We’re somehow big enough to get article submissions. This is an insightful article on attack order, by MrPattywagon, which I have edited somewhat liberally.
This article contains sections on ordering your attacks for information (both denying info to your opponent and fishing it out for yourself), one section about board-state sensitive attack order that you can skip if you know what you’re doing or play Old Man, and a two-part discussion on how attack order matters in dealing damage (preview: it doesn’t) (sometimes) (at least not in the way some of you think it matters). There are limits to the number of words a reasonable person can tolerate in a blog post about an anime card game – scope-creep scales linearly with boredom and tedium – but I hope the examples and explanations that survived the writing process will do something for you.
A scenario before we get into the article proper. (You can CTRL-F or CMD-F for “attack order does not matter” if you’re here to read the section on dealing damage.)
Green Railgun opponent is at 0-3 with two characters on the field and you commit yourself to killing both of them: an<a href=”http://www.heartofthecards.com/code/cardlist important source.html?card=WS_RG/W13-029″> on-reverse drop-search and a best girl oversize. When you defeat the drop-search character, your opponent will search for the card he thinks he needs. He’ll take the damage, his searcher will be reversed and primed, and he’ll go find something. But the opponent might not yet know what he wants to grab with the drop-search: he’s sensitive to his clock and how close he is to level 1.
Let’s explore some damage scenarios and how that affects his search target. Then we’ll talk about why attack order matters here.
If the first attack triggers 2 soul and sticks, dealing three damage and putting him to 0-6, the opponent knows he can put himself to level 1 by clocking during clock phase alone with no help from Takitsubo, freeing up his search for level 1 targets rather than needing to worry about level 0 at all. On his turn, he can kill your characters with level 1s and won’t bother searching for level 0 front-row. Level 0 is over.
If the first attack cancels, still leaving him at 0-3, the opponent knows that unless you trigger soul on the second attack he’ll be at 0-4 when he begins his turn. With only four cards in clock, he’s stuck at level 0 and has to keep playing a level 0 game if he wants to be able to kill your characters and deny you from attacking with them again. If forced to make a decision now – and half in the blind – he has to take a guess that a second attack won’t hit him for a higher soul count off a trigger. And he might be guessing wrong. But using this drop-search to get a 0 rather than chance on clocking into it is a reasonable path.
And in between the 3 damage and no damage possibilities described in those two paragraphs are 1-soul and 2-soul possibilities which I’ll leave unelaborated, with considerations in there about grabbing Takitsubo to advance his clock further. There could be guess work involved. But at least the first attack will give the opponent some information to let him make a good search.
The big information threshold, though, is after the second attack. After both attacks, the opponent has perfect information. He doesn’t have to guess how much damage he’ll take: he knows how much damage actually happened, and at that point can search for the right character accordingly.
Attacking the drop-search last is usually a mistake. It is a mistake because it gives the opponent more information about how much damage is going through and produces a better search. However, that’s not the only point of contention.
It’s a mistake to attack the drop-search last . . . unless you value the opponent’s decompression over this search information. Maybe there is a clear, number one valuable search target that the opponent always needs here and you can divine from mulligans that the opponent’s hand is already equipped to handle level 0. So the extra information to your opponent might not be as relevant as one less non-CX in the deck for your second attack to process through. So you attack the drop-search last.
It’s a mistake . . . unless you are not committed to killing both characters and might side the drop-search or not attack in that row at all. The opponent’s on-reverse drop-search would then live, stranding the opponent without the search and without access to some key character going into the turn. That could be a good thing for you – but maybe you can’t tell if it’s worth it yet. Maybe letting the searcher live depends on your attack against the oversize sticking a certain soul-count, which is based on triggers and a possible cancel that you don’t know, so you reserve the attack on the searcher until you find out what happens. Rather than value information denial to your opponent, your attack order values getting information for yourself. So you attack the drop-search last.
It’s a mistake . . . unless damage processing more cards before killing the drop-search last would substantially reduce the number of viable search targets left in the deck, or unless the opponent wants to pay out a CX from stock into waiting room before your attacks this turn force a refresh, or unless yada yada a bunch of other things that are probably not relevant at level 0.
So… attacking the drop-search first before attacking the oversize is correct, except when it isn’t. Good hustle, guys.
And that’s an example of what this article is about. We’ll learn about attack order and information. We’ll learn about forcing decisions on your opponent before he has better information about his options on the one hand and preserving options for yourself until you have the information you need to make that best decision on the other hand.
And we’ll also learn why attack order doesn’t matter for dealing damage.
I believe the meta on this blog is to put a gif at this point. So enjoy the break and then we’ll get to learning how to turn cards sideways.
