1. The five-second answer
Set 9, Revenants of Azvaldt, launched on August 27, 2026. In the supported day-one simulation pool, Aggro Nightmare came out as the strongest practical starting point.
But there are three different numbers: 81.0%, 65.5%, and 91.2%. They are not contradictions. They come from different experiments.
The deck recommended for everyday play is the constrained 40-card list with 0 傍死のアナテマ・徒姫, 1 ガロダートVSゼット, and no newly added Legendary cards. Its eight-turn score was 65.5%, with a small first/second-player gap.
Its biggest simulated weakness was Dragon. In the play session that triggered this article, Dragon was barely showing up.
So the joke version is simple:
Theory: “Dragon is your bad matchup.”
Reality: “Where is Dragon?”
2. The 40-card deck
Card names are kept in their official Japanese form to avoid inventing unofficial translations.
| Copies | Card |
|---|---|
| 3 | 魅惑のサキュバス・リリム |
| 3 | 残虐の炸裂 |
| 3 | 渇仰の悪魔 |
| 3 | ゴーストエスケーパー |
| 3 | デーモンドラム・ラズ |
| 3 | タイトロープキャット |
| 3 | 夜の唱のライブ |
| 3 | キュートな悪魔・リリム |
| 1 | ダークナイト・マーシャ |
| 3 | 誠実なる呪い屋・スージー |
| 3 | 強襲の特攻隊長 |
| 2 | 幽冥の中尉 |
| 3 | 鉄鎖の繋がり |
| 3 | 激烈の副総長 |
| 1 | ガロダートVSゼット |
Deck hash: 532b50fad4ebb60bac5132cda9c737f2670529cdf45f3f646a2ced7c7de3b709
This is not a truly cheap deck. 誠実なる呪い屋・スージー is a Gold card and the optimized list wants three copies. We saved resources by cutting Anathema to zero and Galodart to one, only for Suzie to appear at the checkout counter.
3. What the deck actually does
This is an aggressive deck, not a board-cleaning simulator.
- Turns 1–3: develop cheap cards and damage the opposing leader.
- Turns 4–6: remove only the cards that start the opponent's win condition.
- Finish with Storm/direct damage, 夜の唱のライブ, and ガロダートVSゼット.
The key lesson is that you do not need to kill every opposing follower. If you spend every turn cleaning the board, the opponent eventually gets to play the expensive card you were trying to avoid.
You are here to attack, not to apply for a janitorial position.
4. Why 81.0%, 65.5%, and 91.2% are all correct
81.0%: the first broad Set 9 comparison
The first report tested 32 fully supported candidate decks across 2,324 engine games. Aggro Nightmare ranked first at an 81.0% average, with a 63.9% worst leader matchup and a 4.6-point play-order gap.
That original deck, however, used three copies of 傍死のアナテマ・徒姫 and three copies of ガロダートVSゼット.
This was closer to a “use the strong cards if you own them” baseline.
65.5%: the constrained everyday list
The final everyday list uses zero Anathema, one Galodart, and no extra Legendary additions.
Against 14 opponent decks—two per leader—the eight-turn score was 65.5%. Going first scored 63.1%; going second scored 67.9%.
A different play policy reached 67.9%, but its split became 76.2% going first versus 59.5% going second: a 16.7-point gap. The more stable 65.5% setup was therefore kept.
91.2%: Pirate Royal specialist mode
This is a different job.
Against 18 Pirate Royal lists only, a dedicated constrained build reached 91.2%, with an 80.6% worst opposing list.
Do not compare a pirate exterminator directly with a general-purpose ladder deck.
5. Matchup guide
| Opponent | 8-turn score | Simple rule |
|---|---|---|
| Forest | 66.7% | Remove combo starters, not everything |
| Sword/Royal | 75.0% | Kill the flag factory, not the flag |
| Rune/Witch | 66.7% | Race before the hand is assembled |
| Dragon | 45.8% | Try to finish before 7 PP |
| Nightmare | 54.2% | Count the opponent's self-damage |
| Haven | 79.2% | Bundle damage before healing |
| Portal | 70.8% | Remove token/copy engines |
Dragon is the obvious weak point. Haven, Royal, and Portal were much more comfortable in this model.
If Dragon is excluded, the simple unweighted average of the other six leader scores is about 68.8%. That is not a real ladder win rate; it is only a way to show how much Dragon drags down the all-leader score.
6. Pirate Royal: do not punch the flag
Against Pirate Royal, the priority targets are:
- 外洋の斥候
- 渦潮の砲手
- 荒波の副船長
- Strong copy/reuse targets
- Everything else
The flag itself cannot simply be attacked away. Stop the followers that produce more flags.
Also respect L'Age d'Or at 4 and 6 PP and Barbaros around 7 PP. Do not donate three or more followers to a board clear just because “wide board looks cool.”
The flag is evidence. Find the factory.
7. What if Dragon is barely present?
Then the practical value of the deck improves.
The simulator says Dragon is the worst leader matchup at 45.8%. If the actual opponents you meet rarely include Dragon, that weakness matters less for your real deck choice.
This does not prove Dragon is globally absent from the metagame. It only means local encounter frequency should affect what you queue.
Metagames move. If Dragon suddenly becomes popular tomorrow, rerun the numbers.
8. Mulligan
Usually keep:
- 魅惑のサキュバス・リリム
- 残虐の炸裂
- 渇仰の悪魔
- ゴーストエスケーパー
- デーモンドラム・ラズ
- キュートな悪魔・リリム
- 強襲の特攻隊長
- 幽冥の中尉
Conditional keep:
- 鉄鎖の繋がり when it can answer an important early engine
Usually send back:
- 激烈の副総長
- ガロダートVSゼット
- duplicated heavy cards that do nothing early
Do not keep a card just because it is powerful. Keep a hand that can actually play the game on turns 1–3.
9. Holding cards and finishing
夜の唱のライブ and ガロダートVSゼット are major finishers.
Do not hold your entire hand waiting for a cinematic ending. Usually keeping one finishing resource is enough.
With only one Galodart, use it mainly when:
- it creates guaranteed lethal, or
- failing to use it means you lose next turn.
Before trading, count current-turn and next-turn damage. In this deck, “Can I kill them?” often matters more than “Can I make the board look clean?”
10. The three-Suzie budget problem
The final 40-card list kept three copies of 誠実なる呪い屋・スージー. The Pirate-specialized constrained list also used three.
So the simulations did not support removing a copy merely because the card is expensive to craft.
However, this project did not yet run a dedicated “Suzie capped at one copy” or “Suzie capped at two copies” budget optimization.
The honest conclusion is:
Three copies are the current performance baseline.
Whether the third copy is worth the resource cost has not yet been tested.
The 41st card in a 40-card deck is apparently your wallet.
11. Do not read 65.5% as a real win rate
The all-leader simulation stops at turn 8. Unfinished games are scored as 0.5.
In the final sample, 32 of 84 games—38.1%—were unresolved at that limit.
Therefore, 65.5% does not mean “expect 65 wins per 100 ladder games.” It is a short-horizon comparison score designed to compare candidate decks under the same engine conditions.
It is also day-one data. Day-one metagames age like milk left in the sun.
12. Final answer
For the first day of Set 9, the constrained Aggro Nightmare list is a reasonable everyday choice.
It survived the resource restrictions, has a clear Pirate Royal plan, and its largest simulated weakness is currently not appearing often in the observed matches.
The remaining pain point is simple:
Before Aggro Nightmare hits the opponent, Red Ether hits you.
If you can afford the list, queue it.
