0. Five-second conclusion
Aggro Nightmare felt excellent on Set 9 launch day. By the next day, opponents appeared more prepared with removal, healing, and Ward, and games that used to end on turns 5–6 were more likely to reach turn 7.
The first reaction was to add more one-drops.
Then reality produced the perfect comedy sketch:
Turn 7, opponent at 3. Topdeck 達筆の怪異: effectively a 1/1.
激烈の副総長 draws two: 怪異, 怪異.
Human Resources delivered two interns after closing time.
After another adjustment, the deck returned to ending games at roughly turn 6. Better still, 副総長 itself won a game on turn 6 with three direct leader damage.
The current practical verdict is:
One 達筆の怪異 — the “Goblin slot” — is enough.
The zero-copy 副総長 plan is withdrawn. Three face damage on turn 6 is a real closing job.
The real optimization target is balancing turns 1–3 consistency with topdeck quality through turn 6.
1. Why aggro felt worse on day two
In short ladder observations, especially against Bishop/Haven-style decks, more removal, healing, and Ward made pure board pressure less reliable. This is not a global server statistic; it is local, anonymized play evidence.
Public rankings also show that the environment is still moving. On August 28, GameWith placed AF Portal, Combo Forest, Evolution Haven, and Experiment Rune in the upper band, while Game8 still rated Aggro Nightmare highly.
The safest interpretation is not “aggro died.” It is:
Opponents started reading the manual.
Day one is a product demo. Day two is the countermeasure meeting.
2. The starting 40 and the strategic idea
The tuning session began from 怪異1, both Lilim at three copies, Mummy3, 渇仰3, Captain3, Lieutenant3, 残虐3, 鉄鎖3, Baal3, Live3, Ansurge3, Suzie2, 副総長3, and VS1.
The strategic idea had already moved away from permanent board control:
Turns 1–3: rent the board and damage the opposing leader.
Turns 4–6: remove only what matters, then switch to burn and Storm.
Around turn 6: finish from hand.
The board is not real estate. It is short-term rental property.
3. If Dragon is at 13 after turn 3, the speed ceiling exists
A clean Dragoncraft game left the opponent at 13 after turn 3: seven damage in three turns.
That does not prove a seven-damage average. It proves the deck has enough maximum speed.
As a practical logging KPI, opponent HP of 12–13 after turn 3 is excellent, 14–15 is healthy pressure, 16–17 is slow, and 18–20 is close to an opening failure. This is not an official threshold or a win-rate formula; it is a thermometer for comparing tuning changes.
Recent losses showed two distinct failure modes:
A. Opponent still at 12–15 around turn 5: early consistency failed.
B. Opponent at 3 around turn 7, but the topdeck is not immediate damage: finishing density failed.
Fixing only A can make B worse.
4. Three 怪異: better mornings, terrible overtime
達筆の怪異 is a one-cost 1/1 that becomes larger when another original-cost-one card is on the field. Played alone on turn 1, it is still a 1/1.
The starting list had seven one-drops: 怪異1 + 魅惑3 + キュート3. Raising 怪異 to three creates nine one-drops. Ignoring mulligans and looking only at the first five cards, the simple chance of seeing at least one one-drop rises from about 63.9% with seven copies to 74.2% with nine.
So the idea was logically sound if the only objective was early consistency.
The hidden cost appeared later:
Turn 7, opponent at 3: 怪異.
副総長 draws two: 怪異 + 怪異.
Adding more early-only cards improves the morning shift while contaminating the turn-6/7 topdeck pool.
Once the deck returned to ending games around turn 6, the practical answer became simple:
One 怪異 is enough.
The Goblin department has moved from three full-time employees to one temp.
5. Why Raz and Rimiru were not the current answer
デーモンドラム・ラズ and バニーな悪魔・リミル can improve board presence, but the observed problem was not “my board is ugly.” It was “the opponent’s leader still has too much health.”
If the opponent remains at 12–15 on turn 5, another sticky small body only matters if it turns into leader damage before removal arrives.
They can still matter in fast mirrors or board-heavy matchups. For the current objective — finish near turn 6 and convert draws directly into face damage — their priority falls.
Leaving a skeleton is cute. Leaving the opponent alive is less cute.
6. Suzie + Live showed why amplifiers cannot kill without ammunition
誠実なる呪い屋・スージー can give +3 attack to a Storm follower in hand, but it needs the Storm card.
夜の唱のライブ can deal two to the opposing leader when its condition is active. Against an opponent at 3, a hand of Suzie + Live with no Storm can leave exactly one health.
That extra turn can let Dragoncraft reach a late stabilizer such as イランツァ and turn the game into removal, healing, and defense.
The conclusion is not “cut Suzie.” It is:
Too many amplifiers, conditional burn cards, and early-only bodies can produce a hand that cannot convert the final few points.
There is still no clean same-39-card, same-seed Suzie 0/1/2/3 copy-count experiment. Invented decimals are not research; they are card-game astrology.
7. 副総長 went home — then returned for three damage
Six cost looks late in an aggro deck. That is why a zero-copy idea appeared.
Then a real game ended like this:
Turn 6. 副総長 deals three to the opposing leader. Exact lethal.
That changes the job description.
