Author Topic: Theorycrafting! (Because I like competitive metagames too much)  (Read 133910 times)

metroid composite

  • m_ACac
  • Administrator
  • Denizen
  • *
  • Posts: 4381
    • View Profile
Re: Theorycrafting! (Because I like competitive metagames too much)
« Reply #975 on: October 13, 2023, 09:51:24 PM »
Firebat

TvT: 0/4.

TvZ: 3/4.  Yeah, honestly, fine they're seen about as often as hydras in this matchup.

TvP: 1/4.  I've seen the "land barracks behind a pylon walled base and build a firebat play" like...maybe 2-3 times in ASL now?  It's quirky, it's rare, but it happens.

ZvZ: 2/4.  It would break the symmetry of ling walls.  Yeah, early firebats in small numbers are not that great against lings, but if your lings are in front, and firebats are behind, it's great.  Obviously the zerg player without firebats can still win--just get to mutas that's still what matters.  Turtle with spines if you need to, firebats can't break those.  But there's a cheese here, and it's probably similar to something like DT rushing.  Although...ZvZ is a full information matchup, due to the overlord sitting in each base, so that blunts the power of cheeses slightly.

ZvP: 1/4.  Is there a world where firebats get used against zealots?  Eh.  Basically the same damage as zealots, but like 40% of the durability.  Cost only slightly less than a zealot.  Would need to hit two zealots with the AoE to be worth it.  But...hitting two targets sounds doable if zealots are coming out of a narrow choke like the walled-in protoss base, punishing the protoss for moving zealots forward against hydras.  I'm not sure if that's better than just...more hydras, but I'll split the difference between 0/4 and 2/4 and call this 1/4.

ZvT: 0/4.  No, just make lings.

PvP: 0/4.  No.  Just make dragoons.

PvT: 0/4.  Possibly the worst unit you could make against vulture tank.

PvZ: 1/4.  If you can catch zerg not having hydras yet, AND you don't let your firebat get totally exposed away from the zealots where zerglings could surround them, yeah, maybe there could be a decent attack here.  Foiled by sunken colonies of course, but there might be a cheese that works against very specific builds here.

---

Firebat: 8/36

Ahead of a bunch of ultra-lategame stuff, cause "well maybe there's a cheese with these things."

metroid composite

  • m_ACac
  • Administrator
  • Denizen
  • *
  • Posts: 4381
    • View Profile
Re: Theorycrafting! (Because I like competitive metagames too much)
« Reply #976 on: October 13, 2023, 10:18:25 PM »
Ghost

TvP: 1/4. Ghosts were experimented with for a bit by Royal in ASL, but he seems to have given up on those experiments; there's now nobody making ghosts.  I guess I'll give them a 1, cause I'm giving scouts a 1, though I hesitate to even give them this much credit cause they have been abandoned.

TvT: 0/4.  A long, long time ago it used to be that nukes could come up in this matchup in stalled out games but...I haven't seen those recently.

TvZ: 0/4.  No.

PvP: 0/4.  Maybe lockdown could be okay on an enemy reaver, but that's so much tech, High Templar are easier to tech to.

PvT: 0/4.  No, there's not even really something you want to lockdown, and usually terran is not spread out enough for nukes to be a consideration--too obvious where to scan, and tanks outrange nukes

PvZ: 0/4.  No.

ZvZ: 0/4.  Well, they're all concussive damage, but maybe that's okay?  Zerg has all small units that take full damage from concussive.  Wait, no that's not remotely ok, their stats are worse than hydra stats.  About the same cost (more gas heavy) but 45 HP.  Their damage is slightly better against small units, but only slightly.  Also, they're super deep in the tech tree, like hive tech probably.

ZvT: 0/4.  Is there a world where you want to use lockdown on science vessels?  Ehh...no probably not; plague or scourge probably a better use of your resources most of the time.

ZvP: 0/4.  If ghosts were an earlygame unit, I'd be a little intrigued by them here, but they are not.

---

Ghost: 1/36

metroid composite

  • m_ACac
  • Administrator
  • Denizen
  • *
  • Posts: 4381
    • View Profile
Re: Theorycrafting! (Because I like competitive metagames too much)
« Reply #977 on: October 13, 2023, 10:56:48 PM »
Goliath

Oh, an actually good unit.

TvT: 4/4.  Just an absolutely core unit in this matchup.

TvP: 3/4.  You'll have occasional games end early without these, but they almost always get made even when carriers are not on the table.

TvZ: 2/4.  Mech isn't that popular against zerg, but when it is made there tend to be a lot of goliaths.

PvP: 1/4.  I'm very skeptical, but maybe you make one to zone out the enemy shuttle thanks to 8 range.

PvT: 0/4.  Terran just doesn't make enough air units to consider these over goons.

PvZ: 1/4.  Made out of the robo makes me fairly unenthusiastic here.  They are better than dragoons agaisnt mutas and lings, but not hydras which are the bulk of the games.  I'll give goliaths a bit of credit, though, they do make robo openers a bit more safe against muta openers since now your robo can help with antiair.

ZvZ: 3/4.  So...a bit of bookkeeping--goliaths would require two larva to make, so they aren't all that larva efficient.  They'd also be...IDK, probably a lair unit, so not really too much cheese potential.  That said, they're fairly good against both mutas and lings.  Not god tier, mutas can dive on small numbers of goliaths.  Lings can surround a small number of goliaths.  But I do think they would be made reasonably often.

ZvT: 2/4.  Would these be made over hydras?  I'm...not sure actually.  They are worse against units that were hit by HP->1 as hydras attack faster and cost less.  But they are better against marine/medic that haven't been hit by HP->1.  And they are better at killing science vessels--8 range is actually huge here.  But on the other hand, they would become biological and that makes them much better irradiate targets than hydras (hydras cost less, so less loss from being irradiated, and also they can cancel irradiate by morphing into a lurker).  They are also not the pick if the terran has tanks.  My gut says goliaths would sometimes get built, but it just wouldn't be that big of a deal.

ZvP: 2/4.  So...here's a question.  Are they better than hydras against zealots?  It takes 10 hydra shots to kill zealot shields, and then 23 shots to kill the zealot.  So one hydra takes about 21 seconds to kill a zealot.  Goliaths kill a zealot in about 13 seconds.  They are more expensive but...still yes that sounds worth-it.  They are also, as a bonus, better against storms, and better against air (shuttles and corsairs, and notably observer pickoffs).  But the tradeoff is that they are a lot worse against dragoons, so you can't actually go pure goliath.  I don't think it has that big of an impact on the matchup cause they aren't that different from hydras, but less larva efficient and need more tech, hydras might even still be the bulk of the army, but yeah, you'd see some goliaths.

---

Goliath: 18

Goliaths below hydras and marines is a surprise.  But...specifically being a factory unit (and thus a robo unit for protoss and a lair unit for zerg) I think is a big limiting factor.  And some matchups the antiair capability just isn't that important, and yeah, their ground-to-ground capabilities aren't bad but aren't special.

metroid composite

  • m_ACac
  • Administrator
  • Denizen
  • *
  • Posts: 4381
    • View Profile
Re: Theorycrafting! (Because I like competitive metagames too much)
« Reply #978 on: October 13, 2023, 11:50:33 PM »
Science Vessel

TvT: 0/4.  I can remember these being used once in a game from like 2006 when nukes were being used TvT.  a patrolling science vessel was used for detection.  Offhand I don't remember them being used in any recent TvTs, though I might be misremembering and they were built for defensive matrix.  Something to research later I guess.

TvP: 2/4.  These get made sometimes for EMP.  Specifically to stop arbiters, but it's also just a usable AoE spell and good if it lands on high templar.

TvZ: 4/4.  Matchup defining unit.

ZvZ: 2/4.  I'm going to rate these a bit below some other anti-air options.  Just...how do you protect them from scourge.  in TvZ you have marines, the science vessels can retreat to marines.  In ZvZ...mutas can fend off scourge, but they do so by retreating with moving shot, and the vessel can't retreat as quickly.  You could lag behind the mutas with your vessel, but if you get flanked by scourge that's a hard punish.  The vessel also stands still for a long time when casting--a time to hit with scourge.  It's also more tech--probably not quite hive tech, but science facility would require a spire, for example, and irradiate requires research.  But...still...better than making a queen with ensnare I think.

ZvP: 2/4.  These would be limited by "does the protoss have corsairs left?"  Cause corsairs kill a science vessel faster than they kill a muta.  Nevertheless, irradiating high templar would probably come up some of the time.

ZvT: 2/4.  Sticking an irradiate on a marine at the center of a marine ball probably means 8 dead marines or so.  But...zerg science vessels would become biological and thus die to enemy irradiates without being able to irradiate the terran vessels back.  That said, I think you can still get good value hiding your zerg science vessels a long way away from terran science vessels and just using defence matrix.  Defence matrix soaks up most (but not all) damage from an irradiate.  Like...over the course of an irradiate, the irradiated unit loses 30 HP (instead of 250).  It's not a flawless counter cause you're using a 100 energy spell to not fully counter a 75 energy spell, but it's ok.  Additionally...EMP is a thought.  If you EMP all the enemy science vessels, that's gotta be good right?  I think we'd see these often enough.

PvP: 1/4.  Maybe this is worth considering over high templar.  EMP pretty good.  Defence matrix on a reaver is pretty solid.  It's not really clear if a vessel would like...die ever being a fast air unit that doesn't need to get close.  But...I mean, 1/4 seems about right--probably would be a very rare tech choice, rarer than high templar.  Same score I gave guardians.

