Pular para o conteúdo

Do prices dip at weekends? We cannot tell yet: the two October weekends disagree

Este conteúdo não está disponível em sua língua ainda.

A lot of us play more on Saturday and Sunday, and a common belief is that the auction house gets cheaper then, because more sellers show up. If that were true, the weekend would be the time to shop and a weekday the time to sell. We checked our own scans to see whether the beta shows it, and what it would take to know for sure.

Do prices fall on Saturday and Sunday, compared with the weekdays around them? The catch is that prices also drift for other reasons, a little each day and sometimes a lot each week. So a weekend dip only counts if prices fall at the weekend, come back up on the weekdays after, and do it weekend after weekend. One weekend that happens to fall is not enough, because it might just be part of a slide that started earlier and kept going.

A useful answer needs at least three weekends of full scans in a market that is not already rising or falling fast, with the days on either side of each weekend scanned too. It should say whether the weekend stands out from ordinary pairs of days, and whether it does so every time.

  • We cannot tell yet. Our own full scans cover 24 days on Horde and 26 on Alliance, but there were no Horde scans from 22 October to 4 November, so only two weekends fall before the gap, and the one after it came while the auction house was refilling.
  • The two October weekends disagree. From Friday 9 October to Sunday 11 October, 1,316 Horde items got cheaper and 912 got dearer, tilted toward cheaper more than any other two days of the month. From Friday 16 October to Sunday 18 October it was 1,089 against 936, an ordinary pair of days.
  • The big drop on the weekend of 7 November was not a weekend dip. The typical Horde price fell 25% from Thursday 5 November to Friday 6 November, before any weekend, then 16.7% into Saturday and 11.5% into Sunday: a slide that was already slowing down.

We used only our own full scans: 8 to 21 October and 5 to 14 November on Horde, and 8 to 23 October and 5 to 14 November on Alliance. The daily history from before 8 October covers only about 320 items a market and is measured differently, so we left it out. Each day is stood for by its last scan of the day. We do not measure how many people play. The closest thing we have is how many units are listed.

We looked at the question three ways.

Two-day stretches in October. For every pair of days two apart, Friday to Sunday, Saturday to Monday and so on, we counted the items priced on both days that got at least 1% cheaper and those that got at least 1% dearer. We chose 1% so that a copper or two of rounding does not count as a move. A stretch is lopsided when far more items got cheaper than dearer, or the other way round. That gave 12 stretches on Horde and 14 on Alliance. If weekends mean cheaper prices, the Friday to Sunday stretches should be the most lopsided toward cheaper, every time. The first one was, on both factions. The second was not: on Horde it came sixth of 12, and on Alliance seventh of 14 (1,066 cheaper against 970 dearer, after 1,280 against 930 the weekend before). Worse for the weekend idea, the two days into Saturday 17 October leaned hard the other way on Horde: 871 items got cheaper and 1,227 got dearer. And after the first weekend, prices did not bounce back: from Sunday 11 to Tuesday 13 October, 1,201 Horde items still got cheaper against 1,019 dearer. A dip comes back up. This looked more like a week that drifted down.

The typical item, meaning the middle item when all of them are lined up by how much they moved, did not move at all over any two days in October: a change of 0% on every stretch, on both factions. Most items sat still or moved a little either way.

The weekend against the days around it. For every item priced on all six days, we compared its Saturday and Sunday price with its price on the Thursday, Friday, Monday and Tuesday around them. Because there are days on both sides, a steady rise or fall cancels out, and what is left is how the weekend differs from its own surroundings. We used an average that one odd price cannot drag far, then took the middle item. On both October weekends, on both factions, the weekend came out at 0%: no different from the days around it. We did the same for units listed, which is where the weekend did stand out: more on that under the charts.