Attack Order for Information Denial
Every decision your opponent makes is an opportunity for him to make the wrong one. Order your attacks in a way that forces the opponent to make choices based on incomplete information now rather than more complete information later. Order attacks to put pressure on your opponent’s options by controlling their timing:
“I’m thinking of playing this counter, but how much damage will I take this turn? What will my opponent trigger?” Countering is a much easier decision if you know what damage level you’re going to be at the end of the turn.
“I want access to green next turn, but the green cards in my clock are important for my later plans. Will the next damage off the clock have some green in it letting me level up with something else?” Deny the opponent that knowledge.
Here are some examples of forcing your opponent to choose a line of play prematurely.
A) When against a character which is worth countering ONLY if your opponent doesn’t level up, front attack that questionably-valuable character first.
Your opponent has a counter and a reasonably powerful character he might protect with it, but the character is obsoleted by level 3s in hand. Front attack the counter-worthy character first if you want to force the opponent’s decision. This way they have to decide before the opponent knows whether a) he’s cancelling enough to keep his Level 1 relevant, or whether b) that’s not the case and that character is not worth the counter, stock, or space on field – and indeed, would be played over if it lived anyway.
B) Against anti-damage counters, with a mix of attacks that have to be front attacks and attacks that don’t have to be front attacks, and against an opponent with limited resources available, front into the anti-damage counter first.
If the opponent has low resources – low stock, low hand – and can only afford either an anti-damage counter to reduce incoming damage/heal OR a finisher to try to kill you if the damage cancels, force the decision to counter before the opponent knows whether the counter is necessary. Crash into your opponent’s level 3 with your doomed-to-never-side character before attacking into an uncounterable slot. You deny information about whether spending that remaining stock on money counter or sac-counter is either required and near essential to survive or actually unnecessary and better used to secure the win on the opponent’s next turn. Force your opponent to counter before your triggers reveal information that would make that decision easier. Exceptions to this attack order priority are Cinderella Girls’ rest-counter and Persona’s tarot reading.
C) When the only front attack in counter-range is between your low-value character and the opponent’s low-value character, you should make that attack first.
You attack with your 1k bond into your opponent’s .5k bond, neither of which have any effects on the board anymore. It’s combat between supports. Not really high-stakes for either player and not as counter-worthy as fights between characters who are put into decks to win combat and get on-reverse effects. But maybe no front attack worth countering will come up. Your opponent’s counter might be an otherwise dead card whose only profitable use must happen right now to open up a slot for a direct attack and extra soul. That might be a fine play for your opponent – it could even be her best play – but you might as well force her to make this decision before your second and third attacks trigger CXs and make it obvious. Make the opponent commit to the direct attack value now before she gets more data to make whatever the right decision is.
D) Force your opponent to level-up with precise damage rather than create an overflow that would make the level-up decision easier.
Against multicolor decks (or god forbid something like 4-color Guilty Crown, aka “Mana Screwed”), send in soul counts that hit the opponent to exactly seven in clock first instead of soul counts that would overflow the clock. Exact damage denies information about what other colors the clock will give “for free”. Make that level-up decision harder.
“I need yellow, and I’d rather not have to clock these good yellow cards in my hand next turn, but the yellow card in my clock that I’d use for the level-up is my bond target . . . should I risk it and hope I get damaged for yellow in the next attack? Oh wait, I have yellow in the overflow damage, so I’m OK to level-up with something else.” All of us have been in that position: of getting leveled up and not liking our level-up choices but feeling comforted when we see that the overflow damage gives us access to our colors. But we’re not here to make the opponent feel comfortable about his decisions. Why attack first with a soul count that is more than the amount needed to level up the opponent? Reserve that mismatched soul count for after the opponent levels up.
And you can make good choices to deny your opponent that information – even when no single attack would appear to level up the opponent – by playing for soul triggers off the top of the deck. Say the opponent is at 1-4, you have a 1-soul attack and a 2-soul attack, and you have some number of 1-soul triggers in your deck. Normally your attack order will have no impact on the opponent’s clock – with no soul triggers, the opponent will make a level-up decision without any knowledge of overflow damage. But a good attack order respects the possibility of a soul trigger on top that can hurt the 1-soul 2-soul attack order (turning it into 2-soul 2-soul and giving 1 overflow of information) and won’t hurt the 2-soul 1-soul order (turning it into 3-soul 1-soul and still getting an exact level-up). Your attack order still has the opportunity to deny information if the top card happens to have a soul trigger, which in the blind is something you can’t count on, but you can only buy into that advantage and insulate yourself from the soul trigger possibility by attacking with the 2-soul attacker first.