PvT: 2/4.  This basically comes down to do I think there's shenanigans with defence matrix.  And...maybe?  I don't think D-matrix on like Zealots is all that exciting, although you might do it cause it's repeated value you can get without putting your vessel at any risk.  But d-matrix on a reaver sounds really spicy.  d-matrix on high templar sounds pretty spicy.  D-matrix on a shuttle could be fairly spicy.  And D-matrix lasts for 56 seconds, you can set all of this up in advance without needing a lot of extra APM.  I think we would see vessels in games that went fairly long--just repeat value that terran can't do a whole lot about.

PvZ: 3/4.  Can protoss protect science vessels from scourge?  Yep, it's called corsairs, or even just cannons.  Is there a potential allin that vessels wouldn't be great against?  Yep, hydra allin.  And hydras in general you would rather make high templar than science vessels until you are stable (science vessels might be better after you are stable cause they can retreat and stay alive, but high templar are much better if your army is about to get wiped out).  Vessels do kind-of invalidate any higher tech than hydras, but I think zerg's ability to cheese, and ability to stay alive in a midgame with just hydras is good enough that I don't really think vessels are an auto-loss.

---

Science Vessel: 18

The surprise, I guess, is that they are not ahead of dark archons by all that much.  But loosely being the third best spellcaster after high templar and defiler is the right general position for them.

metroid composite

  • m_ACac
  • Administrator
  • Denizen
  • *
  • Posts: 4381
    • View Profile
Re: Theorycrafting! (Because I like competitive metagames too much)
« Reply #979 on: October 14, 2023, 12:33:19 AM »
SCV

OK, so gotta think about what would be repairable.  (Other than obviously other workers).  For Protoss I suppose it would be all robo and stargate units.  For zerg, all lair and hive units.  Hmm...would burrowed lurkers be repairable?  Looks like you can't repair mines, so I'm going to go with "no" on that.  Obviously it would be nice to repair burrowed irradiated units, to avoid taking splash damage from the irradiate.  And then all buildings are repairable too, of course.

TvT: 2/4.  Used for scouting and repairing.  Wouldn't be the end of the world if it couldn't be used.

TvP: 3/4.  Used for scouting and repairing.  The really scary question is what do you do about dragoons hitting your bunker, and like...I think you just don't build a bunker, people used to play that way so I don't think it's an auto-loss, but wow it would suck to not have repairing.

TvZ: 3/4.  Scouting still important in this matchup.  Repairing pretty important in the earlygame.  Like...if the zerg tries to bust with zerglings, you repair the wall.  If mutas damage but don't kill a turret, you repair the turret.  I think playing without repair would be very rough actually.  Plus there's some proxy rax bunker rush with SCV pull builds that would be invalidated if you can't use SCVs on offence.

ZvZ: 3/4.  So...lair units are repairable you say?  Yeah, guess what's a lair unit: Mutalisk.  A muta that ate a scourge hit could get repaired to full.  Also, buildings are repairable?  So...repairing sunkens and spores is a thing?  Nifty.  I don't know that the matchup is unwinnable for the zerg without repair, but wow this is a big deal.

ZvT: 3/4.  Retreating damaged mutas and repairing them is just nasty in this matchup.  So is repairing sunkens when terran tries to bust sunkens.  I think terran can potentially stabilize.  In a lategame with science vessels and siege tanks, yeah, repair not such a big deal.  But the midgame would be rough.

ZvP: 2/4.  Hydras would not be repairable.  Lurkers would be, though you'd have to unburrow it.  But zerg kind of does that already--if during a lurker contain a lurker gets stormed, they'll pull that lurker back, and rotate in a high HP lurker.  And yeah, they could repair the lurker in that scenario.  Repairing mutas also probably is not bad.  I don't think repair here is as nice as against terran, where you might just have one very low health muta--usually corsairs spread the damage fairly evenly, and it would take a lot of time to repair everything.  But still, having the option makes muta allins a bit better, and that's nice.

PvP: 1/4.  Dragoons would not be repairable.  Still, you probably would occasionally repair shuttles and reavers.

PvT: 2/4.  Repairing carriers and occasionally repairing reavers, shuttles or buildings is the use case here.  I think the one that impresses me the most, that makes me upgrade this from a 1 to a 2 is repairing cannons while a vulture raiding party is trying to take out a base.  Obviously vultures could manually target repairing SCVs, and sometimes you would do that, but that takes attention, and sometimes vultures do target the cannon just to guarantee the base kill.  Also...I think in this matchup past the earlygame scouting phase you do actually care about having 60 HP workers, just cause it takes 3 hits for a vulture to kill.  You'd keep some probes for building of course.

PvZ: 3/4.  Being able to repair cannons or even just an upgrading forge against a hydra bust would be a game changer.  Being able to repair corsairs would be reasonably nice.  SCV pull against hydras just as a fighting unit would be a lot better than a probe pull.  Repairing observers would occasionally be a thing.  Just...pretty much all of zerg's openers suffer against repair.

---

SCV: 22/36

OK, this is genuinely surprising.  Just...this all being inspired by an artosis video where he puts probes in S-rank.  But man, repair is like...fairly critical for Terran functioning, and really good for other races.

metroid composite

  • m_ACac
  • Administrator
  • Denizen
  • *
  • Posts: 4381
    • View Profile
Re: Theorycrafting! (Because I like competitive metagames too much)
« Reply #980 on: October 14, 2023, 12:43:01 AM »
Valkyrie

TvT: 2/4.  Yeah, you see them occasionally.  Probably slightly less than half of all TvTs, usually fairly late.

TvZ: 2/4.  They have a strong attack timing against muta builds, and are used as a response to muta scourge allins, but skipping them is common too.

TvP: 0/4.

PvP: 0/4.  Worse corsairs which are already not built.

PvZ: 0/4.  Worse corsairs.

PvT: 0/4.  Even if they do go full cheese and make mass wraiths, they're worse corsairs.

ZvZ: 4/4.  There's some vulnerability to scourge, but not too bad to escort these with mutas, and hard to imagine they don't just break the matchup.

ZvP: 2/4.  Yeah, I think they would get made as part of "Ogre Zerg Gamer" where you allin with mutas.  Obviously not a ton of them, but they are more cost efficient than mutas against corsairs (roughly cost even) if you had them in a control group with mutas they would just naturally spread away from the mutas.

ZvT: 1/4.  If you plague a clump of science vessels, you could later make a valkyrie.  It's...a lot worse than the corsair in this role--slower, more expensive when it inevitably dies to irradiate.  But...wider splash.  What did I give corsair?  2?  OK, this gets a 1 then.

---

Valkyrie: 9/36

metroid composite

  • m_ACac
  • Administrator
  • Denizen
  • *
  • Posts: 4381
    • View Profile
Re: Theorycrafting! (Because I like competitive metagames too much)
« Reply #981 on: October 14, 2023, 01:15:37 AM »
Wraith

And this is the final unit.

TvT: 3/4.  They are a fairly core unit in this matchup.  The one thing I will note is that eventually people tend to switch off to them.  Switch to dropships or battlecruisers.

TvZ: 1/4.  2 port wraith openers are apparently becoming a somewhat popular opener, although haven't been seen a ton in ASL other than from Leta.  This feels like a rating that could easily go up in the future.

TvP: 2/4.  Wraiths spent a bit of time being mildly in-vogue against protoss.  They've dropped off again, but still totally reasonable to make a wraith against reaver drops.

ZvZ: 3/4.  I don't think they're going to change a losing position into a winning one, but wraiths are kinda nice in this matchup.  Pretty much cost neutral fighting with mutas, but they have more range, so could take potshots at enemy mutas.  Cloak could also matter.  And they murder overlords--they kill stray overlords much better than mutas.  OK that's the good news, the bad news is that you would need something like 2 larva to make a wraith (2.2 is my calculation).  And any wraiths you make obviously means you don't deal as much ground damage.  Still...I think they would get built and I think they would be solid.

ZvT: 0/4.  I just don't see the use case for these.  Mutas way better against turrets, goliaths, marines, SCVs.  Mutas better against plagued scinece vessels.  Best I can think of is either defending dropships, or if you know the terran is going valkyrie allin without a science vessel, maybe you make cloak and try to snipe the valkyrie faster than the terran can scan.  I don't really believe in either of these scenarios.

ZvP: 0/4.  They get erased so hard by corsairs.

PvP: 1/4.  So corsairs are never made in this matchup even when reaver drops are common.  I'll give wraiths the benefit of the doubt, though, they do kill shuttles a bit faster, and can also slowly kill the reaver after it drops out of the shuttle.

PvT: 2/4.  So wraiths would be good in the same scenario scouts were good--to stop a push that doesn't have goliaths yet while you get starports for carriers.  Except, you know, wraiths are a lot better than scouts--come out earlier, much better vs ground, potentially cloak.

PvZ: 2/4.  Yeah, wraiths would be a legit cheese.  Macro zerg is often prepared for corsairs, gathering their overlord near a spore, but not prepared for an early air unit like a scout hitting drones, and wraiths would come out earlier and be better at drone killing.  Also, in terms of sniping overlords, wraiths kill individual overlords better than a corsair.  And the scary scenario where zerg is hydra busting...a wraith is still bad to build, but better than a corsair.  I think you usually open with a wraith, do the scouting, then mass up corsairs if needed.

