TL;DR: A disciplined weekly driver who waits for the bottom of the cycle saved about $100 a year on ULP over one who just fills when the tank’s empty — 180.90 c/L versus 184.84, annualised over 18 months, not verified across several clean years. Perfect timing would save $469, but you can’t wait for a bottom you can’t see coming, and one fill caught at the crest wipes out a month of the real saving.
“Wait for the trough” is the most-repeated piece of fuel advice in Brisbane. It’s also unfalsifiable folklore until someone runs the tank empty against a calendar. So I did — every ULP price change 430-plus Brisbane metro stations reported over 474 usable days, two strategies, no vibes.
The claim
Two drivers, same car, same commute. One waits for the dip:
“I’m guessing it cycles as an incentive to ‘wait for the trough’- but get ‘caught out by the crest’” — ‘forumer17’, Whirlpool
The other doesn’t play the game — fills when the needle’s near E, whatever the price. The question is narrow: over a year, how much does the waiting actually earn?
What we checked
I built a daily median ULP price for Brisbane metro — the middle station out of 434 to 441 reporting each day — from the QLD Government feed, using each station’s real reported price, carried forward until it next changed. That’s the metro median, not any one driver’s bowser. Window: 15 Jan 2025 to 30 Jun 2026 — 532 calendar days, 474 usable after cutting the March–late-April 2026 ingest gap. Over that window a turning-point pass finds 9 clean interior troughs, a median 38 days apart. Those 9 troughs give 8 inter-trough intervals; the median runs off those 8, and it’s the 38 days. A naïve read might expect ~12 troughs in 474 days at a 38-day cadence — the missing laps are the two intervals we don’t count: the early-window run into the first trough and the one interval that straddles the excised March–April gap (a 96-day span that isn’t a real cycle, just a hole). Drop those two and the eight that remain are the honest cadence. (If you read our cycle-length piece, it counts 14 troughs — that’s over a longer 25-month window back to June 2024. This post starts in January 2025, so it sees fewer laps of the same cycle, not a different one.) The swings themselves run a median of about 45 c/L trough-to-peak — the same amplitude every one of our cycle reads lands on.
Then two strategies, both buying 50 L a week (2,600 L a year):
- Strategy A — fill when empty: buy every 7 days, cycle-blind. Averaged over all seven weekday start-offsets so the luck of which day you run dry doesn’t skew it (that luck is worth 2.05 c/L — near noise).
- Strategy B — disciplined trough-timer: fill when today’s price is at or below the trailing 28-day minimum plus 2 c, but force a fill once 10 days pass because the tank’s empty. This is a real weekly filler watching the app, not a fortune-teller.
I also computed a perfect-trough ceiling: buy every cycle’s fuel on its exact minimum day. Nobody can do this — you can’t know a bottom held until it’s passed, and your tank empties first — but it bounds the prize.
What the data shows
Fill-when-empty landed at 184.84 c/L — that’s the offset-averaged Strategy A, so both figures below compare against the same cycle-blind baseline, apples to apples. The disciplined waiter landed at 180.90 c/L — 3.9 c/L cheaper, which on 2,600 L is about $100 a year ($102).
Perfect timing would have hit 166.79 c/L — $469 a year. That’s the number folklore quotes. It’s real, and it’s a trap: it beats cycle-blind filling by 18.0 c/L, but only by buying every litre on the exact minimum day of every cycle. A weekly filler physically can’t. The tank forces a buy inside ten days, long before you can be sure the price has bottomed — so most of that 18-cent edge is unreachable, and the disciplined waiter captures under a quarter of it. The honest, actionable number is $100, not $469.
The catch
Here’s the part the folklore leaves out. The saving is fragile because of what the crest costs.
The average cycle peak was 201.4 c/L. One 50 L fill caught up there costs +16.6 c/L, or $8.30, versus just filling blind. That single mistake erases about four weeks — a month — of realistic Strategy B’s savings.
So the whole edge depends on discipline and flexibility. Run the sensitivity. Start from the $102 win. Now let the driver get caught empty at the crest on about one cycle in two. That’s a frequency, not a litre-share, and the two don’t match one-for-one: a caught cycle forces a full 50 L tank at the 201.4 c/L peak, while a good cycle is a single cheap trough buy — so the expensive cycles pull more litres than their count suggests. Weight it by the fuel actually bought and about 36% of the year’s litres land at that peak instead of the 180.90 disciplined rate. Blend those — 36% at 201.4, the rest at 180.90 — and the realised rate climbs to about 188.3 c/L, which is about 3.4 c/L above the 184.84 you’d have paid filling blind. On 2,600 L, that’s roughly minus $90 a year: Strategy B stops winning and goes negative. The strategy pays only if you can actually defer the fill when the price is high — and if you can’t, you’re better off not trying to time it at all.
Verdict
Waiting for the trough beats filling on empty — but by about $100 a year for a disciplined driver, not the $469 the perfect-timing math implies, and only if you don’t get caught at the top. Call it true-but-small, and reversible if you’re sloppy.
This whole exercise answers one job — is now a good time to fill? — for the disciplined weekly commuter watching the cycle. But the bigger lever isn’t timing at all; it’s which station, which is a different job: where’s the cheapest fuel near me right now? Strategy A here is the median Brisbane station on a random day, and a driver who already checks the cheapest price near them beats that median before timing enters the picture, which shrinks the marginal value of the whole cycle game. Get the cheap station first. Time the cycle second, if you’ve got the tank to wait.
Honest limit: this is 474 usable days out of an 18-month window, annualised — not several clean years — with the March–late-April 2026 gap cut out. It’s the metro median for ULP/91 only: no E10, no premium, no docket discounts, no loyalty lock-ins. Different trough thresholds shift the perfect-trough figure by a few c/L, but not the shape of the answer: realistic timing captures a small slice of the swing, and mistiming it costs more than getting it right saves.