PLAY'S · META · 24 Oct 2026

Week four: Raid and Jokers ended level, on the last day before Raid's card changed

The fourth week of the format ran 440 games, then 84, then 149 — both ends of everything this site has recorded, inside seven days. On the largest day the top four entries sat within 0.8 points of each other. On the last day the top two were exactly level and the third was nine points back. Three columns moved by more than fourteen points across the week, against three by more than ten in week three.

The week, day by day

Leading named deck is the largest single named entry on that day's chart. The survey's own unnamed bucket is excluded from this column. Three names took the lead across the week, and ジョーカーズ took three of the days outright.

The biggest sample gave the flattest chart

This is the clearest the point has ever been in this series, and it is visible in two days at opposite ends of the week.

15 October, 440 games — the largest day recorded here in any week, full or partial. Jokers 13.9, Raid 13.4, Dragerion 13.2, and the unnamed bucket at 13.1. Four entries spanning 0.8 points. Nothing led it in any meaningful sense; the chart is almost flat across the top.

21 October, 149 games. Raid and Jokers both 20.8, then a gap of 9.4 points to Dragerion and Baraghiara, who are themselves tied at 11.4. Two ties on one chart, and a cliff between them.

At 84 games on the 19th, the smallest full read in the series, a single percentage point is under one game. A deck on 2.4 that day was on two games. The spikes are in the small days and the flatness is in the big one, which is the shape you would expect if a good part of what the chart shows is sampling rather than the format moving.

Level at the end, after the widest week yet

Raid moved 15.8 points between its own high and low this week, Jokers 15.7, Cotton Zenith 14.2. In week three the three widest columns moved 11.4, 11.4 and 10.6, and that article called three columns past ten points the thing to take from the week. Week four has three columns past fourteen.

And the two widest finished the week in exactly the same place. Raid went 13.4, 18.5, 24.1, 15.0, 8.3, 18.2, 20.8 — its format high and its format low two days apart, inside one week. Jokers went 13.9, 10.8, 11.4, 26.5, 14.3, 22.7, 20.8. Two entirely different shapes, the same number on the last day.

Cotton Zenith did it again, and further

Week three noted that Cotton Zenith was the only tracked column to put its format high and its format low inside one week, at 15.4 and 4.8. It has done the same thing again in week four, at both new extremes: 16.2 on the 16th and 2.0 on the 21st.

The fall is not one day. It reads 9.5, 16.2, 10.8, 8.8, 8.3, 3.6, 2.0 — down on five of the six days after its peak. That is the one movement in this week that does not look like sampling, and this deck led a day outright as recently as 12 October.

What 2.0 means at 149 games is about three games, so the floor is not a precise figure. The trajectory is the finding, not the last number in it.

Every tracked deck, every day

Bold is that column's high for the week, grey its low. The seven columns are the same ones week three used, so the two weeks can be read against each other. Magic Tools is a family total — the chart splits it across two entries on the 15th and the 20th, and on the 18th it has no named entry at all, which is an em dash here rather than a zero: below the survey's cutoff a deck sits in the unnamed bucket instead of being absent from the format. Baraghiara is not split on any day this week. Other is the unnamed bucket, printed as it appears. 赤単バルガ and ゼニス天門 are on every day of this week too, at 3.6–9.2 and 2.3–8.2, and are left out only to keep the column set identical to week three's.

The sample moved more than the decks did

Full-read days only. The format's opening week, 24–30 September, is left out of this table because its days are partial six-column reads and the leader share is not measured the same way; for what it is worth its games ran 82 to 386, a 4.7× spread. Week four is the widest of the four by sample and the second widest by leader share.

Two of week three's numbers have been passed

Neither was wrong when it was written, and that page has not been rewritten. Both have simply been overtaken by this week:

• It said Raid's 10.2 on 14 October was its lowest figure in all twenty-one days of the format. Raid read 8.3 on 19 October.
• It said Cotton Zenith's 4.8 and 15.4 were its lowest and highest readings of the entire format. Both have gone: 2.0 on 21 October and 16.2 on the 16th.

A superlative in a weekly piece is true of the week it was written in. This site would rather say that plainly in the next week than quietly edit the last one.

What ends with this week

Week four is the last full week played under the old 《“B-零朱”レイド》. The card was changed at the 22 October maintenance, and the deck named after it led two days of this week outright and shared a third.

Nothing here forecasts what that does. The first full day under the new text is 23 October, whose chart publishes later this morning. And week four is itself the argument for waiting: on 18 October Jokers read 26.5, its format high, and a standalone piece on it was held back because the trigger set for it — Jokers holding twenty per cent on the 19th — failed at 14.3. It then read 22.7 and 20.8. Four readings, no two alike, with nothing changed in the game at all.