---

Wraith: 14/36

And that's the last unit so...

Zergling: 29/36
Siege Tank: 29/36
Mutalisk: 29/36
Zealot: 29/36
Vulture: 28/36
Reaver: 27/36
High Templar: 24/36
Defiler: 23/36
Scourge: 23/36
Dragoon: 22/36
SCV: 22/36
Corsair: 21/36
Overlord 20/36
Marine/Medic: 20/36
Hydralisk 19/36
Science Vessel: 18/36
Goliath: 18/36
Shuttle: 17/36
Lurker 16/36
Dark Archon: 16/36
Arbiter: 16/36
Probe: 16/36
Wraith: 14/36
Dark Templar: 13/36
Queen: 13/36
Drone: 12/36
Dropship: 11/36
Observer: 10/36
Valkyrie: 9/36
Firebat: 8/36
Guardian: 7/36
Carrier: 7/36
Ultralisk: 6/36
Battlecruiser: 5/36
Archon: 5/36
Devourer: 4/36
Scout: 2/36
Ghost: 1/36

metroid composite

  • m_ACac
  • Administrator
  • Denizen
  • *
  • Posts: 4381
    • View Profile
Re: Theorycrafting! (Because I like competitive metagames too much)
« Reply #982 on: October 14, 2023, 01:24:58 AM »
One additional thought about marine/medic: optical flare.

I think specifically ZvP this could be good in a lurker contain--zerg already tends to regularly make scourge to kill observers, and optical flare is a 75 energy spell that basically kills observers (makes them sight range 1 with no detection).  And it's 9 range and there's no risk of scourge being picked off.  One medic is more expensive than one scourge, but can potentially pick off a lot of observers in a contain situation.

Does this increase medic marine in the ZvP matchup?  Mmm...maybe.  They were 2/4 before due to just early marines having some cheese potential--scarier hydra busts.  Maybe this makes them 3/4 cause they continue to have utility in longer games?  I dunno though--this requires an academy and optical flare research, requires building medics with no plan of building units they can heal, and you would need to blind quite a few observers before all of that became more cost-efficient than just plain not getting any of that tech and getting scourge instead.  2/4 probably still fine.

metroid composite

  • m_ACac
  • Administrator
  • Denizen
  • *
  • Posts: 4381
    • View Profile
Re: Theorycrafting! (Because I like competitive metagames too much)
« Reply #983 on: October 20, 2023, 06:20:51 AM »
Starcraft

ASL Finals

https://www.youtube.com/watch?v=KwcdXblqYlc

Nearly every game going to lategame PvZ, which means defilers, reavers, archons.  Soulkey favouring ling lurker over mass hydra.

metroid composite

  • m_ACac
  • Administrator
  • Denizen
  • *
  • Posts: 4381
    • View Profile
Re: Theorycrafting! (Because I like competitive metagames too much)
« Reply #984 on: October 29, 2023, 05:43:47 AM »
Starjeweled

So...I've been doing calculations on Starjeweled units again.  I think I've even posted them in this topic before.  There's a nice table of stats I've added to a wiki I can link people to if anyone is reading this.

But for stuff that doesn't really feel like it belongs on a wiki, I've been messing around with some DPS calculations, and I...have some random thoughts.

The first thought is that HP, and specifically actual literal HP and not "HP relative to cost" is generally a much better measure of what units are good than DPS figures.  Like...muta damage is higher than Banshee damage, but Banshees having 125 HP instead of 100 HP make them usually better, as they survive the 100 damage storm spell.  But the case that really stands out here is zealots, which, on paper, deal more damage to ultralisks than any other unit for their cost, and yet generally aren't even considered cost efficient against ultralisks cause the AoE cleave damage from ultras is too good against them.

One interesting result, however, is that several units are (relative to their cost) fairly similar at damaging cannons.  Zealot, Hydra, Banshee, Immortal, and Ultra are all about equivalent at hitting cannons.  Tank and Roach deal about half the damage of those two, and in practice Muta is also in the same ballpark as roach (their attacks bounce, but they'll never hit two cannons with the glaive bounces).  And then Ghosts and Colossus are very bad at this role (like 13%-20% of the group that is good at hitting cannons).

Another interesting thing to do, since there are often marines on the field soaking up hits but not really doing anything, is look at the time it takes each unit to kill a marine.  The surprising thing to me is that Colossus, if they don't cleave, are actually the second worst at this relative to their cost.  Yeah, they have a huge damage bonus against marines--they deal a whopping 80 damage to them.  But that doesn't matter when marines have 30 HP.  And their slow attack speed really sinks them here.  (Ultras take two hits to kill a marine, but attack almost three times as fast, so while obviously it's quite bad when an ultra gets distracted and attacks a single marine, it's actually not as much of a time waster as when a colossus does it).  Though obviously there is some value to all of the AoE options against marines (Ultras, Colossus) just a surprising result that Ultras are arguably better at marine clearing when their damage bonus is not marines.  For non-cleave units, the ones that take the least time to kill marines relative to their cost by far are Hydras and Zealots, though worth noting Mutas assuming good glaive bounces are also roughly in this group.  The cleave units generally need to hit 5-7 marines to get equal efficiency.  Other low cost units (Ghost/Roach/Banshee) are about half as fast at clearing out marines, although worth noting ghosts can and do snipe marines, so this can make them briefly faster.  Siege units (the ones with bonus damage to buildings) are quite slow at killing marines (siege tank in particular is quite miserable, slower at killing marines relative to its cost than a Colossus that never manages to cleave two marines).

metroid composite

  • m_ACac
  • Administrator
  • Denizen
  • *
  • Posts: 4381
    • View Profile
Re: Theorycrafting! (Because I like competitive metagames too much)
« Reply #985 on: November 01, 2023, 09:32:06 PM »
Thinking about FFT ban-a-thons again.

First ban probably still Calculator.  Although...I mean, I think there is some argument that "well if you're doing a solo challenge and minimizing grinding that maybe Auto-Potion comes online faster, and maybe it's hard to get Calculator JP solo if you have special classes banned, like if Ramza can't Yell+Accumulate?"

And like...ok, let's analyze this a bit.

How hard is the game for a solo challenge without Ramza ubersquire stuff with Auto Potion but without Mathskill?  Honestly, I don't think it's too bad?  I did a solo challenge once that had all three heavily restricted, and fights with a guest were fairly easy, but fights without a guest...the fact that every archer shot, every geomancy hit, every Ninja using Throw was aiming at my one low level summoner made surviving kinda hard.  Auto Potion probably can fill in that gap just fine, soak some hits just like a guest does when they're in the fight.

The downside I guess, is that Auto-potion can fail.  Every once in a while Auto Potion will fail two or three times in a row, and then you'll get a reset and need to try the level again.  This downside can be averted with brave raising.  However...brave raising is probably more grinding than just learning calculator skills.  So if the argument is "well you have to do brave raising of course" then I think Chemist is already in a losing position here.

Let me just sanity check that claim.  Being able to CT5 in Math Skill takes 400 JP (you start with about 150, so that's like...15 actions or so to get to 400.  15 actions of Praise gains you...somewhere between 8 and 9 brave depending on your MA.  But CT5 doesn't actually save you until like...chapter 3, using math skill early on you actually need to target.  If you need all the math skills, it's 2200 JP.  That is like...82 actions.  With a slow character.  Getting brave to 97 from 74 takes like...about 40 actions.  So actually no: raising brave to 97 is arguably less effort than mastering Calculator in terms of raw actions.  You do need to spread it out over several fights, but like...in every fight where there's a competent guest, you could drop a couple big summons, get the fight under control, and then spam praise on yourself while you let the guest clean things up.  Then you don't even need to do your grinding outside of story fights.

I guess there is an argument for mathskill, and that's that it can take an extra layer of challenge on top of the solo challenge.  Like if you add some kind of stat penalty on top of being a solo challenge, maybe require Ramza to step on a level down trap at level 1 20 times, or maybe require you to be in the mediator class for every story fight, or maybe gameshark away the ability for characters to gain EXP so that all player characters stay at level 1--yeah, Mathskill still going to cruise through that challenge, whereas a summon user with auto-potion...not so much.  Yeah, it's a bit more grinding, but Mathskill is just so much more powerful than anything else when you get it.

And...sure, I think I'm sold by that argument.

1. Calculator

metroid composite

  • m_ACac
  • Administrator
  • Denizen
  • *
  • Posts: 4381
    • View Profile
Re: Theorycrafting! (Because I like competitive metagames too much)
« Reply #986 on: November 01, 2023, 09:49:52 PM »
FFT ban-a-thon, second ban

OK, so like class to ban after Calculator...

It's probably still Chemist, but...I do at least want to think about it some.

If we ban summoner instead, what does it look like?  Well, we would have time mages using black magic with Bolt/Bolt 2, until they build up to short charge Meteor.  And they'd have auto-potion, and item in general so like Ether and Hi-Ether for more uses of Meteor.

Yeah...you'd suffer a bit in Chapter 2 due to using Bolt 2 instead of Ramuh, but like...whatever, I think you're in good shape.

And specifically, only Calculator and Summoner banned, I think you could probably still do that as a solo challenge relatively comfortably.  But Calculator+Chemist ban...I think that's a non-trivial solo challenge, at least without substantial grinding.


2. Chemist

So for third ban...

