The focus for Shadow Paladin's witches is on grade 0 manipulation, and this extends beyond mainstream Shadow Paladin's capacity to superior call grade 0s to fuel retire skills; the witches can also cast spells that transform the opponent's rearguards into grade 0s. This gives a form of true field controlling surpassing its aggressive counterparts Kagerou and Narukami, which allows you to effectively punish the opponent for guarding, lock them out of recycling trigger units through legion, and deny them of their key rearguards while ensuring that their lanes can't actually deal damage to your own vanguard. These come at lower cost compared to retire and lock skills, in exchange for being highly conditional and occasionally dependent on chance.
Fianna is the primary early to midgame boss for the deck. She has a once per turn Nemain-like advantage skill built into her, retiring a unit with a same name as one of the vanguards to draw two cards. Because of her wording the possible retire targets for this expand from three to six the moment she is in legion, but the primary use is to retire extra copies of herself to draw more cards than you sacrificed and build towards having four cards in the drop zone to initiate legion with.
While in legion, Fianna gains a second skill. For counterblast 2 she can reveal the top ten cards of the opponent's deck and force the opponent to call two grade 0s over two occupied rearguard circles of your choice, sending the original rearguards to the drop zone in the process. This skill remains useful throughout the fight as a means to consistently weaken the opponent's rearguards, and over multiple turns can reduce them to swinging with 10000-power rearguard columns. If they choose to call over the grade 0s with new rearguards during their own turn, then for counterblast 2 you've effectively retired two units of your choice at no extra cost while also removing trigger units from the deck--a level of control that Narukami could only dream of. However, Fianna is an extremely costly unit to use, and effectively only activates twice per game under the current circumstances we've described. The witch subclan does have unflipping options that make her substantially more viable, but as discussed further down these options may not be to every fighter's liking.
Rias' on-legion skill retires one of your own witch rearguards to reduce the opponent's power by -5000 for each of their grade 0 rearguards, at the minimum giving the mechanical equivalent of a Soul Saver Dragon power boost to your own field. Rias can easily surpass the likes of “Яeverse” Daiyusha and other power-decreasing cards by using her skill when the opponent has three or four grade 0s in play, reducing the typical 11000 power base to -4000 or -9000 power. The fact that negative numbers can come into play alone makes her a highly compelling grade 3 option, as with four grade 0s on the field those 19000 power lanes from subclan attackers will be very relevant, requiring ~20000 shield to block. It's telling that even in Extreme Fight format, Rias is the most popular grade 3 option when all of the Genesis witch bosses are on the table.
While Fianna and Rias are both solid grade 3 options, a chief obstacle to the witch decks is the need to support their counterblast costs and make rearguard formations that can compensate for these cards not having innate power skills. Other legion vanguards of the format will attack for at least ~22000 power to demand high shield counts from the opponent, but all of the witch legions at this time are limited to 20000 power exactly, which on the typical defense is no different from a 16000 or 18000 power lane.
The intended first vanguard for this Shadow variant is Witch of Banquets Lir, but Creeping Dark Goat is also viable for this deck. Generally your choice of FVG will swing based on your trigger lineup. Lir's skill lets you put her into the soul to search the deck for two grade 0 units and call them at rest after she boosts an attack that hits while your vanguard is in legion and if the opponent has at least two grade 0 rearguards. This may appear useless on the surface, as it's a +1 that removes trigger units from the deck, but the rationale behind Lir will become more clear when you consider her potential call targets.
The witches as a whole do not require especial counterblast, which gives them access to the entire gamut of Shadow Paladin triggers. That means you can do things like 4 draw 8 critical lineups with three types of draw and five types of critical trigger to deceive the opponent into thinking it's 6 draw 6 critical. However, witches also have access to the first skilled trigger unit in Shadow Paladin, Witch of Goats Medb.
Medb is a stand trigger that when placed on the rearguard can soulcharge 1 to unflip a damage if the opponent has two or more grade 0 rearguards, and if you have a witch vanguard in play. The soulcharge skill is useful because of how dependent the witches are on utilizing the soul to pay for costs, and two Medb can be searched out by Lir's skill, building soul and card advantage so that Medb can be sacrificed for retire skills later on. As witches are already very effective at drawing cards and cycling through the deck, you can effectively substitute Medb in for draw triggers to offset the high counterblast costs of the deck's boss cards. It may be tempting to run a rainbow setup of triggers with 4 heal 4 draw 4 stand and 4 crit, but as a whole the combination of stands and critical triggers are a more cohesive way to fight when your deck already has draw power built into it. Critical triggers will get the opponent to the point where stands are the most dangerous trigger in the game faster than draws will.
How often you'll have an opportunity to use early game skills is debatable. Grade 1 soulblasts are unlikely to go off on the turn that you would ride one of your sentinel-bearing cards, and 10000-power attackers usually make no difference on the first turn because most grade 1 vanguards will be at 7000 power in the current format. Hence, the only time that Graine activating subclan attackers will matter is against base 8000 vanguards, base 5000 vanguards and when the subclan attacker is boosted. For reasons discussed further down, it's a generally bad idea for witches to have a booster on the field on the early turns of the game, so as a whole Mac Lir is preferable to Grainne.
