PLAY'S · META · 4 OCT 2026

Day nine: the tally and the tier list name the same two decks

For six days running, the most-encountered named deck in the New Division format was less common than the most-encountered deck the day before — 26.8% on 24 September down to 13.3% on the 29th, just under half. That stopped. On 1 October the figure was 22.0%, the second-highest reading of the nine days. And on 1 and 2 October the two decks people ran into most were the two decks the published ranking puts in its top tier, which before this week had happened on one day only, 29 September.

What this page covers, and when Nine days: 24 September to 2 October 2026, the first nine days of the set 37 New Division format, 2,039 games in all. The daily counts are one person's survey, published the following morning. The ranking is a different site's and was last updated on 3 October at 12:30 JST; the figures here were read on 3 and 4 October.

The six-day page covering 24–29 September is left as it was. This is a separate page, not a rewrite of it.

Corrected 4 October Published earlier the same day with three statements the table on this page contradicts. It said Raid holding the lead on 30 September and 1 October was the first time any deck had led two days running — Magic Tools led the first three days. It said Jokers had fallen once on its way up — it fell on 25 September as well as 30 September. And it said the top two decks had never been Jokers and Raid during the first week — they were, on 29 September.

The leader column now says named deck. The survey's own unnamed bucket was larger than the leading deck on 27 and 28 September, so that column is the leading deck, not the largest slice. All four were caught by recomputing the claims from the table rather than re-reading the sentences.

Day ten, added 4 October The 3 October reading went up on the morning of the 4th, after this page was published, and it breaks the finding below. The two most-encountered named decks on 3 October were Raid at 21.5% and the Baraghiara loop at 17.5%. Jokers was third, at 15.4%. The pairing this page describes has now held on 29 September, 1 October and 2 October, and not on 30 September or 3 October — three days out of the last five.

Three other things in that reading. The survey was back up to 325 games from 124, so it is the firmest single day since 27 September. The leader's share was 21.5%: every reading since the 13.3% low on 29 September has been above it, though not rising every day — it fell on 2 October. And the Magic Tools entry read 4.6%, the lowest single Magic Tools reading of the format, with no second Magic Tools entry beside it this time. That is not a clean comparison with the day before, when there were two entries adding to 12.1; part of the missing 3.2 may now be sitting in the unnamed bucket rather than having gone anywhere. Cotton Zenith was 9.5%, its highest since 29 September.

The bucket count moved the other way this time. The survey itemised 10 named decks on 3 October against 13 on 2 October, and the three that went — one of them the second Magic Tools entry — were worth 9.6 points between them. "Other" rose by 11.5, from 4.3% to 15.8%. So 9.6 of that 11.5 is the bucket count, and 1.9 is not. The same caution applies as on 2 October, in the opposite direction.

The table on this page still covers nine days and has not been changed.

Day eleven, added 5 October The pairing was back on 4 October. The two most-encountered named decks were Raid at 17.3% and Jokers at 13.5%, which is the pair in the ranking's top tier. Over the six days since 29 September it has held on four of them and not on two.

Read the Baraghiara line carefully. The survey split it: a new entry, Cabbage Baraghiara, appears at 4.2% beside the loop's 11.2%. Taken as one family that is 15.4% against 17.5% the day before — down 2.1, not down 6.3, which is what the loop line on its own says. One bucket was added and none was removed, 10 to 11.

Otherwise, from 312 games: Cotton Zenith reached 11.2%, its highest since 28 September, on a third straight rise — and it is still in no tier at all. Jokers fell for a second day, 17.7 to 15.4 to 13.5 — its longest run of falls since the format opened, and still only two. The leader's share was 17.3%.

Nine days, one row each

Date	Games	Most-encountered deck	Its share
24 Sep	82	Magic Tools	26.8%
25 Sep	308	Magic Tools	20.5%
26 Sep	333	Magic Tools	19.5%
27 Sep	386	Raid	15.8%
28 Sep	216	Cotton Zenith	13.9%
29 Sep	294	Jokers	13.3%
30 Sep	132	Raid	15.2%
1 Oct	164	Raid	22.0%
2 Oct	124	Jokers	17.7%

Share of games the single most-encountered named deck accounted for, by day. The survey's unnamed "Other" bucket is excluded from this column because it is not a deck; on 27 and 28 September it was larger than the leading deck.

The slide stopped on day seven

The first week's story was that nothing held. Every day the leader was a little less dominant than the leader before: 26.8, 20.5, 19.5, 15.8, 13.9, 13.3. Six falls in a row, and by the sixth the top deck was barely more than a tenth of the field.

Then 15.2, 22.0, 17.7. The thing that had fallen every day for six days did not keep falling. The 1 October reading is higher than every day of the format except its first.

A format that flattens for six days and then concentrates again on the seventh has not settled. It has changed direction once.

Raid held the lead two days, and grew while holding it

Between 27 and 29 September the lead changed hands every single day — Raid, then Cotton Zenith, then Jokers. Raid took it back on 30 September and kept it on 1 October, and it did not just hold: it went from 15.2% to 22.0%, about half again as common. That is the first repeat since the opening stretch, when Magic Tools led for three days running — and the first time any deck has held the lead on consecutive days since 26 September.