We're sort-of in the territory where you could do...not a solo challenge, but a 1.5 character challenge.  You do a playthrough with a newbie, tell them they'll control one character, and you'll control one character, and let them experiment with any class they want.  This sort-of solves the durability issue that arises when you try to do a solo summoner build playthrough with low grinding.  You'll have a second HP bag, they'll take some hits, and you're good to go.

For a 1.5 character playthrough where most of the cheese is banned...you need to pack the damage.  And specifically you need AoE damage.  I've seen many times on two character playthroughs (both in FFT and LFT) that if you don't have AoE...you just end up getting swarmed.

So...the options for that are Draw Out from Samurai, Summon, or Short Charge Meteor.

And the pick's gotta be Summon right?  Like...arguably the best of the three anyway.  And it's 200 JP instead of 2300 JP (Short Charge + Meteor).

3. Summon

metroid composite

  • m_ACac
  • Administrator
  • Denizen
  • *
  • Posts: 4381
    • View Profile
Re: Theorycrafting! (Because I like competitive metagames too much)
« Reply #987 on: November 01, 2023, 11:36:29 PM »
FFT ban-a-thon 4th ban

So...ok, this is definitely one I've been wondering about.  Assuming we're still going into this with a small party, 2 or 3 people, is there a world where the right ban for the 4th ban is not Wizard, and is instead like Time Mage or even Samurai?

And...I think it depends how we want to loosen restrictions.  Like...yeah, if you're still trying to do this with 2 characters, or 1.5 characters, and you are just dealing with the small party size by grinding more, absolutely I think there is an argument for Time Mage ban next.

If, on the other hand, you are dealing with the greater restriction on classes by being stricter about low grinding, and looser on the number of characters, maybe 3 characters with a limitation like "don't even do random encounters, just story fights" I think the answer is probably Wizard.

Ultimately I think the latter style of restriction is the one I've done and seen more, and getting up to a full party or closer to a full party gets things closer to how the game was intended to be played, which matters for stuff like party buffs that are unnaturally less impressive than they should be in a 2 player party, so yeah, that's probably still the way to go.

4. Wizard

OK, so fifth ban....

In the past I've put Squire here for Gained JP Up, as every good build that isn't banned requires JP.

More recently, for the most recent list, I decided that just about every build even physically focused builds were going to set either Time Magic (for haste) or White Magic (for raise) for a long time, and that the first really powerful units you get would not come from JP at all, but from equipment as you steal Agrias and Gafgarion's armour and slap them on a Lancer who gets overtuned spears in chapter 2.  Making one of Time Mage or Priest the next ban.

But what about Lancer?  Is the ban here ever Lancer?

And that's...an interesting question.  Like...what's better in early through mid Chapter 2, a Knight with haste cast on them, a Geomancer with Haste cast on them, or a Lancer?  I think it's Lancer right?

Which means banning Lancer would make the game harder for...part of Chapter 2, at least until you get Ninjas unlocked or learn short charge+meteor or whatever you're going for.  But...simultaneously harder in chapter 1 without whichever of Haste or Raise you're getting rid of, and harder in the latter 60% of the game, when you still want haste and raise, and might also have Meteor or Holy or Teleport or MP Switch too.  Jump is a fine skillset of course, but not going to outshine other lategame skills.

And I mean, honestly, Rainbow Staff and one Wizard Robe show up at Zirekile falls, so like...while being a Time Mage in chapter 1 is going to be a bit rough compared to being in Knight, it's fine in Chapter 2.  And Geomancy starts hitting hard for a little while thanks to Wizard Robe.  And guns for Mediator show up.  These aren't as good as "Lancer in early Chapter 2", but they're...fine.  I think the chapter 2 options are actually looking a bit more robust than the chapter 1 options.

Which means...probably again no changes from the last list, unless I want to switch up Time Mage and Priest here.  Do I?  Ehh...don't think so.  Debates could be had about whether banning Raise hurts more than banning Haste, it probably hurts a little more, sure, but Time Mage has other exciting stuff like Teleport, MP Switch, Short Charge Meteor.

5. Time Mage

And yeah, I mean, at this point probably your whole party slaps on white magic for raise while heading to...IDK, whatever jobs they're heading to.  And this helps in Chapter 1, which still looks rougher than Chapter 2, so it's not Lancer.

6. Priest

OK, 7th position.  Last time I gave this to Squire due to Gained JP Up.  But should it actually go to Lancer, for being this big bailout in Chapter 2 thanks to nutty equips that pull the party out of the dark age of earlygame?  Mmm...I kind of do buy that argument, actually.  What else are you headed towards?  Ninja where...once you get there you aren't going to care about Gained JP Up?  Oracle where gained JP Up is admittedly pretty nice?  Samurai where Gained JP up is mandatory?  Monk where...if gained JP Up is banned you actually don't mind slapping on Equip Armour instead?  Dancer where...like Ninja once you unlock the class you don't mind a lack of Gained JP Up?  Geomancer where...you'd live without Gained JP Up soon enough cause you'd learn Attack Up?

I think the path here is like...two characters go to lancer and carry the party, and the other characters go to...something like Ninja, and take advantage of the fact that the Lancers are carrying the party to gain JP in weak classes like Thief and Archer.  And I think this sounds a bit stronger than trying to send all five characters down "weak now, strong later" pathways, even if those pathways pay off a bit sooner thanks to Gained JP Up.

7. Lancer

So...ok, is it Squire now, or is there a remaining build that would hurt more than hitting Gained JP Up.  It would have to be something that is strong early on, strong without gaining a ton of JP, so like...we're talking Knight and Geomancer basically.  Maybe Oracle deserves a look too--maybe a lot of these physical builds are setting Yin Yang as a ranged option; even one use of Paralyze isn't a joke.  And in Chapter 2 with Wizard Robes and Sticks showing up, of course Oracles start hitting like a truck.  I don't think it can be Ninja--while Gained JP Up isn't used much after unlocking Ninja, pushing back the unlock time of Ninja by 50% hurts that build a lot.

I don't know, none of these are really jumping out to me.

Like...if Oracle is banned, but you have Gained JP Up, ok, whatever big deal, just ignore the magic side of the job tree, have a couple Geomancers hold down the fort while Ninja is speedily unlocked with Gained JP Up.

If Knight is banned...ok so this hurts a bit for about 6 fights, then you unlock Geomancer part way through Chapter 1 and you no longer care.

If Geomancer is banned...I mean, you just use some Knights when you need a bit of help in chapter 1, accept the fact that Knight JP is not great, and then in Chapter 2 you have Oracles with sticks and those can carry you until you unlock stuff like Ninja and Dancer.

If Squire is banned...just every plan is worse cause it doesn't have Gained JP Up, and also doesn't have Move+1 which will also hurt early on for basically all of these builds.

8. Squire

So I mean, with Squire banned, we're looking at low JP investments to make our lives not painful as early as possible, that also hopefully have some long-term relevance, which basically means Oracle, Geomancer, Monk.

Oracle comes online earlier than both.  Yin Yang even with one Yin Yang magic known is probably going to be the secondary skill of choice for a lot of characters given all the other bans.  A lot of remaining classes one of their key payoffs is damage for boss fights, but like...you could also just get Life Drain.  High damage from other classes is mostly required for non boss fights, where maybe one party member dies, you don't have revival and you need to kill all remaining enemies...but lots of different remaining classes are good in that scenario.  Also even one focused Oracle in the party could get Defence Up from spillover for the whole party, and that's pretty good too.

Monk...I do think a significant selling point of Monk is that you can have them in Knight for all of chapter 1, get Equip Armor, move them over to Monk, and then have them learn Monk stuff.  This build path is probably the best of the three in chapter 1, but then just...falls behind for a while cause you're grinding out 500 Knight JP, and then you start grinding out the Monk skills which are all kind-of high JP and you've got a 3 move melee attacker who has more HP than an oracle but hits less hard for most of Chapter 2.  Or I mean, you could equip battle boots instead of power wrist and be 4 move, at the expense of dealing...quite a bit less damage because you'd have 7 PA.  (50 with punches, 40 with wave fists, 30 with earth slash not that you'll have JP for earth slash).  I guess you do that until you learn wave fist, or if you haven't hit 7 base PA yet, but once you do you probably want 3 move, 8 PA for 60 damage wave fist.  They are also, of course, the only remaining revival, even if it's bad and costs a ton of JP.

Geomancer...I think the issue with Geomancer is that it's going fewer places than the other two.  Maybe it has a period of time when it brings the most power as a unit.  Probably in late chapter 1, maybe Chapter 2 too, but it's not much better than Oracle in Chapter 2 if it is.  Maybe lots of builds will want attack up long term--Monk likes it if they don't multiclass out of Monk, Ninjas or Geomancers would certainly use it.  Gunning mediators would use it.  But Geomancy is generally outclassed by Yin Yang if you just want to give your character some relevant range.  Yin Yang is available earlier and it's also kinda just better.

Yeah, probably still Oracle here.

9. Oracle

So...so far the only change seems to be slipping Squire down below Lancer.  Looking back in the topic...yeah, I didn't analyze that one too deeply.  I think the next few bans probably go the same way (next probably Geomancer, for example--remaining good Wizard Robe class for Chapter 2.  Remaining place to take Draw Out if you go that route).  So copying the old list that looks something like...