If 副総長 is “a six-drop that recovers the board so we can play a long game,” it is too slow.
If it is “a turn-6 clear plus three face damage that closes the game,” it is a finisher.
Prior local research also does not justify casually cutting it to zero. The constrained all-leader baseline used three copies. The 5,904-game Pirate Swordcraft KEEP_BASELINE used three. The August 28 cross-source 31-card public core also contained three.
An older legend-light Pirate specialist used one 副総長 and three 相承の意思, but many cards changed simultaneously, so the gain cannot be assigned to cutting two Deputy Chiefs.
The evidence-safe conclusion is:
Zero copies is unproven.
副総長 stays.
The exact count in the current shell still needs direct testing.
We fired management, then called management back for the closing shift.
8. Do not mix old simulation numbers
The first broad Set 9 screen covered 32 supported builds and 2,324 engine games. Aggro Nightmare scored 81.0% average with a 63.9% worst matchup. That is a directional launch ranking, not a real ladder win rate.
The constrained all-leader test produced an eight-turn score of 65.5%, with 63.1% going first and 67.9% going second. It compared 14 opposing lists, and 32 of 84 games were unresolved at the turn cap.
The Pirate-only optimization tested 96 candidates over 5,904 games, including a 1,944-game holdout. The baseline scored 91.4%, 92.6% going first, 90.1% going second, and 69.4% against the worst opposing Pirate list. That result belongs only to those 18 Pirate Swordcraft lists.
An older legend-light Pirate specialist scored 91.2% versus a 75.6% comparator, but it changed many cards at once.
The old all-leader matrix also showed Haven/Bishop at 79.2%. That number must not be pasted onto today’s Evolution Haven matchup. Same class name does not mean same experimental object.
9. “Research-based” must be separated from “this card looks good”
Use three evidence levels:
- Direct comparison: same conditions, measured difference.
- Successful elsewhere: used in another successful experiment, but isolated contribution unknown.
- Card-text hypothesis: worth testing, not proven.
相承の意思 is level 2.
旅立ちの退出 is level 3.
副総長0 is level 3.
怪異3 also started at level 3 and later gained negative practical evidence through dead late draws.
怪異1 is currently a practical tuning conclusion because the deck returned to roughly turn-6 finishes, not a universal same-seed theorem.
An honest blank cell is stronger than invented precision.
10. What the next experiment should log
The next A/B test should track opponent HP after turn 3, opponent HP after turn 5, turn-6 finish rate, turn-7 reach rate, the probability that a turn-6/7 topdeck becomes immediate face damage, six-plus-cost cards stranded in the opener, late cards that do nothing immediately, first/second split, and the worst matchup by leader.
The new key KPI is turn-6 completion rate.
If games are already ending around turn 6, a card that is “great on turn 8” is applying for a job at a company that has already closed. 副総長 survives on the boundary because it itself produces three face damage on turn 6. 怪異 copies two and three are much more likely to become 1/1 topdecks at that point.
The question is not “is the card cheap?” It is “how many points does this card become on the turn I want to win?”
11. Current practical verdict
- 達筆の怪異: one copy. Early insurance without overloading late topdecks.
- 激烈の副総長: zero-copy plan withdrawn. Turn-6 face three produced real lethal. Exact count remains untested.
- Raz/Rimiru: lower priority for the current objective. Board presence is not the same thing as leader damage.
- Immediate damage, burn, and Storm remain central.
- The practical deadline is around turn 6. The longer the game goes, the more likely healing, Ward, and large removal catch up.
The tuning process became a complete comedy loop:
“副総長 is too slow. Go home.”
“副総長 drew 怪異 + 怪異. Everyone go home.”
“Games are ending on turn 6 again.”
“副総長: three to face, lethal.”
Deputy Chief: “I already left.”
Operations: “Sorry. Just come back, deal three, and you can leave again.”
12. Final conclusion: aggro is not “cheapest cards first”; it is “damage before the deadline”
The question is not whether a card is cheap in isolation.
Ask:
Does it damage the opponent on turns 1–3?
Can it convert into burn/Storm on turns 4–6?
If topdecked on turn 6, does it still do a job?
Three 怪異 looked attractive for opening consistency. Once the deck returned to near-turn-6 finishes, one copy became enough.
副総長 looks slow because it costs six. But if six is the closing bell and it deals three face damage, it is not a late-game card. It is the employee holding the keys to the front door.
The best list is not “the cheapest 40 cards.” It is:
the 40 cards most likely to turn draws into opponent HP loss before turn 6.
Sources
- GameWith Rotation Tier: https://gamewith.jp/shadowverse-wb/497197
- GameWith Aggro Nightmare: https://gamewith.jp/shadowverse-wb/503018
- GameWith Evolution Haven: https://gamewith.jp/shadowverse-wb/565666
- Game8 Aggro Nightmare: https://game8.jp/shadowverse-beyond/698338
- Local:
SET09_AGGRO_NIGHTMARE_ALL_LEADERS_LEGEND_LIGHT.md - Local:
SET09_AGGRO_NIGHTMARE_VS_PIRATE_ROYAL(2).md - Local:
SET09_AGGRO_NIGHTMARE_VS_PIRATE_ROYAL_LEGEND_LIGHT.md