The subclan's grade 1 attacker Scathach can help generate this aggression, by exploiting the idea set in the opponent's mind that damage and defeat are one and the same. Her attack persuades the opponent to drop a trigger unit to avoid taking early damage, in the process exposing themselves to Dana and Fianna's curses in the following turn, playing off of the fact that they consider losing their field more acceptable than taking damage.
Normally grade 1 attackers serve to get the opponent to drop a shield unit early that later on would have been used to defend an on-hit, preemptively activating that skill later on, but Scathach serves a dual purpose by making the opponent vulnerable to having their field controlled. Versus witches it's often better for the opponent to take damage than it is to guard at all, because the moment one starts defending with grade 0s they instantly become unable to play rearguards without losing them. The alternative is to defend with grade 1s and 2s, which removes useful rearguards from the hand while leaving less useful grade 0s still around, unable to be used either as rearguards or for legion.
If you don't do this and try to fill the field early to use multiple Arianrhod, Scathach or Dana, when it comes time to take advantage of what may be your only chance at Lir going off, the current field will be too clumsy to support it. When using Lir with a full field you'll stumble into calling a Medb over your existing rearguards, weakening your formation and preventing you from building a +1 off of Lir's skill. If the only open rearguard circles are the one behind the vanguard and in a column that already has a grade 1 on it, you may be forced to call over the rearguard column later.
Phantom Blaster “Abyss” decks have successfully used Arianrhod in the same conservative way while running a playset of her, but four copies is not strictly necessary in this deck. Three can be sufficient enough to get her out regularly, especially since Dana's soulblasting can serve the same purpose. At the same time, using either Dana or Arian early game can clue the opponent in to the fact that they won't be seeing the other one anytime soon. Which one you use depends heavily on the kind of hand you draw into, and how well you can read into the current situation.
Witch of Quests Securna is where things get complicated. Her actual effect is to retire one of your other witch rearguards to give the opponent's vanguard -5000 power. While she can only be used if your vanguard is in legion and if the opponent has at least two grade 0 rearguards, her skill is actually useful at every point in the game in legion format because of how much more valuable each point of damage is. The deck's acceleration options mean she can go off as early as turn 3 or 4, and she's vital to on-hit skills.
So you don't really want her at four copies, except that you do. At least three copies will help get her off consistently, but two is also an option, and how many you run comes down to individual experience. You also need to account for Securna's 6000 power base, which necessitates having either a 12000-power attacker or base 10000 power unit in front of her.
While she falls short of being as much of a threat to crossride units, the number of crossrides in the post-November game is severely diminished, so Femme will not be going out of style anytime soon.
Witch of Attraction Adora is the one witch that the English format has yet to see. Released as a promo card in part of Cardfight Pack 14 in Japan, and expected to be released in Pack 12 in English, Adora is the first in the witch series to require especial counterblast. When her attack hits a vanguard, her counterblast 1 reveals the top 5 cards of the opponent's deck and forces them to call a grade 0 unit from there over an occupied rearguard circle of your choice.
Adora effectively achieves Rias' skill for less counterblast in exchange for being an on-hit, and synergizes well with the witches' existing impetus to make early attacks that capitalize on their stand triggers and get the opponent to drop shield before Adora comes out. Her primary weakness is the especial counterblast cost, which limits your trigger pool much more significantly and encourages using witch quintet walls over non-witch Mac Lirs. Adora is an ideal early game ride and a counterblast that you don't have to dedicate to until your attack hits, and unlike the other non-legion support she's live the moment she hits the field, so you don't have to wait for an opponent to defend with a grade 0 first.
She has also aged well compared to the rest of BT04's card pool. Nemain's weak 3000 power base, so low that no other grade 2 in the game has anything on part with it, has been overhyped for what it can do to your turn 2 ride. The reality is that most opponents will not have the optimized hand for filling the field if faced with her, and trigger rushing leaves them exposed to activating your Securna and Rias for you further down the line. Her strengths outweigh her weaknesses. The opponent seeing you ride her also lulls them into a false sense of safety, causing them to believe they have already won based on these early turns when they ought to have their guard constantly up against a witch deck.
Overall the witch series is more difficult to play but also more rewarding than their Revenger counterparts. A witch deck played poorly will do significantly less than an “Abyss” or Raging Form deck played poorly, but a witch deck played well will exceed their contemporaries in every respect. The subclan is made and broken by the skill level of the cardfighter commanding them, requiring unorthodox planning around grade 0 calls and an appreciation for the properties of legion. Those more familiar with the Revengers will find themselves challenged and potentially turned away by the witches' opaque card skills, but seasoned Shadow Paladin cardfighters that first cut their teeth on Phantom Blaster or Ildona will find themselves right at home with the mix of advantage-based play and careful consideration of costs. As with their predecessors, there is no direct "win condition" compared to the competing restand decks within the clan, but a combination of related skills that can each push the game in their own respect, until you've won before the opponent can ever realize it.