1. Calculator
2. Chemist
3. Summoner
4. Wizard
5. Time Mage
6. Priest
7. Lancer
8. Squire
9. Oracle
10. Geomancer
11. Knight
12. Ninja
13. Archer
14. Monk
15. Mediator
16. Samurai
17. Dancer
18. Thief
19. Bard
20. Mime

I might want to run back through my notes for some of these--like looking at my current analysis I do see the logic train of Geomancer and Knight above Ninja--they just help more in Chapter 1 and 2, but I'll probably re-read my notes on Archer above Monk cause I'm not quite remembering how I came to that conclusion.
« Last Edit: November 01, 2023, 11:52:31 PM by metroid composite »

metroid composite

  • m_ACac
  • Administrator
  • Denizen
  • *
  • Posts: 4381
    • View Profile
Re: Theorycrafting! (Because I like competitive metagames too much)
« Reply #988 on: November 02, 2023, 03:08:05 AM »
So...ok, let's just list out what is left

Monk, Dancer, Bard, Archer, Mediator, Samurai, Thief, Mime

I am certainly thinking about charge guns at this point.  And maybe Monk finally, they still have revival, and now some of the best blitz damage.

Equip Gun Archer with punch art for revive and stigma magic sounds reasonable.  Just building a monk also sounds reasonable, maybe ending up in Archer with Martial Arts secondary, or in Monk with Concentrate.  Arrow Guard and HP Restore both good reactions.

I think it's going to for sure be one of Monk or Archer.  Just the two classes that aren't pathetic in chapter 1, while playing different roles.  (Mediator is pathetic in Chapter 1).

So like...I intuitively feel like this really ought to be Monk.  They're just generically seen as a good class, and Archers are generically seen as a bad class.

The thing that's making me hesitate on just slamming Monk is that like...I'm kind of feeling like Archers probably bring a bit more to Chapter 1, a bit better at handling Sand Rat Cellar, and Monks are really suffering from gear woes in Chapter 2.  But surely the long term must be better for Monk?  And like...yeah I think it is, but Mediator SCC kinda stomps most of the game pretty good once it gets guns, and add charge and shields and arrow guard, and you can slap equip gun on people while they unlock Bard and Dancer and Samurai for whatever you want out of those classes, and man, that really doesn't sound all that bad.  Archer just seems like it flows better into the remaining classes.

I guess Archer is also now the one remaining way to get a shield.  Which matters for like Balk fights, and not too much else.  White robes are also decent for those fights (but those are now only available on Mediator and Samurai, and if you ban Archer, then Mediator drops in value a notable amount).

Let's see...I think Archers are better in Chapter 1 (though not by a lot, I just value the 4-5 range), worse for Zaland/Barius Hill in Chapter 2 (monks only need to get to level 5 for 7 PA, and Power Wrist shows up after Zirekile Falls, bringing that up to 8 PA for 60 damage wave fist, whereas Archers are still looking at 5 WP here, so like 30 damage--longer range and higher HP of course), worse for the next four fights that matter in Chapter 2 cause Romanda Guns are storebought. Chapter 3 I think Archers are ahead until Bracer (even with power sleeve, monk's hitting like 10 PA, so like 90 damage wave fist.  Yeah, give me Mythril Gun over that--64 damage base, and charge+3 brings that to like 88.  With 8 range instead of 3 and no chance to miss).

And then bracers surely put Monks ahead for the last four fights of chapter 3, and then earth clothes as well in chapter 4 for earth healing, and also somewhere around part way through Chapter 3 Monks start having revival for emergencies.

But in terms of how much of the game each class is better, this is a surprisingly even split.

The numbers gap is going to be quite wide in Chapter 4 and late Chapter 3, so that's one argument to break the tie towards Monk.  But Archer routes will still have the range edge and the option of shields and the ignore evade edge (against earth immune enemies only, since earth slash is ITE, but there's lots of earth immune enemies), and the edge in terms of portability (Equip Gun+Charge unlocking bard/dancer/samurai much more comfortably than martial arts+punch art).

I...feel like I'm leaning Archer here?

13. Archer

OK, this is what I wrote out last time for Archer over monk.

Quick note on a typo:

"worse for the next four fights that matter in Chapter 2 cause Romanda Guns are storebought."

I think this should be "better for the next four fights that matter."

Anyway, do I buy all of this logic--do I buy that a 15 damage bowgun is better than a 36 damage punch?  Eh, I can kind of see how I came up with that, like you don't need to get hit by 3 move knights in Sand Rat Cellar, but also...that's over double the damage, and almost certainly preferable for Dorter.

I'm curious what the HP difference is at Sand Rat Cellar actually.

Looks like it should be pretty similar.  Monks have about 10 more HP than Archers at level 1, and Feather Hats provide 16 HP.  So maybe a 6 HP difference in favour of Archers, but also if you're level 2 or 3 that would help Monks more than Archers.  (Loosely Monks gain about 1.5 HP more per level than archers).

Additionally, though, while obviously archers would rather have a longbow, the fact that they're stuck with a 3 WP crossbow does allow them to use a shield, the best available shield being the...escutcheon (10% evasion).  Eh.  I mean, you still wear it, but it's the equivalent of like...5 HP, honestly.

I think I'm going to disagree with my previous self and say that Monk is better before the first longbow comes out.  Certainly better for Dorter where you might need to rush down some wizards.  Maybe worse for Sand Rat Cellar but not by much.

Although...hold on, I should consider charge.  By the time you can have monk unlocked at all, you can also have Charge+3.  (200 Archer JP vs 200 Knight JP).  Charge+3 with a 3 WP crossbow brings the damage up to 24.  That's...a lot less bad.  Now, granted, if you use charge, you are not moving.  You can't use charge on a knight, and also back up and have them not hit you back.  The move command is disabled after using charge.  But...even when kiting as a strategy is not an option, some range is still better than no range.  Maybe you can't avoid getting hit by one Knight, but you don't get hit by all the Knights.  Maybe you wouldn't have an attack, but the range gives you an attack.

That probably does slide archer into still being at least a bit better for basically all of Chapter 1 yes.

Another question I have is...should we maybe be banning Mediator instead?  Even if I think Archer is a little bit better in Chapter 1 (which I do) how much worse is Mediator without Archer?  They can still grab Equip Gun, and then go play around in any class they want, probably Samurai and some Dancer.

So...if Mediator is the ban...what happens?

Well, for one thing, Monks in Chapter 1 would get an easy secondary.  They can probably, without even spending time in Archer, get Charge+2, and then punch for something like 48 (?) instead of 36.

Let's see...looking up the BMG, charge interacts with unarmed punches like this:

Bare Hands: [((PA + K) * Br) / 100] * PA

So...that's interesting, the K modifier is on the brave PA and not the martial arts PA.  Means if you have 6 PA, and 72 brave, Charge+1 does something for you, and Charge+2 does nothing.

Charge+1 with 6 PA should get you 36 -> 45 damage.  Charge+3 with 6 PA should get you 36 -> 54 damage.  OK, interesting.  Weird but interesting.

So...that's a pretty big boost in Chapter 1.  And Archer continues being fairly valuable to Monk later on (probably more vaulable than a Mediator would be to a monk build).  If the Monk needs shields or hats, Archer has those with 110 PA.  Concentrate's maybe somewhat interesting to a monk.  Arrow Guard is maybe somewhat interesting to a monk.

I think you do end up pretty close to a Monk SCC, but like...with Charge, and Concentrate, and Move+2, and a way to get shields and hats if you want them, and that should be very smooth sailing later on (like mid Chapter 3 onwards), while also being substantially better in Chapter 1.

So what's the selling point for banning Mediator over one of the other two here?  I guess it's the period from late Chapter 2 when you have guns, to early Chapter 3 before Power Sleeve and Bracer show up but when you still get Mythril Guns.  And like...maybe also ultra late Chapter 4 your gun users will have gone and gotten something cool out of Samurai, but any of these should be smooth sailing that deep into the run.

OK, yeah, I don't think banning Mediator makes much sense.  Archer and Monk work decent together.

So I think it's ban Archer or ban Monk.

Going over that again...

Archer probably has the edge in Chapter 1.

Monk probably has the edge before guns (early Chapter 2) although might depend a little on the fight.  Zaland Fort City probably a bit more Archer favoured cause you can jump on the wall with longbows, but Barius Hill probably Monk favoured.  So not a strong lean.

Archer should have the edge up until bracers (second half of Chapter 2, much of Chapter 3).

And then I've been assuming Monks pull ahead after that, but I mean, let me actually run some numbers.  A level 18 Monk with Power Sleeve and Bracer has a 133 damage wave fist.  (Will be 168 damage at level 21).  Eh...is 133 really blowing away gun damage?  Mythril Gun with Charge+3 is like 88 damage, but also ignores evade, and is 8 range instead of 3.  Obviously you will hit level 21 for 168 damage, but that'll be like...second half of Chapter 4 maybe?

Also worth noting, the gun user can use Angel Rings or Sprint Shoes instead of Bracer.  Now like...don't get me wrong, the Monk definitely wants a Bracer.  90 damage wave fists...no, don't lose 40% of your damage to equip angel rings.  But it is a nice benefit the gun user gets.

I guess there is another question of just how much is Archer really bringing to the gun user.  What if Archer gets banned, and we just end up going for a gun using build anyway?  How much is actually lost compared to what Monk offers?

