← Duel Masters English

PLAY'S · META · 5 OCT 2026

The tally and the ranking are not counting the same decks

This site reads two Japanese sources about the same format: a daily survey that counts what people queued into, and a published ranking that says which decks are strongest. The numbers page put them side by side. Eleven days in, that is harder than it looks — they do not use the same names, and they do not cut the same decks in the same places. Four worked examples.

What this is from Daily encounter counts for 24 September to 4 October. All eleven charts were read directly for this page, and every figure in this site's published eleven-day table was checked against them. The ranking as it stood when this page was written: updated 5 October at 18:18 JST, read about ninety minutes later. Player usage is from public posts on 2 and 3 October.

Neither source's table is reproduced here. The names below are quoted because the names are the subject.

1. Jokers has four names and one deck

The survey calls it . The ranking calls it . Players posting about beating it call it . The survey also changed its own mind partway through. For the first three days its bucket was — 4.9%, 4.2% and 5.4% on 24 to 26 September. From 27 September it has been . The eleven-day series this site plots as one line is, on the source's own labelling, two. This one is almost certainly the same deck — the two qualifiers both mean Nature — and this site has treated it as one. That is a judgement, not something either source states.

2. Magic Tools is one entry or two, depending on the day

On 26 September, 28 September, 30 September and 2 October the survey had two Magic Tools entries. On 24, 25, 27 and 29 September, and again on 1, 3 and 4 October, it had one. Four splits, on alternate days, with single days in between. Nothing was announced; the chart simply had a different number of slices.

Added together across 29 September to 2 October, the four days read most closely, the family read 9.2, 10.6, 9.1 and 12.1. Read off the single line instead, the same days say 9.2, 7.6, 9.1 and 8.9. One of those series falls and the other does not.

The split is not new, either. On 28 September the two entries read 13.4% and 4.6% — a family of 18.0% behind a headline of 13.4%.

3. Baraghiara was split twice, and not along the same line

On 4 October the survey added a second Baraghiara entry, so the family that read 17.5% the day before now reads 11.2% plus 4.2%. Taken together that is 15.4% — down 2.1 on the day, where the headline line on its own says down 6.3.

The ranking also splits Baraghiara in two, but not into the same two. It carries two separate Baraghiara entries in two different tiers, divided by colour and by what the list leaves out. Neither split maps onto the other, and neither source says how it decided.

Two sources, two different ways of cutting the same archetype, on the same day. Any line drawn between them is drawn by the reader.

4. Cotton Zenith, which this site has reported as ranked nowhere

The survey has carried every day of the format. It was the most-encountered named deck on 28 September at 13.9%, fell to 4.9% by 1 October — about a third of its peak — and has risen on each of the three days since: 5.6, 9.5, 11.2% on 4 October, which puts it joint third among named decks.

This site has said more than once that the ranking does not list it. That is still literally true: as of the 5 October update there is no entry called or anywhere on the page.

But there is a Lionel entry now, and it may be the same deck The second tier now contains . It was not there on 3 October, when a different Lionel deck, , sat in the tier below instead; that one is no longer listed at all. The ranking's own update history records neither change — it credits the 5 October update to . Players posting ladder finishes on 2 and 3 October called their Cotton decks and , tagged one , and thanked by name. Light-and-dark and white-black describe the same pair of civilizations.

This site cannot confirm they are the same deck. Settling it would mean comparing the two sources' deck lists, and this site does not reproduce either source's lists. What can be said is that the sentence "the ranking does not list it" rests on a name, and names are what this page is about.

What this does to the finding on the numbers page

That page reports that on several days the two most-encountered named decks were the two in the ranking's top tier. That finding survives this one, but for a reason worth saying out loud: Raid and Jokers are the two decks whose identity is least ambiguous across the sources. The agreement was easiest to check exactly where the names happened to line up.

Everywhere else — Magic Tools, Baraghiara, Cotton — a reader comparing the two sources is doing work neither source did for them.

What would settle it, and why this site will not

Deck lists would. Both sources have them; neither publishes them in a form that lines up with the other, and reproducing another site's compiled lists is something this site does not do, which is the same rule that keeps tier tables and poll aggregates off these pages.

So the honest position is the one stated at the top: these are two sources, read separately, and the joins between them are this site's own. Where a join is uncertain, it will say so.

Corrected 5 October, later the same evening This page first said the 5 October update was what added to the second tier. The ranking's update history does not say that. It lists the 5 October change as an update to , and does not mention the Lionel entry at any date. All that can be checked is that the entry is there now and was not there on 3 October.

Which is, unhelpfully, another instance of the thing this page is about.

Sources

Daily encounter counts, 24 September to 4 October 2026: the New Division survey reports published at BEANS, every day's chart image read for this page.

Deck ranking: the New Division ranking at Kamigame, as updated 5 October 2026 at 18:18 JST. Only the names needed for the argument are quoted; the ranking itself is not reproduced.

Player usage: public posts on X from 2 and 3 October 2026, described rather than quoted at length, as on the community page.

Deck names are given in Japanese wherever the point being made is about the name itself.

Corrections and questions

If something here is wrong, please say so — that is the most useful thing you can do with this page. Comments use GitHub, so you will need an account, and everything posted lives in this site's own repository rather than on a third-party service. No tracking, no ads.