Of course, in the worlds where none of your attacks go in, these decisions won’t get you anything. And in the worlds where some attacks cancel and some go in, your attack order was a coin flip and the thought you put into choosing an attack order didn’t bear fruit. But in the worlds where both attacks go in, you’ve bought into the opportunity to deny information with a smart attack order.
One more example of information denial, with a story:
Delay your own battle phase decision until after your opponent’s decision if the opponent could use that information.
My opponent had an early play, and I responded with an early-play killer (Inori) which forced a backup. Actually, I had two Inoris attacking, and with the salvage CX combo. But I front attacked the early play with an Inori first and then attacked with my second Inori in another row afterwards to deny my opponent the information that I was committing my second salvage to grabbing an anti-change sac counter.
So far as he knows, I might or might not want to use that anti-change counter against the early play. There’d be other uses for it that game, later, as a sac-counter. And, sure, salvaging the counter last also meant I delayed that salvage until a) after I knew he countered to save the early play character and b) after I knew he was still going to be level 2 because he canceled enough, both of which were necessary to keep the anti-change live (attack order for information gain, the next section).
But if I can trust my opponent to care about what I’m salvaging, my second salvage target was information I didn’t want to give away for free and wanted to deny until after his decision to protect the early play unless I had good reason to attack in a different order.
Then again, should I not respect my opponent enough to expect him to know he can direct attack into the empty row he could create with his counter and thereby avoid the anti-change counter wrecking him, I might salvage the counter first as a rattlesnake and bait him into letting his early play die, lol. You play your opponent. Every decision your opponent makes is an opportunity to make the wrong one, and it’s up to you to decide what information you let the opponent have.
Attack Order for Information Gain
“I’m about to make some attacks, but right now I don’t know something important that I’ll only learn partway through the attack phase. So let’s set myself up for that future with a useful attack order that delays my board-state sensitive decision until I know that information and can act on it.”
Mostly this means doing obvious front attacks or direct attacks first to get more information about characters that could profitably side (or profitably decline to attack at all, or profitably adjust the use of some on-attack effect) if the board state changes.
One big “level-up” for Weiss players is the first time they reduce soul counts for exactly lethal damage. Good attack orders preserve the option to customize soul counts, i.e., front attacking into level 0 defending characters, front attacking with low soul counts into high-level defending characters, or attacking with on-reverse effects first before sending in characters that could side instead for reduced soul. But we’ll get to more on damage optimization later in the section “Attack Order Doesn’t Matter for Damage (Usually)”. Here are some other ways to choose an attack order to get information:
A) Make valuable attacks first so that a decision to hold back your final attacker hurts less.
Sometimes it’s worthwhile to decline an attack (maybe to keep the opponent at a lower level or to not trigger your last card in the deck. If that is a reasonable possibility, you should do yourself the favor attacking in a way that would sacrifice your least valuable attack rather than your best.
If you make the least valuable attacks first – ones that don’t kill characters, don’t get on-attack or on-reverse effects, or whatever – and something happens to the board state (a soul trigger, or an opponent’s clock-cost backup, or two attacks that stick, or two attacks that don’t stick), that can strongly affect your next decision to turn a card sideways. You now have to weigh the potential downside of not attacking against the gains from your most important attack. You are in the sad position of giving up your best attack in order to not die to level 3s that the opponent could then play, or to not give the opponent two Mikan reversals, or to not give up the first play for position in a counter-based defender’s advantage match-up, or whatever (whoops, I did elaborate a bit). If such an important thing could happen to you and would be worth not attacking to avoid, then get your most valuable attacks in while the getting is good.
B) Attack with a character last to preserve its option to side attack if that character needs to survive some uncertain-to-happen board state.
Chris from Symphogear, Chitoge from Nisekoi2, Kanata from Little Busters, and Saya from GF Beta – these card profiles get their value from living through their own attacks to provide an effect on the opponent’s turn. They don’t want to risk getting reversed by the defending player… but their powers are high enough that, unless the opponent has a full field of level 3s, they should be able to find somewhere to front attack profitably. It’ll still come up for those guys, but we’ll look to Sayaka’s Wish for a better example.
Attack with your undersized Sayakas last and with other characters first to get more information for the Sayaka attack and your Sayaka’s Wish plan. If those other attacks go in and bring the opponent into lethal range, there’s value to crashing that Sayaka in for lethal. If those first attacks are canceled, you know to give up on front attacks and to side with the Sayaka for less damage but more counter-phase survivability. Similar scenarios can pop up with Index’s Healing Magic, Im@s CG’s rest-counter and other field-requirement counter-step options.