How good is Charge for a gunner really?  Well...so I've generally been assuming Charge+3, but up to Charge+5 will happen (104 damage).  But the flip side is that if you are not using charge, you can attack and wait on spot, and you'll get an extra 20 CT from that that you don't get from charge.  So like...you can think of a character not using charge but waiting on spot as dishing out 80 damage per 100 CT instead of 64.  But...104 is still 30% more damage than 80.  88 (Charge+3) is still 10% more damage than 80.  You can't always stay in-sync with enemies if you wait on spot, which means enemies can land big charges or spells or jumps on you.  Also...80 damage per 100 CT sounds nice on paper, but it is more back loaded where turn 1 you are still dealing 64 damage.  And that still comes at the cost of never moving.  I think saying charge is worth 30%-40% more damage is still probably about right, and that's...substantial.

And then also, Archer is a pretty nice class for a gunner to be in.  You get the charge skillset by default, and can set another secondary.  You get shields, hats, etc.

What does Monk offer to a gunner party, by contrast?  Some revival (0 vert, can miss, probably don't learn till chapter 3).  Some status curing (still 0 vert, but AoE and basically can't miss unless you're like worst compatibility--130% hit rate).  Healing which will be available a lot earlier than someone who spends time in Archer then gets 750 mediator JP for equip gun, and then finally goes back to unlock Samurai.  Like...probably talking Chapter 2 vs Chapter 4 on when those healing skills are available.  Some reactions like HP restore, but Arrow Guard on gunners is probably also in the same ballpark of good.

Still...overall no: I think gunners prefer Archer over Monk here.  Would I rather deal 30%-40% more damage, or would I rather have some safety options like a bad revival for longer fights?  I would rather have more damage, just end the fight before people crystallize rather than use 0 vert low hitrate revival.

So...yes, I do think the points when guns are good are also going to be points when you'd rather have Archer in the class list.

Also as maybe an interesting option if you're in chapter 4 and up against a zodiac that really needs to die fast, Gastrifitis Concentrate archer with Charge+3 is like...120 damage at level 18.  Or, I guess, if you equip Twist Headband and Bracer too, then it's 170 damage.  (140 without Charge+3).  To be clear, I think you almost never equip Gastrifitis over a Mythril Gun.  But like...if there's concerns that maybe Archers can't really blitz down...I don't know, Adramelk or something because 168 damage wave fists are so much more damage...ehh...no there are options for shorter range higher damage.  I'm not sure if there's actually a fight where I'd take Gastrifitis over Mythril Gun, though--Adramelk I think I'd rather just spread out.

Anyway, Archer over Monk and Mediator does seem correct.

---

Is Monk still 14th?  Yes, Monk is still 14th.  While Monk and Archer are kinda close in value in Chapter 1, Thief and Mediator are a lot worse in Chapter 1.  So once Archer is banned, Monk is kind of the only thing holding the Chapter 1 and early Chapter 2 together.  Monk basically doubles party effectiveness for Chapter 1 and half of Chapter 2.  And after that...when guns show up...losing Monk still hurts for the rest of the game.  Like...even if you focus on guns after that, you probably get HP Restore from monk, and probably have one or two people pick up Revive.  Also, it's less clear that focusing on guns is definitely the plan--the damage gap widens without charge.  133 wave fist to 88 Charge+3 gun where the 88 ignores evade and is much better range doesn't sound too bad.  133 to 64 on the other hand...now we're talking double the damage.
« Last Edit: November 02, 2023, 03:12:55 AM by metroid composite »

metroid composite

  • m_ACac
  • Administrator
  • Denizen
  • *
  • Posts: 4381
    • View Profile
Re: Theorycrafting! (Because I like competitive metagames too much)
« Reply #989 on: November 02, 2023, 11:10:24 AM »
Some other errant thoughts on the low-ish ranks:

Am I too easily writing off Wiznaibus?

So...on the one hand, Wiznaibus in some fights is clearly better than guns, right?  Without a bracer you can deal about 15 damage, with a bracer you can deal about 20 damage, and it'll tick roughly twice per turn in late chapter 3/early chapter 4, which means 30 damage per turn without a bracer, 40 damage per turn with a bracer.  Full all enemy AoE.  There are a number of fights where this is probably better than 64 damage gunshots from Mythril guns.

But...not assassination missions, which are often the harder fights.  Especially not zodiac demons with summons, as they will often be caught charging, and take 96 damage.  And the fights where Wiznaibus could be argued as better than guns, where the goal is kill all enemies, Nameless Dance tends to trump wiznaibus.

And of course, unlocking dancer is a problem, makes for a character that is weak for a long time...unless you use Equip Gun.

One additional note on Nameless Dance--in a weird way I think having access to it as a strong option devaules Monk revive.  The kind of fights that tend to stretch on long enough for revival to be needed are often also the same fights that get flattened by Dance, and specifically Nameless Dance.  Except maybe Altima.

---

OK, next question, how good is Slow Dance on Altima?  Obviously form 2 here.  Assuming a full 4 dancers (The max).

After 8 clockticks she gets her turn, and then slow dance goes off, lowering her from 12 to 10.

After 8 more clockticks slow dance goes off again, she's at about 80 CT, lowering her from 10 to 8.  Then she gets her second turn.

After 8 more clockticks, slow dance goes off again, she'll be at about 40 CT.  Lowering her from 8 to 6. 

After 8 more clockticks, she'll be at about 90 CT.  Slow dance goes off again, lowering her from 6 to 4 CT.  She'll get her third turn

And yeah, she should hit 1 speed before getting a fourth turn, at which point she is easy persuade bait.

So she gets about three turns if no one is disabled this whole time.  How does this compare to I dunno, just shooting her with Mythril Guns over four turns?  If 5 characters take 4 shots with Mythril Guns, that's...1280 damage.  She has over 3000 HP.  So...yes, actually the speed breaking plan is actually fairly reasonable.

---

OK, next question...what about other mediator shenanigans.  With invite or whatever.

Like, I've been dunking on Mediators in Chapter 1, but what if you Invite some monster that doesn't suck.  Is that like...a thing that actually exists without going way higher level than you need to?  Well...maybe not actually...a level 1 red chocobo has like...30 HP, which is low, and a choco meteor that deals 20 damage, which is...fine, but not that exciting.  Choco Ball from a black Chocobo deals 32 damage, but they'll have less HP than a red, and 5 speed.  Bear in mind, for mediator to move up, they'd have to move ahead of like...Monk probably, and so you would want a monster that really carries a level 1 party harder than a monk to the point that you don't mind that you're gaining no JP and levelling a unit that you won't use when guns come out.  I'm...not impressed by Chocobo numbers anyhow.

The other thing I've considered is...I mean, one thing I outlined for Mediators as a possible workaround if grinding is needed is like...inviting a porky, and then poaching chantages in chapter 3.  And I think that's reasonable--Chantage is the common poach, so I mean, you end up resetting a couple times in Chapter 2 searching for the uribo, and then you do a couple randoms in chapter 3 to poach some porkies.

But there is another option, and it's to just go get yourself some elemental guns.  From Germinas peak or whatever.  Might even be faster, IDK.

And there is yet another option that is interesting, and that's to get yourself some Tiamats.  Notably here, triple bracelet will take off half of a Zodiac's health.  So that's an option if you are worried about zodiacs.

Does any of this move Mediator above Monk...?  Ehh...probably not.  Monks are probably still in the region where like...you really don't need to grind you can just play a Monk SCC with extras.  But it probably does cement Mediator over other grind heavy classes like Samurai.  Mediator with a bit of grind can break the game harder than Samurai.

---

Is there a way I am underestimating Bard?  Mmm...if I am it would only be as a Monk spoiler.  "Monk can get healing way earlier than someone who goes Equip Gun first then unlocks Samurai."  "Oh yeah? well what if they take Equip Gun and unlock Bard for Life Song?  What now huh?"  But honestly...this barely moves the needle I think.

---

Is there a world where Dancer should be above Samurai?  If you grind JP hard, unlock Samurai, get all the relevant draw outs, and two hands, and blade grasp, and go grab Move+2 from Thief, I do think it's correct to say you'll get more out of Samurai than Dancer, and specifically, Blade Grasp handles a lot of mook fights, and Samurai brings more damage to the table for big boss fights.

But...that's also very JP expensive.  Dancer, you unlock it, you spend 100 JP on Nameless Dance, and maybe you stick around for 300 JP of stuff, and there you go.

Now, I mean, worth noting, Dancer's unlock requirements are harsher (about 400 JP harsher in fact) and you need to be female, which makes you worse in these physical classes.

So...ok, let's look at single target damage, to say, Velius.  Kiyomori is the equip available here, 12 WP, Samurai should have 8 PA by this point, and probably want sprint shoes.  That's...60 damage.  With Two Hands, 120.  Alternatively, you can draw out for...you will have 4 MA which you can raise to 6 with a Wizard Robe, and then Heaven's Cloud if you know it is 6*14 = 84 damage, but doesn't deal 1.5x damage to a charging Velius.

What does dancer carpet damage look like?  5 PA, 6 MA, 10 WP, leaving the accessory free it's like...70 damage.  Although worth noting to match samurai speed you could go green beret and then bracer, and that would be 80 damage.

OK, what about a male thief punching?  6 PA base.  Higher speed, so can actually justify not wearing sprint shoes probably.  So...slap on all the PA gear, 13 PA.  117 damage.

Um...ok, Thieves can, in practice cause Samurais need to wear sprint shoes, match Samurai two hand damage?  Bro...why the heck would I unlock Samurai which takes ages, and then spend 900 JP getting two hands, when I can get these numbers out of a Thief with zero JP investment?