November, at the same hour each day. After the scan gap the auction house started almost empty and refilled day by day, from about 1,700 priced Horde items on 5 November to 6,031 in the newest scan. Prices fell fast while it filled. Here we have many scans a day, so we took the scan nearest 15:00 UTC (10 in the morning, US Eastern) on each day and compared each item with the same hour the day before. The short answer uses each day’s last scan instead. Both tell the same story: on Horde the typical price fell 33.3% into Friday 6 November, 20% into Saturday, 11.8% into Sunday and 6.1% into Monday, then between 1.8% and 3.3% a day since. Alliance is close behind. The weekend sits on a smooth curve with no kink in it. This morning, Saturday 14 November, the typical item was 2.5% cheaper on Horde than at the same hour on Friday, and 2.8% on Alliance. The day before, Friday against Thursday, was 3.3% and 4.8%. So far, no extra Saturday drop.

Bar chart of every two-day stretch in October on Horde, showing the first Friday to Sunday stretch as the most tilted toward cheaper and the second as middling

Each bar is one two-day stretch. A bar to the right means more items got cheaper than dearer, counted per 100 items priced on both days. The orange bars run from Friday to Sunday. The first weekend tops the chart, but the second sits in the middle of the pack, and the stretch just before it, Thursday 15 to Saturday 17 October, leans furthest the other way. One weekend for, one weekend against.

Line chart of the daily change in the typical price after the November restart on both factions, falling steeply at first and flattening near minus two to three percent a day

Each point is the change in the typical price from the same hour the day before. Both factions fell hardest on the first day and eased every day after, through the weekend of 7 and 8 November and out the other side. If weekends were cheaper, the weekend points would sit below the curve. They sit on it.

Bar chart showing units listed on Horde were higher on both October weekends, against the days around them, than on any other pair of days

This is the one place the weekend shows up. Each bar compares the units listed on a pair of days with the two days either side. Most pairs come out above zero on this measure, so read the bars against each other, not against zero. On Horde the two weekends were up 25.7% and 21.5%, and every other pair of days was between a 1.5% fall and a 14.6% rise. Alliance looks the same: 27.8% and 18.9% at the weekends, between a 6.2% fall and a 15.8% rise otherwise. That fits more people playing and listing at the weekend. But more listings did not reliably mean lower prices. For live listings and prices, see The Terminal and Charts.

Three or more weekends in a row, scanned in full, in a market that has stopped refilling. The November slide has slowed to a few percent a day, so the weekends of 14, 21 and 28 November, each with the Monday and Tuesday after it, are the first real chance.

We would say yes, prices dip at weekends, if on at least three of those weekends, on both factions, the Friday to Sunday stretch leans clearly toward cheaper, the Sunday to Tuesday stretch leans back toward dearer, and the weekend comes out below the days around it rather than at 0%. We would say no if the weekend stretches keep landing in the middle of the pack, as 16 to 18 October did. If units listed keep jumping at weekends while prices stay put, the answer becomes: more goes up for sale at the weekend, and buyers take it at the usual price.

Two things would make the test sharper. Scans at the same hours every day, so that time of day cannot pass for day of the week. And a sold-a-day estimate kept for every day, not just the newest scan, so we could see whether more actually sells at the weekend.

This is beta data under a level 30 cap, with no raids and a small population, and the beta is wiped before launch. When people play in a beta is not when they will play at launch, so even a clear beta weekend pattern might not carry over. Only the order of things carries over, never the numbers.

In October we have one scan a day, taken at different hours, mostly between 20:00 and midnight UTC but as early as 16:00 on some days, such as Sunday 18 October on Horde. A single scan is a snapshot, and prices and listings move within a day, so part of what looks like a day-of-week difference could be a time-of-day difference. There were no Horde scans from 22 October to 4 November, which breaks the run of weeks in two. The November days describe an auction house that was refilling, where everything was falling, and that trend swamps any weekend effect. Sold a day is an estimate and exists only in the newest scan, so this report cannot say whether more sells at the weekend, only whether more is listed. How we scan and what the fields mean is on the methodology page.