And at a basic level – you don’t get a counter step to play counter-step heals and anti-damage if you can’t keep characters worth being front attacked on the field. You can’t mill-counter to push through your deck, you can’t Leafa counter, and you can’t shuffle CXs back in at counter step with Glasses if you can’t counter at all.
Above, the hypothetical board state was “I don’t have lethal,” and the choice you delayed until you knew you didn’t have lethal was crashing a valuable character. Alternatively, the uncertain-to-happen board state could be “I might trigger a CX”. Nao from Charlotte benefits from a good attack order in this way: attack with Nao last to see if you trigger the pants CX (which would turn on her no-front-attack now that you could pay the cost) to better evaluate whether a front attack is worth risking Nao eating a counter.
And hey, even a mill runner can be a valuable character to keep alive later in the game if your deck’s compression is worse than your waiting room’s compression.
C) Delay a character’s attack until later to preserve options for its on-attack or on-reverse effect.
There are lots of these, but many are damage-based and I’m putting off talking about damage until later. But searches and salvages provide a lot of fodder for good attack orders. A counter could protect your level 1 field, but delay grabbing it until after you know that those attacks have gone smoothly. Same goes for salvaging a 2/1 2-cost refresh counter: not immediately too valuable if the triggers end up clean and your deck remains compressed, but the best card you could have in your hand if you trigger multiple CXs and have none left in the deck. Some cards can have high value, but that value ceiling is dependent on a specific board state, a board state that you can’t know will happen until partway through the attacks.
Attack Order Based on Board State
Attack order is sometimes a matter of simple interactions between specific cards and zones, with everything on the table and no information to fish for or deny. Sometimes you choose an attack order for nothing more complicated than “this card gives power to my other characters when it attacks, so I guess I’ll attack with it first.”
None of the examples below deal with information, and most will be obvious to experienced players, but an article that purports to explain and demystify attack order has to mention them or it does a disservice to people learning the game. This section has to be here and that’s just how it is, but I’ve limited us to the need-to-knows and stuff I just happen to think is interesting.
Compression (if you believe compression is meaningful, there’s a lot of attack order rules for it):
- Attack for enough soul to send the opponent’s clean clock to waiting room to decompress the future deck before attacking for a higher soul count that would refresh the deck immediately and leave those non CXs out.
- Side or direct attack first to force a refresh before making counter-worthy front attacks to deny the opponent from paying out CXs from stock into that refresh. Or if that stock is clean, make the front attack first.
- Front attack a counter-worthy character first against self-mill counters when the opponent’s deck is highly compressed and the mill would refresh the opponent’s deck into a less compressed state. The opponent either saves the character or loses out on the value of those guaranteed cancels. Or if the mill counter would mill through a clean deck, delay that attack until it won’t.
- Attack with early plays before any other front attacks to insulate yourself from anti-change counters. Or attack with early plays before the opponent even hits the level requirement to play anti-change counters.
- Blind stockcharge last to pay out those possible CXs into a refresh more easily.
- Delay on-attack pay1 stuff, like Asuna Invites to Party searchers, until later in your attack order to pay out CXs that are triggered by earlier attackers.
- And for searchers specifically (like Greenodera or Yuuki) or climax-back-to-deck effects, attack with them last to delay the increased compression until you’re almost done triggering cards to stock. i.e., save that compressive effect for the opponent’s attacks rather than waste more of it on your own triggers than you have to.
- By contrast, attack first with searchers that have limited search targets (Little Busters Rin, or any searcher if you only have one copy in deck of the search target in mind) to avoid triggering them to stock and losing out on your search.
- Attack into 1/1 2.5 splittable counters with an order that prevents the opponent from efficiently allocating that split power. Force the 1.5k boost on the character who only needs 1k.
- Attack with salvagers last to avoid anti-salvage.
- Order your clockbombs and antichange bombs to connect when the opponent’s clock/level will still meet your bomb condition. This is easy to forget about and easy to screw up.
- Build stock with other attackers first to meet stock thresholds for characters like the Clannad cost-bomb or Kuro or characters who need to use that stock to pay for expensive effects.
- Or attack with other characters first to hope for CX triggers: plussing CXs if you need cards in hand to discard for expensive effects, wind triggers to open up a direct attack or make a front attack safer.
- Shuffle cards back into the deck with these on-attack Honoka effects first to set up more soul triggers or clean triggers – whichever is appropriate. This is low on the list for a reason. Don’t do this unless you need to.
