When should you buy concert tickets? What our time-series actually supports

"Buy now — prices only go up" is the most effective sentence in ticket marketing, and it is testable. OnStage has tracked the entry price (the cheapest listed ticket) of 9,829 events across 11 countries since August 9, taking a snapshot on every price change and every six hours: 98,432 snapshots in all. For movement claims we use a clean cohort: 7,684 events with at least three price snapshots and an entry price above $10, each compared only against itself, first observation to last. This article says exactly what that data supports about timing a purchase, and — just as carefully — what it does not.

What the data supports

1. "Prices always rise" is false at the entry price — in fact, falls beat rises. In that 7,684-event cohort, 6,762 floors — 88.0% — ended the window exactly where they started. Among the movers, 528 events (6.9%) got cheaper while only 394 (5.1%) got more expensive. For the typical show in a typical week, waiting cost nothing; the second most likely outcome of waiting was paying less.

2. When the floor does rise, waiting is expensive. The rises at the top of our published table were not gentle: Tritonal (US dance) went $24.95 → $100.70 (+303.6%); a four-band alternative bill went $17.00 → $40.46 (+138.0%); Justin Silva went $39.83 → $63.82 (+60.2%). The distribution of outcomes is: usually nothing, sometimes a discount, occasionally a step change you cannot undo — none of our top risers had returned to its original floor by our last snapshot.

3. The extreme risk is concentrated, and so far North American. All 25 entries in our top-risers table are US or Canadian events. If you are buying for a show in the UK or EU, our window contains zero named examples of the buy-instantly penalty at the entry price.

The strategy this supports is boring and useful: default to patience; make exceptions deliberately. Buy early when the specific show has scarcity written on it — small room, rising act, final date, headliner that just outgrew the venue (the Tritonal profile). For everything else, the floor will very likely be there next week.

Transparency section: what our data cannot say

This is the part most coverage of ticket prices skips, so we are making it a standing feature.

We cannot tell you the best week to buy. Our raw data contains a seductive curve: average entry price bucketed by weeks-until-event falls from $27.79 at eight weeks out to $23.55 in event week. Printed naively, that "shows" prices drop 15% as the show approaches — wait and save! We refuse that reading, because the curve is composition-biased. Each bucket holds a different mix of events: the near-term bucket is dominated by small, cheap, short-horizon club shows that go on sale days before the date; the far bucket skews toward bigger, pricier productions announced months out. The curve measures what kind of events exist at each distance from their date — not what happens to the price of any ticket you are watching. A per-event, cohort-correct "weeks-out" curve needs each event observed across its whole sales window; at two months of tracking, our series are not long enough. When they are, we will publish that curve, whichever way it points.

Other limits: our "first price" is the first we observed, not necessarily the on-sale price, so pre-tracking moves are invisible to us; we see the primary market only (resale timing is a different game with different math); and two months cannot see seasonality — a window containing an on-sale-heavy autumn may not generalize to spring.

The honest answer, in three sentences

For 88% of events in our cohort the entry price did not move in two months, and a fall (6.9%) was more likely than a rise (5.1%) — so routine urgency is unjustified. For the rare show with real scarcity, the penalty for waiting ran as high as +303.6%, and it arrived as a sudden step, not a drift you could react to. Know which kind of show you are buying for — and distrust any advice, including "wait for a drop," that claims the timing question has one answer for all events, because our data says it does not.