0. The 5-second answer
As of midday on August 28, 2026, my starting point is to keep 3 Baal (Genso no Kyomei Baal) and 2 Suzie (Seijitsu naru Noroiya Suzie), and let Raz (Demon Drum Raz) and Rimiru (Bunny na Akuma Rimiru) fight over the flex slots.
If it's strictly "3 Baal or 3 Raz," go with 3 Baal for now. The latest sample lists on both GameWith and Game8 run 3 Baal, while only Game8 runs 3 Raz.
If you only have four slots to split between Raz, Rimiru, and Suzie, the baseline for a general ladder environment is 2 Suzie + 2 Raz. If you're seeing more mirrors and other aggro decks, try 2 Suzie + 2 Rimiru.
One caveat: I haven't run a head-to-head on the same 36 cards and the same random seed for 0 vs. 2 Suzie, or 3 Baal vs. 3 Raz. If I made up a win rate here, this wouldn't be a strategy guide anymore. It would be fortune-telling with card art.
1. What is the 31-card shared core?
The "fixed 31" didn't come down from the mountain on stone tablets. I took the two Aggro Nightmare 40-card lists that GameWith and Game8 updated around 11 a.m. on August 28, 2026, laid them on top of each other, and kept the minimum number of copies that both lists agree on.
| Copies | Card |
|---|---|
| 3 | Miwaku no Succubus Lilim (Succubus Lilim, the alluring one) |
| 3 | Cute na Akuma Lilim (Lilim, the cute demon) |
| 3 | Akuratsu no Lesser Mummy (the vicious Lesser Mummy) |
| 3 | Kyoshu no Tokko Taicho (the assault squad captain) |
| 3 | Zangyaku no Sakuretsu (Savage Burst) |
| 3 | Genso no Kyomei Baal (Baal) |
| 3 | Yoru no Uta no Live (the Night Song live show) |
| 3 | Sukitooru Shinnen Ansage (Ansage, the pure-hearted) |
| 2 | Seijitsu naru Noroiya Suzie (Suzie, the honest curse-seller) |
| 3 | Gekiretsu no Fuku-Socho (the fierce vice-chief) |
| 2 | Garodart VS Zett |
That's 31 cards in total. So it's not "never touch these." It's 31 cards that two separate guide teams each kept independently. If you reconsider Baal, that isn't a contradiction. You're just re-bidding on a reserved seat.
2. The other 9 slots are completely split
GameWith's 9 slots: 3 Yumei no Chui (the netherworld lieutenant), 3 Tessa no Tsunagari (the iron-chain bond), 2 Dark Knight Marsha, and a 3rd Garodart VS Zett.
Game8's 9 slots: 3 Tappitsu no Kaii (the calligrapher's apparition), 3 Katsugo no Akuma (the devout demon), and 3 Demon Drum Raz.
They shake hands on 31 cards, then flip the table over the last 9. That's normal right after a new set drops. If anything, it shows that "the one correct 40" hasn't settled yet.
3. 3 Baal vs. 3 Raz
Baal is a 3-cost 2/2. When played, if you're the aggressor, he gives himself and your other followers +1/+1. If there's an annoying blocker on the other side, you can pick that instead and deal 3 damage to a random enemy. He doesn't wipe the whole board. He only removes whatever stands between you and their face, which is exactly the job aggro wants done.
Raz is a 2-cost 1/2. His Last Words leave a 0-cost 1/1 Skeleton behind, and when he evolves he deals 3 damage to an enemy. His clear strengths are a steady turn 2 and staying power on the board.
Right now GameWith runs 3 Baal and 0 Raz, while Game8 runs 3 Baal and 3 Raz. That doesn't prove a causal win rate, but the latest published lists agree on Baal more than on Raz. If you have to cut one three-of entirely, cut Raz first.
That said, if you keep whiffing on turn 2, the answer changes. Raz isn't getting cut for being weak. The 2-cost slot is just crowded at the moment, and he's getting squeezed out of a seat.
4. Rimiru, Suzie, and the four-slot problem
Rimiru is a 2-cost 1/1. If your health is higher than your opponent's, she summons two Bats, and the same effect triggers when she evolves. Hit the condition and you get three bodies for two mana. If this were a job listing, applicants would be lining up around the block.