Lastly, don’t walk into Old Man. This card profile demands you put actual, serious thought into the whole attack phase and especially into your attack order – and this game is all about setting up and executing profitable attacks, so you could say once this effect is on the board that the entire game becomes a game about Old Man. And games in which both players have Old Man create pretty interesting multi-level-thinking combat math and are a real joy to play. Wait, sorry, I meant to say those games are truly cancerous and the worst experiences in Weiss I’ve ever had. Fuck this effect. People don’t play these cards because they’re good. People play them because they get perverse pleasure watching their opponents calculate and recalculate powers fruitlessly, waffle back and forth between different attack orders, and shuffle characters around the playmat until they resign themselves to being unable to field-wipe and settle for what looks like the best option only to realize after the first attack that they still fucked it all up.
Go away, Old Man.
Attack Order Doesn’t Matter for Damage (Usually)
Attack order does not matter for customizing your damage or getting different amounts of damage to stick. Sometimes. At least, it doesn’t matter in the way you think it does if you think like this:
“I only need to do 1 damage, so I’ll attack for 3 first to get any CX near the top out of the way, and then the 1 damage afterward will be more likely to go in.” <— no, that’s nonsense.
“He triggered a CX last turn, so ordering my 2-soul attack before my 1-soul attack this turn takes advantage of that weakness.” <— no, that’s wrong.
“Forcing” climaxes on X damage to get through Y damage is incoherent except in special circumstances, with more information about how CXs are distributed in the piles of cards in front of you, and interpreting “forcing climaxes” pretty loosely. Attack order can matter for damage, but – so far as teaching you something goes – it’s better to treat those as exceptions to the general rule, exceptions which long-time players have internalized and which I’ll explain later. As a rule, though: attack order does not matter.
Intuition is different for different people. For me, this is intuitive, but it might not be for you. I’m not a good enough math person to do this the proper way, let alone explain that proper math in a way that fits your intuition. So I’m going to try two different approaches. First, a brute-forced spreadsheet of attack orders against all possible configurations of some deck set-up. Second, a thought experiment. OK? Let’s go.
Say you have two direct attacks, with no other decisions available to you: a 5-soul attack and a 2-soul attack — it’s a weird setup, but that isn’t important, what is important is the lesson it teaches. Here’s the CX distribution for 2 CXs in the 10 card deck your attacks are up against.
45 possible configurations of CXs and non-CXs (though I’ll note for a second that if you particularly care whether you cancel off a book CX as opposed to a stock-soul CX, you can treat otherwise identical cards as nonidentical and all these numbers go up proportionally). 45 different ways that the opponent’s deck can have its CXs and non-CXs distributed. 45 shuffle possibilities.
Against those 45 configurations we’re throwing two attack order variants. World One is the 5-soul attack first, then the 2-soul attack. World Two, we send in the 2-soul attack first, then the 5-soul. Or, putting it another way, World One is Akane and World Two is Pumpkin. We work those attacks out against each row on the table and see what happens.
So…how does our damage pan out? How much gets through the opponent’s cancels?
10 configurations of no damage at all,
25 configs of 2 damage,
7 configs of 5 damage, and
3 configs of 7 damage.
And in World Two (Pumpkin) – with the 2-soul first, 5-soul last:
10 configurations of no damage at all,
25 configs of 2 damage,
7 configs of 5 damage, and
3 configs of 7 damage.
Those results are identical. There is literally no difference. In the blind, out of all possible shuffled configurations of the opponent’s 2 CX 8 non-CX deck, with no more specific information, and no option between siding or fronting – choosing an attack order to select a preferred amount of damage is not possible.
And we can do this again with three attacks.
Say your attacks are 3 soul, 3 soul, and 4 soul, with no soul changes from triggers or anything else. And the opponent is at 3-3, with 10 cards in deck and 2 climaxes. Does attack order matter? Again, no. And I’ll hope you’ll humor my only crunching a set-up with 3 unique attack orders rather than 6 because it really is tedious doing this by hand. I listen to podcasts while I do it, and occasionally it’s kind of zen, but still. Spreadsheet is here.
These non-customizable attacks in the blind can’t be ordered to deal more damage in a rational way. I leave a more rigorous explanation for why that is to a person qualified to do it. Maybe that person will be motivated by this post. Until then you can settle for my hand-done tables or move on to this next explanation.
Suppose I have 5 cards in a deck: a book CX, a stock-soul CX, a level 0 character, a level 1 character, and a level 3 character. There are no level 2s in the deck because no one outside my playgroup uses those things and I don’t want to confuse you with a thought experiment too unrealistic.
You attack me three times. The first attack is canceled on the first card – the book CX – leaving 4 cards in the deck: 1 stock-soul CX and 3 non-CXs. You have a 1-soul attack and a 2-soul attack still available, with no options to customize soul by siding with them or anything else to affect their soul. That’s important. No other soul customization. 1-soul and 2-soul. That’s what you’ve got. But does the attack order matter?