OK, time out, there is still one remaining argument for Samurai, and that's that it is the last robe wearing class.  This matters for some Balk fights, although Dancer is pretty good against both of those.  But more importantly this matters for the Wiegraf duel because Samurai is the last class that can wear Chameleon Robes.  Some SCCs need to level to near level 50 to beat Wiegraf, and like specifically Thief and Mime jump to mind here.  I think there does remain a non-Samurai out, though, and that out is Bard who can handle the Wiegraf fight by running away and buffing.  Not my preferred strategy but whatever, it gets the job done.

Alright, yeah, I think I'm sold.  It's a bit weird, cause I'm pretty sure Dancer is considered to have the harder SCC than Samurai, and Thieves are a bottom tier SCC, and Bards are generally also considered quite awful as an SCC, but I guess they plug each other's holes here.

16. Dancer

Is 17 Samurai?  Pretty sure yes.  Take whatever you would do with other classes.  Add Blade Grasp and Kiyomori.  It's better now.  I think you have to assume a decent amount of grinding--like if you ban Samurai, the other classes just need a lot of levels anyway, so getting Samurai JP is reasonable.

17. Samurai

And like...ok, previously I put Thief above Bard because "Which SCC is harder?  Nobody remembers?  Uhh...what now?  IDK, Thief unlock sooner I guess?"  But...Bards don't need nearly as many levels to get past Wiegraf, and that is something tangible.

Something else tangible is that unlocking Mimes isn't like...completely off the table, and Bards both have an easier time unlocking Mimes (they can sing in any class and it gets JP fairly quick) and bards also benefit more from unlocking them (Mimes, like Thieves, punch medium hard--complements Bards better than Thieves.  Mimed harps are not bad, because Mimes manage to have better stats than Bards.  And Miming Sing is a known okayish combo).  I think I will move Bard up one as well.

18. Bard


This makes the updated ban sequence...

1. Calculator
2. Chemist
3. Summoner
4. Wizard
5. Time Mage
6. Priest
7. Lancer
8. Squire
9. Oracle
10. Geomancer
11. Knight
12. Ninja
13. Archer
14. Monk
15. Mediator
16. Dancer
17. Samurai
18. Bard
19. Thief
20. Mime

metroid composite

  • m_ACac
  • Administrator
  • Denizen
  • *
  • Posts: 4381
    • View Profile
Re: Theorycrafting! (Because I like competitive metagames too much)
« Reply #990 on: November 02, 2023, 08:36:29 PM »
Starcraft:

Artosis casting a game in which ghosts are used for nukes against Zerg by pro level players.

https://www.youtube.com/watch?v=1cwp3bOC5pQ

So the basic idea is this--Zerg needs to go "crazy zerg", where they skip lurkers get carapace upgrades and go straight to ultras, getting a high number of sunken colonies to survive until a critical mass of Ultras with carapace can be made.  Then, with 7-8 sunken colonies at each base, Terran proceeds to drop a nuke on one set of sunkens (instantly killing most of them) and then hopefully from there just winning the game.

I guess this does raise Ghost up a little bit in my ranking of them, to 2/36.

Trying to think if there's any other matchup where this much static defence might be made, though, and...like...maybe protoss defending against a hydra rush might get that many cannons, but by the time Zerg could tech into nukes protoss would have high templar so wouldn't need cannons anymore for defence.  In fact, might even kill their own cannons.

Zergling: 29/36
Siege Tank: 29/36
Mutalisk: 29/36
Zealot: 29/36
Vulture: 28/36
Reaver: 27/36
High Templar: 24/36
Defiler: 23/36
Scourge: 23/36
Dragoon: 22/36
SCV: 22/36
Corsair: 21/36
Overlord 20/36
Marine/Medic: 20/36
Hydralisk 19/36
Science Vessel: 18/36
Goliath: 18/36
Shuttle: 17/36
Lurker 16/36
Dark Archon: 16/36
Arbiter: 16/36
Probe: 16/36
Wraith: 14/36
Dark Templar: 13/36
Queen: 13/36
Drone: 12/36
Dropship: 11/36
Observer: 10/36
Valkyrie: 9/36
Firebat: 8/36
Guardian: 7/36
Carrier: 7/36
Ultralisk: 6/36
Battlecruiser: 5/36
Archon: 5/36
Devourer: 4/36
Scout: 2/36
Ghost: 2/36

metroid composite

  • m_ACac
  • Administrator
  • Denizen
  • *
  • Posts: 4381
    • View Profile
Re: Theorycrafting! (Because I like competitive metagames too much)
« Reply #991 on: November 03, 2023, 07:55:27 AM »
FFT

Thinking about just putting some overtuned percentages on a few things I think in vanilla FFT are overtuned.

I'll try to restrict this to one variable, so e.g. while LFT might nerf stuff along multiple dimensions (like JP cost and MP cost and ctr and effect) I will try to focus on if just the effect was tuned down, with the same JP cost and everything else, where would it be?

---

Bolt 1/Fire 1/Ice 1

Overtuned by 40%

So...I said recently that I think these could go down to 10 mult in LFT.

I think there actually is some logic behind 10 as well.  While comparing spell multipliers to weapon multipliers is typically hard to do, fire/ice/bolt feel built for Chapter 1, and so comparing to say, 4 WP bows kind of makes sense to me.  Typically faith goes something like your faith is 70, enemy faith can be anywhere from 45-74 which averages 59.5.  0.7*0.595 = 0.4165.  So 10*faith here would typically be about 4.

Now, I mean, to be clear, Wizards in chapter 1 would still deal more damage with a bolt 1 than archers would shooting an arrow--they get their gear (thunder rods) early, and have sky high MA.  But like...a Priest with Black Magic secondary is no longer going to deal more damage to one target with bolt than an archer shot in Chapter 1.  (Would be about the same with a 10 mult).

And like...I don't think fire/ice/bolt would be unusable with these values--still more mana efficient, still very JP cheap, still good at picking off a low health enemy.

---

Auto Potion

Overtuned by 400%

I considered just limiting it to Hi Potions, but the problem is that only makes it balanced relative to other endgame skills, it doesn't actually make it balanced for the 400 JP cost (as seen in LFT, Auto Potion with Hi Potions is still a very solid lategame reaction--and it probably would be better in vanilla cause damage is lower in general).

I considered an intermediate amount, like a 50 HP potion.  But that would still make it stand out compared to other earlygame options like weapon guard, counter, absorb used MP, Caution.  Ultimately, I think this was actually probably correctly balanced as an earlygame ability around 30 HP potions, and people just found an exploit.

Math Skilled CT5Holy

Overtuned by 525%

Is an infinite ranged skill that can hit everything just inherently busted?  No, dance with Wiznaibus shows that it is not, the damage just needs to be low enough.

So...ok, in LFT we let people use a 12 mult evadeable move with mathskill, and that is generally considered fineish.  How much do we need to reduce the damage to justify a non-evadeable move?  Well...if a typical enemy is wearing a feather mantle, they'll have 30% magic evade.  12*0.7 = 8.4.  Sure, let's try out a multiplier of 8.

I'm going to assume you get to 12 MA, use MAU, you strengthen with 108 gems, you wear a chameleon robe.  You can reach 12 MA this way with a level 12 Wizard, so you know, not too hard to hit.  This makes the damage 67, and the healing 80 if your party is all 70 faith.  Yeah, I think that's probably in the range of "fine".  It also heals the party, but like...I think the benefit of healing the whole party is largely counteracted by the fact that the whole party needs to wear chameleon robes.  Also, I mean, 70 AoE party healing, that's...not bad, but wouldn't actually be a jaw dropping action on its own.

Now, I mean, there is an interesting mathskill trick where if you use CT5PrayFaith first, and you hit everyone on the field, then you deal 160 with the same setup.  But I mean, maybe that's an issue with CT5PrayFaith IDK.

Anyway, Holy's multiplier is 50.  So 50/8 = 6.25

MathSkilledRevival

overtuned by 840%

OK, so, there's an argument that Phoenix Down in FFT is mildly overtuned.  We certainly nerfed it in a few ways in LFT, and it's still excellent.  But IDK, maybe this is Bolt1 level of overtuned so 40% overtuned, not sure how that would translate into in-game changes.

Phoenix Down that doesn't require throw item to have 4 range is more overtuned.  Again, hard to translate this into numbers, but maybe it lets you have Magic Attack Up or Attack Up, so a 30% buff to your character, sure let's call that 30%.

This is cumulative, so like...we're now looking at like...1.3*1.4 = 1.8, so like 80% overtuned.

A instant version of Raise is more overtuned than that.  Adding like a 100 HP heal onto an already good action...that's better.  But then you do need to worry about faith and internal party compatibility or you might miss, that's a downside.  I think net this is still better, but it's minor, like a 10% boost.

Now 1.1*1.8 is around 2, so like 100% overtuned.

Instant speed raise 2 is better still of course.  Some downside in hitrate, but not much and can still quite easily get to 100% if you are already looking for good internal party compatibility or a high faith party.  Feels like a bit of a bigger buff.  Call this a 20% boost.

Now 2*1.2 = 2.4.  like 140% overtuned.

So...if going from 1 range to 4 range is 30%, how much is going from 4 range to infinite range?  Hmm...I dunno, let's just give it another 30%.

Now 210% overtuned.

And then there is the AoE aspect.  I think AoE res kind of just inherently breaks the game in FFT.  Realistically I think in a lot of fights you're hitting 2 targets, but I have certainly hit three, and I'm going to score this as if it's hitting 3.  So...200% boost.