On 2 October it lost the lead again, to Jokers at 17.7% against its own 16.9%. Jokers has now gone from 4.9% on day one to 17.7% on day nine, more than three times its opening figure. It fell twice on the way, on 25 September and again on 30 September.

Every deck this site has been following, every day

Date	Jokers	Cotton Zenith	Raid	Magic Tools	Loop	Other
24 Sep	4.9	4.9	24.4	26.8	3.7	8.5
25 Sep	4.2	5.5	16.6	20.5	9.7	12.7
26 Sep	5.4	7.2	16.5	19.5	15.0	15.1
27 Sep	7.5	9.3	15.8	13.7	10.4	16.4
28 Sep	11.1	13.9	11.6	13.4	8.3	15.8
29 Sep	13.3	9.9	12.6	9.2	9.5	13.2
30 Sep	9.8	5.3	15.2	7.6	13.6	12.3
1 Oct	14.6	4.9	22.0	9.1	9.1	17.7
2 Oct	17.7	5.6	16.9	8.9	15.3	4.3

Percentages of games surveyed. "Loop" is the Baraghiara loop deck; "Other" is the survey's own unnamed bucket. Read the Magic Tools column with the caveat below.

The two lists now point the same way

On 2 October the two most-encountered named decks were Jokers and Raid. The ranking, as updated on 3 October, puts exactly two decks in its top tier: 自然ジョーカーズ and 火光レイド. Same two decks.

That pairing has now held on three of the last four days — 29 September, 1 October and 2 October. The exception is 30 September, when the deck second behind Raid was the Baraghiara loop.

Agreement is not confirmation These remain two different kinds of claim. A tier list is an opinion about how strong a deck is; an encounter tally is a count of what people queued into. One can be right while the other is wrong. All that changed is that for the first time in nine days they are not pointing at different decks.

The two sources also use different names for the same decks — 火光レイド against 赤白レイド, 水魔導具 against 青魔導具. That is a reason to read them as two sources, not one dataset.

Cotton Zenith is still in no tier at all

Cotton Zenith was the most-encountered named deck in the format on 28 September, at 13.9%. It is 5.6% on 2 October, still ahead of most of the field. The ranking does not contain it — not in the top tier, not in the one below, not anywhere on the page. That was true when the ranking first covered this format and it is still true after the 3 October update.

This page is not in a position to say which reading is right. It can say that one source counted it as the single most-played deck in the format and the other has never listed it.

Three things in this data changed shape, not size

1. The survey got much smaller 386 games on 27 September; 124 on 2 October. At 124 games one game is worth 0.8 of a percentage point, so an entry reading 3.2% is four games. The small figures at the bottom of the chart should not be read as precise.

2. The unnamed bucket fell 13.4 points in a day, and that is mostly a counting change "Other" reads 17.7% on 1 October and 4.3% on 2 October. The survey itemised 13 named decks on 2 October against 11 the day before. The two decks that appeared account for 7.2 points of the 13.4-point fall. The remaining 6.2 points are not explained by the bucket count, and this data cannot say what they are. Do not read the fall as the unnamed part of the format disappearing.

3. Magic Tools is sometimes one entry and sometimes two On 29 September and 1 October the survey had a single Magic Tools entry. On 30 September and 2 October it had a second one beside it. Added together, the family reads 9.2, 10.6, 9.1 and 12.1 across those four days — roughly flat, not falling. The Magic Tools column in the table above is the single entry only, so on two of those days it is lower than the family actually was.

What else was running at the same time

The Legend Battle opened at 17:00 on 1 October and runs to 21 October; October's sanctioned tournaments run all month. The 1 October row therefore covers the same day a large solo event opened. Whether that moved players out of ranked queues, and which decks they were playing, is not something an encounter tally can answer.

Two competitive events were announced on 1 October — 「ミノガミ杯」 from 8 October and BATTLE ARENA 2026 AUTUMN from 26 October. Both are All Division. Everything on this page is New Division, so none of it describes the field those will be played in. The dates are on the calendar.

This is a count, not a verdict

Encounter rate is not win rate. A deck can be everywhere and lose; a deck can be rare and win. The survey is compiled by one person, from their own games, with every rank band mixed together. It is the best daily record of this format in public, and it is still one person's sample.

Sources

Daily encounter counts, 24 September to 2 October: the New Division survey reports published at BEANS, read on 3 and 4 October 2026. The figures are read off the chart image in each day's post; the posts carry no table. Sample sizes and the compiler's handle are printed on each chart.

Top tier, and the absence of Cotton Zenith: the New Division deck ranking at Kamigame, as updated 3 October 2026, 12:30 JST. Only the contents of its top tier and the fact of an absence are reported here; the ranking itself is not reproduced.

Event dates: the publisher's own announcements of 28 September and 1 October 2026. See the calendar.

Deck names are given in English where this site has used an English name before, and in Japanese where it has not. Nothing on this page reproduces another site's table, ranking or deck list.