“I want to attack with the 1-soul attacker first and hope I get lucky enough to pick off the stock-soul CX so that my 2-soul attack can get through.” ← no. That’s silly.
“You just canceled, so I bet the stock-soul CX in your deck is pretty far away from the top. I’ll attack with the 2-soul attacker before the 1-soul attacker to take advantage of that.” ← no, and that’s even sillier.
You can work out for yourself that against all possible configurations of those 4 cards the damage works out identically whatever the attack order. You can do the tables for yourself, see how much damage Attack Order A does and Attack Order B does, and you’ll realize you’re flipping a coin. But I’ve already attempted that kind of explanation so let’s try something different.
The original 5-card deck, with 2 CXs, had 120 different configurations if we treat each CX as unique cards and each non-CX as unique cards. That 5-card deck can be shuffled into 120 different outcomes.
Of those 120 configs, any one card will appear on top one-fifth of the time, or 24 times. So 24 of those 120 configurations have the book CX on top. Well, hey, that’s the world we’re in: the world of having the book CX on top, cancelling with it, and leaving 4 cards remaining in some random order.
My question to you is whether any one of those 24 card arrangements we’re talking about now – whether any particular order of stock-soul CX, level 0 character, level 1 character, and level 3 character left in the deck – is more likely than another.
The answer is of course not. No arrangement of those 4 cards is more likely than any other.
So why do you care? Why try to justify to yourself making the 1-soul attack first or the 2-soul attack first as if it were more or less likely that the remaining CX is near the top of the deck? as if one 4-card arrangement were more likely than any other?
There are 6 configurations of the last CX being on top, 6 of being second, 6 of being third, and 6 of being on the bottom. They’re all equally likely. The fact that I just canceled is relevant only in so far as the climax ratio of cards remaining in my deck is much worse for me as the defender and much better for you as an attacker than when the turn started. I’m less compressed – sure. Your attacks are, on average, more likely to go in now – yes. But this gives you zero insight into how cards are specifically distributed throughout my deck. That a book CX was on the top before gives you zero indication as to whether the stocksoul CX is at the top, is second, is third, or is at the bottom of the 4-card deck left in front of you.
You cannot differentiate between attacking with the 1-soul attacker or the 2-soul attacker first except by which of them is signed, because we know attacks from prettier cards are more likely to go in, or – and I will only say this once and never bring it up again, because this should not be a part of the game if you respect yourself – you are shuffling your deck and your opponent’s deck poorly enough for them to be insufficiently randomized and are knowingly taking advantage of that.
To customize damage, you need more information than a black box. And you need the option to use that information for attack order to matter.
What kind of information makes attack order matter for customizing damage?
Because, yes – having made my attempt to emphasize the rule before talking about these exceptions – there are several ways for your attack order to matter and customize your damage. Let’s go through them.
A) You have options between siding or fronting that you are willing to choose in response to large changes to your opponent’s compression mid-attack phase.
You make a first attack (preferably one that’s “obvious,” whose value is so great or so one-sided in favor of front attacking or side attacking that you’d never change that plan – see our previous discussion on information gain). Whether that first attack cancels or whether it sticks, your opponent’s deck is different than it was when the turn started. You just now damage-processed some cards off the top of it. The deck’s different. And that’s information you can use. You have information about how much more or less compressed your opponent’s deck is, which can affect your future decisions to side attack or front attack to customize your soul count.t, and you bought into that opportunity to use this information by preserving those side attack options for later in your attack order.
Say you’re against a moderately compressed deck with few cards (a small deck is useful for an instructional hypothetical because even one card removed from the deck has a significant impact on the ratio, given how few cards were there in the first place). So anyway, small deck, decent compression, but your field of attackers is mostly low-power characters boosted by a global soul CX. If that first attack cancels on the first card, crashing your undersized characters into walls becomes pretty attractive now that the chance of actually sticking that front attacking damage is improved.
If the first attack gets through for 4 or 5, though, those crashes look awful. Because the deck you’re facing has 4 or 5 nonCXs sucked out of it, the CX ratio is much better for the opponent and frontal attacks for 2 soul might now be so likely to cancel that it’s not worth opening up your rows to free direct attacks or decompressing your WR with dead characters going into a refresh. So maybe you side for one and at least that chip damage might get in.
What’s so different here from the “attack order doesn’t matter” scenarios that introduced this section is that here you have the option to customize your soul count on-the-fly in some of your slots. Delaying your soul-customizable attacks until later in the attack order preserves that option to customize them. You preserve the option to side with attacks that could side if the situation ends up being right for that. Attack order mattered in that you ordered your non-customizable attack first.
