncurry | 0.7 c/L |
| Gladstone | 0.7 c/L |
| Grafton | 0.7 c/L |
| Gympie | 0.7 c/L |
| Hervey Bay | 0.7 c/L |
| Karratha | 0.7 c/L |
| Kurri Kurri | 0.7 c/L |
| Lismore | 0.7 c/L |
| Mount Gambier | 0.7 c/L |
| Muswellbrook | 0.7 c/L |
| Nambour | 0.7 c/L |
| Newcastle | 0.7 c/L |
| Nowra–Bomaderry | 0.7 c/L |
| Port Lincoln | 0.7 c/L |
| Airlie Beach – Cannonvale | 0.8 c/L |
| Batemans Bay | 0.8 c/L |
| Bundaberg | 0.8 c/L |
| Burnie–Somerset | 0.8 c/L |
| Dalby | 0.8 c/L |
| Dubbo | 0.8 c/L |
| Kalgoorlie–Boulder | 0.8 c/L |
| Mackay | 0.8 c/L |
| Maitland | 0.8 c/L |
| Maryborough | 0.8 c/L |
| Moama | 0.8 c/L |
| Rockhampton | 0.8 c/L |
| Shepparton–Mooroopna | 0.8 c/L |
| Swan Hill | 0.8 c/L |
| Tamworth | 0.8 c/L |
| Taree | 0.8 c/L |
| Tweed Heads | 0.8 c/L |
| Victor Harbor | 0.8 c/L |
| Warrnambool | 0.8 c/L |
| Wodonga | 0.8 c/L |
| Wollongong | 0.8 c/L |
| Albury | 0.9 c/L |
| Bacchus Marsh | 0.9 c/L |
| Bowral–Mittagong | 0.9 c/L |
| Broken Hill | 0.9 c/L |
| Cessnock | 0.9 c/L |
| Colac | 0.9 c/L |
| Crafers–Bridgewater | 0.9 c/L |
| Devonport | 0.9 c/L |
| Echuca | 0.9 c/L |
| Emerald | 0.9 c/L |
| Gawler | 0.9 c/L |
| Geelong | 0.9 c/L |
| Gracemere | 0.9 c/L |
| Horsham | 0.9 c/L |
| Kempsey | 0.9 c/L |
| Lithgow | 0.9 c/L |
| Morisset–Cooranbong | 0.9 c/L |
| Morwell | 0.9 c/L |
| Mount Barker | 0.9 c/L |
| Port Macquarie | 0.9 c/L |
| Portland | 0.9 c/L |
| Ulverstone | 0.9 c/L |
| Wagga Wagga | 0.9 c/L |
| Armidale | 1.0 c/L |
| Bairnsdale | 1.0 c/L |
| Bathurst | 1.0 c/L |
| Bendigo | 1.0 c/L |
| Goulburn | 1.0 c/L |
| Launceston | 1.0 c/L |
| Orange | 1.0 c/L |
| Singleton | 1.0 c/L |
| Torquay – Jan Juc | 1.0 c/L |
| Traralgon | 1.0 c/L |
| Wallan | 1.0 c/L |
| Griffith | 1.1 c/L |
| Melton | 1.1 c/L |
| Mount Isa | 1.1 c/L |
| Ocean Grove – Barwon Heads | 1.1 c/L |
| Raymond Terrace | 1.1 c/L |
| Sunbury | 1.3 c/L |
| Warragul | 1.3 c/L |
| Ballarat | 1.4 c/L |
| Wangaratta | 1.4 c/L |
| Blue Mountains | 1.5 c/L |
| Moe–Newborough | 1.5 c/L |
| Gold Coast | 1.6 c/L |
| Mildura | 2.1 c/L |
| Sale | 4.7 c/L |
| Castlemaine | 5.4 c/L |
Where the forecast is weakest
From the latest replay
- Perth and Melbourne. They had the largest next-day U91 misses of the capitals, 1.4 c/L and 1.1 c/L on average.
- Sale E10. The largest next-day miss of any city and fuel, 14.0 c/L on average.
Changes to the forecast, newest first
- Week-ahead outlooks show their own accuracy line.
- A new city model runs in test and goes live only if it beats today’s model over 28 days of new prices.
- Price-cycle peaks need a 2c fall to confirm.
How we test the forecast
Every week · the latest 90 days
We rerun today’s method on past dates, using only prices known at the time, and score each estimate against the city average reported later.
Start with past reports
Use only city prices that were available before the forecast date.
Estimate the next price
Apply the same forecast method used for the current outlook.
Check the later price
Compare the estimate with the city average reported afterwards.
- These checks recalculate past dates with today’s method. They are not a record of the exact forecasts people saw on those dates.
- The checks use past city averages. They do not recreate every wholesale price change or event that may affect today’s outlook.
- The results compare city averages. They cannot tell you the exact price or availability at one service station.
Forecast accuracy questions
What does average miss mean?
The gap between our estimate and the city average reported later, in cents a litre. A 2 c/L miss is much smaller than a 20 c/L miss.
Why compare with prices staying level?
Because it is the simplest guess there is. If an estimate cannot beat “prices stay the same” over the same dates, it is not telling you anything new.
Why are some results missing?
Too few checks. A city and fuel grade appears once at least three dates have a later reported average to compare.
How often does this page update?
Every week. We check the latest 90 days each week, and the newest completed check replaces the previous results.
Does a small past miss guarantee the next forecast?
No. Price-cycle resets, wholesale changes and disruptions can move the next city average more than recent results suggest.