The catch is that this deck sometimes wants to hurt itself to make Garodart VS Zett cheaper. The plan of lowering your own health and a Rimiru who gets stronger the more health you have occasionally clash. In mirrors and against other aggro decks she's easier to turn on. In builds that rush self-damage, she's harder to turn on.
Suzie is a 4-cost 3/3. She gives one enemy -0/-3, and when she evolves she gives one follower in your hand +3/+0. That lets her fatten Ansage or VS Zett, so she's not just a board card. She's a relay station for lethal from your hand. Both GameWith and Game8 currently run 2 copies.
If you only have four slots, here's my priority order:
- 2 Suzie + 2 Raz: the baseline for a general environment.
- 2 Suzie + 2 Rimiru: for when there are lots of mirrors and fast matchups.
- 1 Suzie + 2 Raz + 1 Rimiru: flexible, but every role gets thin.
- 2 Raz + 2 Rimiru: light, but you lose Suzie's +3 damage route, so it ranks last until someone tests it.
"Give everyone one copy so nobody's upset" keeps the peace in a meeting. In a 40-card deck, you can't draw it when you need it.
5. How to read the old simulation
In the old all-seven-leaders comparison, the old restricted 40-card list scored 65.5% on an 8-turn judgment. By opponent: Elf 66.7, Royal 75.0, Witch 66.7, Dragon 45.8, Nightmare 54.2, Bishop 79.2, Nemesis 70.8. However, 38.1% of games were still undecided at turn 8. That does not mean you'll win 65 out of 100 ranked games.
On top of that, the old 40 included 3 Suzie and 3 Raz, so it's built differently from today's 31-card core, which centers on 3 Baal and 3 Ansage.
In a separate test aimed only at Pirate Royal, the old baseline 40 scored 91.4%. Swapping a single Marsha for a single Baal made that worse, at 88.0% or 87.2%. All you can take from that is that under those Pirate-only conditions at the time, that one swap was bad. It does not mean Baal is weak overall.
6. Your build goes with your game plan
My current game plan is: borrow the board only early on, don't try to control the board in the midgame, and finish with direct damage and Storm (the keyword that lets a follower hit the enemy leader the turn it comes down).
On turns 1 to 3, chip away first with small bodies. On turns 4 to 6, deal only with the card your opponent's win condition hangs on. Against Witch in particular, on turns 5 and 6 a pointless small follower can become a healing target for Rush (attack enemy followers the turn it comes down) plus drain. So there are moments when you have to stop playing something just because you can.
Baal switches between the bare minimum of removal and pressure. Suzie adds 3 damage to what's in your hand. Raz steadies the early game. Rimiru can blow up your board size when her condition is met. The yardstick isn't how flashy a card looks on its own. It's how many turns sooner you can reach your opponent's face.
7. The answer for now, and the next test
| Situation | Starting point |
|---|---|
| General environment | Keep 3 Baal and 2 Suzie |
| 3 Baal vs. 3 Raz | 3 Baal |
| Four slots among Raz, Rimiru, and Suzie | 2 Suzie + 2 Raz |
| More mirrors and other aggro | Try 2 Suzie + 2 Rimiru |
| Turn-2 stumbles are increasing | Add more Raz |
| You're pushing self-damage hard | Cut Rimiru |
| Lots of Witch | Don't overload on midgame board-only cards |
If you want to settle this with numbers next, lock the 31 cards, then swap only the 9 slots against the same opponents, the same seed, and the same game plan. At minimum, run these against all seven leaders: 2 Suzie + 2 Raz, 2 Suzie + 2 Rimiru, 1 Suzie + 2 Raz + 1 Rimiru, 2 Raz + 2 Rimiru, a build that keeps 3 Baal, and a build with 0 Baal and 3 Raz.
Untested means untested. A blank cell isn't a loss, and it's a hundred times more honest than a made-up decimal point.
My final conclusion: keep 3 Baal. Keep 2 Suzie. Let Raz and Rimiru fight over the remaining seats. The final interviewer in this hiring meeting isn't a tier list. It's the opponent's remaining health.
