TL;DR: The felt injustice is real, and it splits in two. A single servo rockets up in one report — the median restoration jump is +45 c/L, and there are 6,772 such jumps of 25 c/L or more on record. But the whole market’s median level doesn’t finish rising for about 9 days, because different servos restore on different days, then it grinds back down over about 32. That’s a 3.6x asymmetry on the market median across 14 clean cycles: down takes roughly three and a half times as long as up.
Every driver who’s watched the bowser for a year already knows the shape of this, and Vinnie on Whirlpool put it better than any chart:
“when the price is going down it takes a week to ten days to come down by 20-30 cents, but when it goes up its just matter of hours to go up by 30 cents” — ‘Vinnie’, Whirlpool
Up in hours, down over a week. That’s the grievance. This post measures it — and it turns out Vinnie is describing two different things at once, both true.
The claim
Prices rocket up and drift down. The rise is quick and brutal; the fall is a slow grind that always seems to stall a few cents above where the last trough was. The asymmetry is the whole reason timing a fill pays.
What we checked
Every price here is the Queensland Government price feed — the actual price each station reported charging, not a driver’s guess at the bowser.
We took every ULP (regular 91) report from the Brisbane-metro stations inside 30 km of the Brisbane GPO — 473 of them, the same in-range set our same-corner and cheapest-suburbs posts use — from January 2024 to now. For each station we carried its last known price forward day by day, then took the median across the panel for every calendar day. That gives a clean daily series with no gaps: 933 days, 156 to 439 stations present per day (thin only in early 2024; 380-plus daily later in the run). Then a turning-point scan walked the peaks and troughs — requiring at least a 15 c/L retracement before it would call one, so noise doesn’t get counted as a cycle — and we measured every up-leg and down-leg in days and in cents per day. That yielded 16 trough-to-peak-to-trough cycles, 14 of them once the two distorted stretches below are held out.
One reconciliation up front, for anyone who read our cycle-mechanics post four days ago. That post measured the same swing — Brisbane ULP market median, trough to peak — and put the up-leg at about 7 days and the fall at “more than four times slower”, a 37-day lap. This one says about 9 days up, about 32 down, 3.6x. Same quantity, different numbers, and the reason is the window and the threshold, not a disagreement. That post ran July 2025 to now and leaned on five recent, sharp cycles; this one reaches back to January 2024 and folds in slower, mushier early cycles, and it only counts a leg once the price has retraced a full 15 c/L. Widen the window and lower the bar and the average up-leg lengthens and the ratio softens. Both are honest reads of the same market; this is the longer, more conservative one.
Separately, and this is the part that resolves Vinnie, we looked at the moves one station at a time — 31,000 upward moves against 120,000 downward ones — to compare how a single servo behaves versus how the whole market’s median behaves.
Two stretches are held out of the clean summary: the 2024 holiday drift (a slow 39-day creep, not a genuine sharp restore) and the 2026 super-swing that straddles the known March-to-late-April data gap. Both show up as records below, but neither is allowed to skew the median cadence.
What the feed shows
Here’s the resolution to Vinnie’s line. There are two clocks running, and drivers feel both without separating them.
One servo: up in one report. At the level of a single station, the rise really is overnight. The median restoration jump is +45 c/L in a single report, and there are 6,772 jumps of 25 c/L or more on record. When a servo goes back to the top, it doesn’t climb — it snaps. Then it grinds down in many small steps: across every station, the median upward move is +4 c/L and the median downward move is only -2 c/L, but there are 120,000 down-moves against 31,000 up-moves. Few big jumps up; many small steps down. That’s the servo you actually stand at.
The whole market: about 9 days up. The market median looks smoother than any one servo, because servos don’t all restore on the same morning. The North Lakes end can snap on a Monday and the far south not until Thursday. So the cross-station median — the level that describes “the market” — takes a median of 9 days to finish rising. Then it grinds back down over a median of 32 days.
Two clocks, two ways to read the speed, and it’s worth being precise so the arithmetic closes. Take the whole leg and divide by its length: 45.5 c/L up over 9 days is about 5 c/L a day, and 46 c/L back over 32 days is about 1.4 c/L a day. Look instead at the median of each single day’s move inside a leg and you get a bit less — about 4.9 c/L a day up and 1.2 c/L a day down — because the down-leg has a few big drop-days that pull the leg total up while most days move less than that. That’s why 32 days times the typical 1.2-a-day looks short of the 46 c/L the leg actually gives back: the per-day median and the whole-leg total are two different averages, and the leg total is the one that has to square with the up-leg.
The asymmetry. Divide the down-leg by the up-leg and you get 3.6x — down takes about three and a half times as long as up, on 14 clean cycles. That’s the honest market-median number. It’s not the 5-to-10x some drivers quote, because that bigger figure is the per-station effect (one report up versus weeks of decline), not the market’s. Both are real; they’re just measuring different things.
The two legs cover about the same ground: the up-leg restores a median of 45.5 c/L and the down-leg gives back about 46 c/L. Same distance, very different speed — that’s the asymmetry in one line.
The records
A few edges, because they’re on the feed and worth knowing. The fastest market-median rise on record ran +54 c/L in 5 days (5-10 June 2024) — 10.8 c/L a day across the whole market, the closest the median ever gets to a single servo’s snap. The slowest decay crawled at -0.53 c/L a day over 57 days in spring 2024, a fall so gentle you’d never notice it week to week. And the longest single down-leg on record ran 93 days and gave back 108 c/L — but that’s the post-gap super-peak unwinding, held out of the clean summary for exactly that reason.
The verdict
The claim holds, and it’s sharper than the folklore. Up fast, down slow is exactly what Brisbane does: a single servo restores in one report of about +45 c/L, and the whole market’s median level takes about 9 days to finish rising and about 32 days to fall — a 3.6x asymmetry. Vinnie was right on both counts; he was just feeling the servo clock and the market clock at the same time.
So what
The asymmetry is the timing edge. Because the fall is slow and the rise is fast, the risk isn’t symmetric: get caught on the up-leg and you pay near the top for the sake of a day’s wait; sit on the long down-leg and a few days of patience is worth real cents. If your tank can wait, the back half of that 32-day grind is where the cheap days live — and because it’s slow, you don’t have to nail the exact bottom to do well. The up-leg is the opposite: once a snap starts, waiting a day can cost you the whole 45.
You don’t have to model any of this in your head. Watch the metro median and your own regular servos on the map at fill.pinly.com.au and read the number, not the calendar. The Wrap will keep a running tally of where the cycle sits — and if you run a diesel, the asymmetry doesn’t apply to you at all; that’s a separate story, and we’ve measured it too.
The honest limit
Two things to keep this straight. First, the market-median slopes are aggregates — about 4.9 c/L a day up, 1.2 down — and they are far gentler than what you see at the pump. A single servo’s up-move of +45 c/L happens in under a day; the 9-day figure is only how long it takes the whole market to finish restoring, because servos restore on different days. Don’t read the 4.9 as what your corner does. Second, the carry-forward means on a quiet day the median is a stale-adjusted estimate, not a fresh reading, and early-2024 days lean on thinner panels. The 3.6x rests on 14 clean cycles after holding out the two distorted stretches; a longer clean run would firm it up. This is ULP/91, Brisbane metro only — diesel and premium aren’t measured here, and diesel runs a different, far flatter cycle that none of these numbers carry to.