And if you follow through on the “Information Gain” section, this scenario won’t need to be planned for or even conscious in your mind – you’ll naturally choose to make your “obvious” attacks first, it will naturally happen that your “obvious” frontal attack might swing that compression one way or the other, and you will be handed the opportunity to make a good decision with that information. And it’s not only the opponent’s compression that can change mid-attacks – your opponent’s clock is changing, too, and that’s what siding for lethal is all about. So, you’ll front attack with a Musashi because her on-cancel burn has built-in insulation against a front-attacking soul count while reserving your option to side with your other level 3s and go for exact lethal, and in doing this your attack order fishes for information about how much more damage you need to do while keeping in reserve the best options you have to use that information once you get it.
B) You know what the top card of the opponent’s deck is and can package your preferred amount of soul damage into a lower soul basket.
You topdecked a non CX from waiting room with a Gandr Shooter Rin, or noticed that the opponent’s last attack was an on-reverse top check that failed and revealed an event, or you reversed a character with topdecking from waiting room as a drawback, or whatever. If you have that information, the math from those Akane/Pumpkin spreadsheets earlier completely changes. Of the 36 remaining configurations (here we’re back to treating individual CXs as interchangeable and nonCXs as interchangeable because it’s easier and fine to do) you end up with skewed, imbalanced damage counts depending on attack order. Neither attack order is strictly better or worse, but they are different and they target customizable amounts of damage.
Here’s the breakdown:
In the Akane-style 5-soul attack first, 2-soul attack last World One, against a 2 CX, 8 nonCX deck with knowledge that the top card is a nonCX:
8 configurations of no damage at all,
18 configs of 2 damage,
7 configs of 5 damage, and
3 configs of 7 damage.
5 configurations of no damage at all,
25 configs of 2 damage,
3 configs of 5 damage, and
3 configs of 7 damage.
Basically, you can put a 5-soul attack into a 4-soul basket: you can try to get 5 damage to stick without the actual risk of a 5-soul attack, since you’re only working with 4 unknown cards rather than 5. Or you can play conservatively. Apply that credit to secure the 2-soul hit by putting it in a 1-soul basket while leaving the 5-soul attack to brave it alone and unassisted.
So attack order matters a lot if you have that extra top-card info, and – whether you value X amount of damage or Y amount of damage – you might choose one attack order over the other. You might make a sacrifice and go to World One/Akane which has three more configurations than World Two/Pumpkin of doing no damage at all if you really need to squeeze the math toward the configurations of 5+ damage this turn to have a chance of winning.
Or, if you need merely a single point of damage to win because your opponent is at 3-6, then of course you should leverage that top card to make the 2-soul attack even more likely to go in with World Two/Pumpkin. If against someone at 3-6 with 2 CXs in a 10 card deck with a top card known to be clean, attacking with the 5-soul first wins only 28 out of 35 times, while attacking with the 2-soul first wins 31 out of 35 times.
Alternatively, if you know the top card is a CX and your first attack will be canceled for sure – well, send in whichever soul count is your least valuable. Or a Musashi burn.
C) You know your attack phase will force a refresh after your attacks begin but before they end and the pre-refresh and post-refresh CX ratios will be very different (creating what I’ll call a “compression cliff”).
A mid-attack phase refresh will split your attacks between two CX ratios: some attacks will fight against Deck Composition 1 pre-refresh and some attacks will line up against Deck Composition 2 post-refresh. Maybe it’s as simple as your damage counts are high enough and numerous enough to damage process the rest of the deck despite some cancels, but there could be a counter-step refresh lurking in your opponent’s hand. And Either way, before that refresh, the waiting room’s CX ratio is open information. If you also know the deck’s CX ratio beforehand, you can compare the compression of those two stacks of cards and choose an attack order that targets a specific damage amount: if you know a square hole is coming up, you can save your square pegs until then.
We can hash this out with anan example. Opponent has a 2-card deck (one CX and 1 nonCX), and a waiting room of 3 CX and 24 nonCX cards. Compression is decent for the opponent right now but it’s about to fall off a cliff. We have attackers with 1 soul, 2 soul, and 2 soul. Even though some of our attacks have no chance at all of getting through the 2-card pre-refresh deck, some attacks have an easier time post-refresh than others, creating skewed and imbalanced damage profiles.
If you want to wade through themy numbers and experience mild discomfort when you see that differences in damage customization exist but aren’t proportionate to the amount of effort put in to show it, go here. Rounding errors mean the %s won’t add to 100.
