PLAY’S · META · 29 Oct 2026

One column moved, and the rest of the chart was the steadiest it has been

Week five, 22 to 28 October, is the first full week under the changed 《“B-零朱”レイド》. 赤白轟轟轟(レイド無し) moved 18.4 points between its own high and low. No other column moved more than 9.5. In each of the three weeks before this one the two widest columns finished within 0.8 points of each other, so a gap that size has no precedent in this series. Around it the chart was unusually still: the leading share held its narrowest band of any week, and HADESドラゲリオン moved 1.3 points across the six full days.

The week

The seven columns that appear on every day of the week, plus the survey’s own unnamed bucket, printed as it appears. Bold is that column’s high for the week, grey its low. The red rule marks the card change: play on the 22nd began at 16:00 with the new text, so every row from there down is under it. その他 has no entry on the 22nd because that chart has no bucket. Decks that do not appear on all seven days are left out of the table and some are discussed below.

Every week before this one had two columns moving together. This one has one

The useful measure of a week here is not which deck led it — that changes with the sample — but how far each column travelled between its own high and its own low. Run that over the four full weeks of the format and the shape of week five is immediately different:

Columns present on every day of that week only, so the comparison is like for like. The format’s opening week, 24–30 September, is left out because its days are partial six-column reads.

In the first three weeks the two widest columns are 0.2, 0.8 and 0.1 points apart. Whatever was moving, it was never moving alone. Week five breaks that: one column at 18.4 and then a drop of nearly nine points to the next.

Part of that 18.4 is the part-day. The 4.3 that sets the low is the 22nd, which is 46 games and two of them. Take the 22nd out and measure the six full days on their own and the same column moves 11.1, against バラギアラループ at 7.9 — a gap of 3.2 rather than 8.9. That is a smaller claim and it is the one to carry: the gap is still the largest of any week, and it is not the four-fold thing the seven-day figure suggests.

The deck that did not move

Across the six full days HADESドラゲリオン reads 10.1, 9.6, 9.3, 10.4, 10.0, 9.1. That is 1.3 points, through six days whose samples ran 227, 146, 290, 309, 110 and 132 games, in which the lead changed hands four times between two names and 赤白轟轟轟(レイド無し) moved 11.1 points.

It is the flattest any column has been across a full week in this record, and it is not close: the flattest column of each earlier week moved 1.8, 3.9 and 5.6 points. Reading that as the deck being stable is a guess — a column can sit still because the deck is steady or because two opposite movements cancelled, and a pie chart cannot tell those apart. What can be said is narrower and still worth saying: on a chart where everything else was moving, this one did not.

The leading share was the narrowest of any week, and the sample the widest

Two more numbers point the same way as the table above, and they point in opposite directions from each other.

Leading share is the largest named slice on that day; the survey’s unnamed bucket is excluded from it. Full-read days only.

The widest sample spread of any week and the narrowest leading share. Week five holds both the fifth-largest full read in the series, 309 games on the 26th, and the smallest, 46 on the 22nd — ranks 5 and 28 out of 28. Week four took the sample-spread record five days ago with 5.2×; this week passed it.

The leading share is the one that is genuinely new. In week four the top of the chart ran from 13.9 to 26.5, a band of 12.6 points, and that article made the flatness of the biggest day its spine. Week five’s top runs 16.5 to 22.7 — six points, half of week four’s. Whoever was in front, they were in front by about the same amount every day.

Three names led, and the run that looked like the story ended before the week did

ジョーカーズ led four days outright and shared a fifth. バラギアラループ took the part-day of the 22nd, and 赤白轟轟轟(レイド無し) took the 27th at 22.7, its highest reading anywhere in the record.

Worth recording because it was nearly written as a finding: from the 23rd to the 26th ジョーカーズ led or shared the lead four days running, which is the longest such run in the whole thirty-five-day record. On the 26th, with five of seven days read, that was the obvious spine for this page. The 27th ended it, and the 28th gave the lead back. A week written from five days would have had the wrong story in its headline.

The deck the card belonged to appears once

赤白レイド is a named slice on exactly one day of this week: the 22nd, at 6.5, which is its lowest reading anywhere in the record and is three games. It is absent from all six full days.

That is covered in full on the page written for it, which also sets out what the chart cannot say: an absent deck is below the chart’s smallest named slice, not at zero, and what the two red/white builds hold together cannot be measured from a column that counts an unnamed build as nothing. Nothing in this week’s figures changes either of those. The player side is separate again.

Two of week four’s numbers have been passed

Neither was wrong when it was written and that page has not been rewritten. It now carries a dated note for each.

• “440 games, then 84 — both ends of everything this site has recorded.” The 440 still stands. The 84 does not: the 22nd is 46. That page went up at 01:22 on 24 October and the 22nd’s chart was recorded here at 01:45 the same night, twenty-three minutes later, so the sentence was true when it was published and false within the hour.
• “Week four is the widest of the four by sample.” It was, at 5.2×. Week five is 6.7×.

Week four’s other records survive this week: 440 games on 15 October is still the largest day, ジョーカーズ at 26.5 is still the highest single reading on any full-read day — 青魔導具 read 26.8 on 24 September, but that is one of the seven partial six-column reads at the start of the format and is not comparable — and week four’s three columns past fourteen points are still three more than any other week has managed. Week five has one.

What this page is not

It is not a verdict on the card change. The notice of 20 October states the change as balance adjustment and names no deck, so there is nothing stated to measure a week against. The change also was not the only thing that happened on the 22nd — the maintenance did eighteen things at once, by its own list — and this survey has one column for the outcome and none for the cause.

It is not a measurement of what decks people are building. The survey counts decks met in ranked games. A column rising means the deck was encountered more often in whatever games the surveyor played, which is not the same as more people playing it, and on a 110-game day it is not far from the same as noise.

And one week is one week. This site has written that sentence in every one of these pages and has been made to mean it twice: week four’s seventh day changed two of its findings, and week five’s sixth day ended the run that would have been this page’s headline. Week six is 29 October to 4 November, and its first day is the day the SP Rule Match and a new story chapter both open, which is one more thing the chart will not be able to separate.