World One combined (1-soul first, 2-soul second and third):
1.08% for 0 damage
0% for 1 damage
25.22% for 2 damage
42.46% for 3 damage
30.89% for 4 damage
i.e., only 1 damage sticking would be a disaster. I must get in at least 2 damage this turn, so I’ll set my 2-soul attacks last so that at least one of them is positioned against the weaker CX ratio (and when I look at the two stacks of cards in front of me, the weaker CX ratio happens to be post-refresh). OR I must get in 4 damage this turn and (ignoring soul triggers) this is the only attack order that can even get numbers that high.
World Two combined (2-soul first, 1-soul second, 2-soul last):
0% for 0 damage
23.73% for 1 damage
5.13% for 2 damage
71.11% for 3 damage
0% for 4 damage
i.e., I must get in something, anything, this turn – maybe I need just 1 damage on top of refresh point to get lethal. So I’ll position the 1-soul attacker against the weaker CX ratio. OR I actively do not want 4 damage to happen. 4 damage would be a disaster.
World Three combined 2-soul first, 2-soul second, 1-soul last:
.8% for 0 damage
16.62% for 1 damage
11.14% for 2 damage
71.11% for 3 damage
0% for 4 damage
i.e., I’m a confused person who chose a damage profile that’s sorta normalized and in the middle of World One and World Three. I don’t know what I want.
Anyway, bottom line being: attack order does matters when facing a compression cliff – a discrepancy in CX ratio between the stack of cards in the opponent’s deck right now and the stack of cards in the waiting room, a waiting room that will become the new deck in the middle of our three attacks. And yYou can choose your attack order to position your preferred damage amount up against the lessor of those two compressions. Arrange some characters to attack into the heights of the cliff and others to attack into the valley.
D) Your opponent can manipulate the top of the deck during your turn, i.e. defensive scry.
This scenario has two conditions but they’re pretty straight-forward. First, there needs to be enough nonCXs in the opponent’s deck to create deck configs with nonCX pockets large enough that two back-to-back attacks could optimistically go through together (if you ignore the defensive scry). This is almost always the case: unless your opponent has almost no nonCXs in the deck at all, there will be some configurations of the deck that put a few of the nonCXs next to each other.
And second, your opponent has on-cancel scry, like Chris from Symphogear.
So, given all that, since one attack’s success can set up the next attack for failure, sacrifice equity on your less valuable soul count by sending it in after your preferred soul count.
Normally, with no scrys going on, every configuration of the deck has an equal and opposite counterpart to make these attack order decisions a wash and even out. We saw earlier that attack order doesn’t matter (usually). So for those of you who respond that any attack’s success sets up the next for failure – you’re sorta right. But defensive scry makes this actually matter to us and our decisions.
If you have a 2-soul attacker and a 3-soul attacker, a 5-card nonCX pocket lets ‘em through and both attacks will stick no matter the order. But if your opponent has a Chris, any 5-card non-CX pocket now becomes a tough fit and only the first attack will stick, and there’s no parallel configuration of the shuffled deck that would take that same attack order but let through the second attack instead of the first.
Were this as simple as “in these top 5 cards, the CX is either near the top or the bottom,” your attack order is a non-decision and doesn’t matter (despite one attack’s success making subsequent attacks more difficult) because any deck configuration that lets one attack order through for X damage and a different attack order through for Y damage has equal and opposite deck configurations that lets those same attack orders swap damage numbers. There’s no decision-making there.
But against Chris, you face a difference in damage output and the two attack orders aren’t equal. You’re making a real decision.
E) You know what the top card(s) of your deck are.
This is the least interesting and most obvious kind of information so it comes last, but still bears mentioning. Your topchecks, scrys, and look at top 6 ditch 3 rearrange the rest effects tell you your future triggers. That trigger information lets you customize your soul counts. By attacking with a certain character first and buffing it with the soul trigger you planned to handle, you can put extra soul in a useful place (a 1-soul character when you need to hit for 2-soul) or the least risky place (dumping the soul trigger on a Musashi burn to insulate you from the added risk of cancelling that an extra soul creates).
The fun part of all this is when multiple attack orders are justified by different reasons – when there’s a tension between several desirable outcomes, like with the on-reverse drop-search scenario that opened this article. Weiss is full of this kind of decision-making. It’s a part of the game that I enjoy. And having read this many words about attack order, maybe now it’s a part of the game that you hate. I don’t think it’s possible to wade this far into a topic without bearing a grudge against it for a while.
But once the grudge is gone, the hope is we’ve internalized a few things, damage will make room for other concerns when we pause before turning cards sideways, and the game will be richer for it.