Now 840% overtuned.

If you don't start with the assumption that Phoenix Down is a little overtuned, divide 9.4/1.4 mathskill revival would instead end up like 570% overtuned.

Shiva/Ramuh/Ifrit

Overtuned by 70%

OK, so we can look at what LFT does, but LFT does a lot of things (raises MP cost, raises JP cost, lowers damage, raises ctr).

If all we change is the damage multiplier...what do these need?

Now, I mean, one comparison is to bolt 3, but in fairness bolt 3 is kinda bad.  Like...I think Bolt 3 could have the AoE size of Ramuh, and nobody would be like "stop, that's overpowered, you can't do that."

So...first issue with bolt 3 is that it's evadeable.  If we assume 30% magic evade again (feather mantle, or more minor mantle and a shield) 24*0.7 = 16.8.

But also, Shiva/Ramuh/Ifrit are 4 CTR and Bolt 3 is 7 CTR.  You can actually make these line up just fine by using Short Charge with Bolt 3, and magic attack up with Ramuh.  That said...I don't want to count this CTR gap as a full 30% damage gap.  You can still use Magic Attack Up with Bolt 3, 7 ctr is useable.  Call this 20%.

So...we end up with a damage multiplier of 14, or an overtuned amount of about 70%.

Teleport

Overtuned by 50%

So Teleport's about the same JP cost as Move+2.  Let's just accept Move+2 as correctly tuned--I realize it was nerfed quite a bit in LFT but that was to try to get people to care about movement abilities like Jump+3 and Move on Lava and Move HP Up.  Those are just a lost cause in vanilla so whatever.  And...in Vanilla FFT, Move+2 is...fine, I learn it occasionally, I don't ultra prioritize it, probably around balanced.

Anyway, Teleport is about the same cost as Move+2, and in terms of what it does it's probably about the equivalent payoff of Move+3, on a sort of medium difficulty setting.  Which would make it about 50% overtuned.

OK, I should explain what I mean by a "medium difficulty setting" weighing the value of Teleport can be a little strange, cause on extreme challenges like SSCCs, Teleport is like Move+5 levels of value.  But people who plan to make the perfect unbeatable lategame party with a 100% winrate, I've seen value it below Move+2--if they eliminate all risks, then they never make a risky teleport.  "Medium difficulty setting" just means you've got a real chance to lose, so a reason to gamble with teleport's fail chance, but also your chance to lose isn't -that- high, you still expect to win more often than you lose, you're not going to reset 10 times to get the perfect teleport sequence.

Under such circumstances, Teleport's raw movement ability if you have some risk tolerance is maybe Move+2ish maybe Move+1.5ish, but then it also goes through walls, goes through enemies, ignores height, ignores speed reduction terrain like deep water.  Basically all the benefits of Fly, but also with added movement tacked on (Fly makes your movement very high quality, but it's spoiled by just not moving very many panels; Teleport fixes this).

I guess I should justify "maybe Move+2ish maybe Move+1.5ish".  So like...one of the classic calculations, which I think is wrong, is people just imagine a long racetrack, 1000 tile map, and figure out that a 3 move character who wanted to get to the end of that track as fast as possible should teleport 6 or 7 panels (for a 70% or 60% chance every turn) and move 4.2 panels per turn on average.  And then they declare the net value of risky teleports to be Move+1.2 (cause 4.2 is 1.2 higher than 3).  Here's why I don't fully buy into that number--FFT, if you've got a crisis where you need to get into position or lose--like need to revive an ally before they hit 0, or need to kill a key enemy.  If you've got to get there in one turn, or lose, and it's a random distance away, Teleport is more like Move+4.5.  If you've got to get there in two turns, Teleport is like Move+3.  If you've got to get there in three turns, Teleport is like Move+1.7 (realistically a fair bit higher than that if you've got a huge map, but I capped the number of panels you need to travel at 15 cause maps aren't usually that big).  Teleport is not great at long-distance running, but it's pretty good at sprinting to a fixed destination, which is probably a measure that more closely matches in-game scenarios.

But the flip side of that is while Teleport is quite good at going to a destination, it's not great at fleeing from a destination.  If there's a summon locked on, and you want to not stand in the summon, you should basically always use a 100% teleport, and then you're getting Move+0 out of Teleport.  So that's worth factoring in too.

Another interesting application to Teleport is when other random elements of FFT mix in.  Like if there's an enemy knight, and you want to attack the knight from the back to avoid hitting the shield, very often it's worth it to teleport additional panels, cause the added Teleport fail chance is less than the shield block rate.  Like...a 100% teleport to hit them from the front or an 80% teleport to hit them from the back, you take the 80% teleport.  Or, let's say there's a Knight that a Move+2 character can just barely reach, and have a 60% chance to hit through a crystal shield.  The teleport character could do an 80% teleport to get to the same spot...or they could do a 60% teleport to get to the back.  Ending up with the same overall odds to hit as the Move+2 character.  Or, potentially, better--if the Move+2 character is one square away from reaching the back, so they're stuck with a 60% chance to hit, meaning the teleport character could reach the side with a 90% teleport, then they can hit the back with a 70% teleport, actually ending up with a higher effective chance of hitting.

FFT also often has dead zones.  Like...not moving might be fine, you stay with your party and out of range of most enemies.  Teleporting onto a roof to kill an archer might be excellent.  But standing in the middle might be death.  Classic case for a long range teleport attempt.

Anyway, all of that has me kneejerk risky teleports as closer to Move+2 than Move+1.2 in value.  (And the value of Fly is Move+1ish or a bit better especially with the added movement).  Which makes teleport roughly Move+3ish in value.

Mathskilled deadly status

Overtuned by 400%

So I was looking through Calculator SSCC logs recently, and being a solo challenge with calculator stats, often damage was just not enough to shut down a fight.  But status often was.  Three that stood out here--Frog and Petrify, which have the same hit rate (120).  And also Sleep, which has a monster hit rate (170).  And sometimes Paralyze (185 hitrate) because of Thief Hats.

So I think the comparison to make here is probably Nameless Dance.  Nameless Dance has a 50% chance to hit, and 50% chance of being a goodish status (frog, sleep, confusion, stop).  But it's also 10 CTR and not instant, and that's a big deal.  Also, Calculator can wait on spot for CT, but Nameless Dance being 10 ctr often just makes that not even an option.  Also, like...in a non-challenge setting you can just go on the fastest class, like an H-Bag Ninja, get hasted and really spam these statuses--not an option for Nameless Dance.  Also, Nameless dance is like...pretty solid already.

I think probably a 1/8 chance per target of debilitating status is probably fair-ish here.  Half that of Nameless Dance, but you can also spam this about twice as often as namless dance on a hasted character that waits in place.

Now, ok, that's what the hitrate should be.  What's the actual hitrate?  Mmm...like 76% for sleep?  53% if they have feather mantles.  Like 55% for Frog or Petrify, 38% against feather mantles.  Worth noting Frog is like this giga highroll on Nameless Dance, so being able to pick it (or the sometimes better Petrify) is pretty eyebrow raising.

Anyway, 1/8 chance is 12.5%.  I guess the Sleep hitrate is probably the one to look at, probably more comparable to non-highroll outcome of nameless dance.  Averaging the mantle and non-mantle hitrates, that's like 64%.  Which is about 5x higher than 12.5%.  So...400% overtuned.
« Last Edit: November 03, 2023, 08:06:20 AM by metroid composite »

metroid composite

  • m_ACac
  • Administrator
  • Denizen
  • *
  • Posts: 4381
    • View Profile
Re: Theorycrafting! (Because I like competitive metagames too much)
« Reply #992 on: November 03, 2023, 09:29:23 AM »
FFT.

So...thinking about just where Mathskill is on the overtuned scale...and specifically thinking that looking at individual aspects of it one at a time might be underselling how good the versatility is.

If there was an "all magic" skillset that just had all the calculable spells but you cast them with mana and cast time, and all the spells became singletarget, it would be quite good, but maybe not overpowered.  I mean, outright better than white magic right cause all relevant white magic is calculable.  But it would be missing a couple of the greatest hits, like Life Drain and Meteor, so like there's situations when I'd set Yin Yang or set a big AoE spell (Meteor or a summon) over an "All Magic" skillset.  Or when I'd set something else to get the AoE back (like Haste).  But still...certainly an exciting option.

What if just...all of it costs no mana?

OK, now it's a bit nutty right?  Like...specifically no mana Holy spam even with charge time, even singletarget, sounds fairly solid.  No mana ranged revival--if you're just looking for a revival secondary, now this is probably the best one.  Like...sometimes currently on a Summoner or on a Meteor focused Time Mage I might set item over white magic cause it costs no MP.  But I'd definitely set this All Magic over both other revival options.  And no-mana Holy when you're out of MP would be very strong...like probably on the level of Draw Out strong.  A bit more damage, but singletarget and still has cast times.  I don't know that I would necessarily rate it above Summon, but the no MP cost makes it highly synergistic.  Already about as good as vanilla summon.

What if it also all had no charge times on top of no cast time?

The revival is now way, way too good, like outright uncomfortably better than item.  Instant speed no MP holy I feel fairly confident in calling better than Draw Out, even as a singletarget skill.  All the status moves like Sleep become way better cause they can interrupt charging enemies.  Yeah, probably runaway best skillset in the game already.

Still two more things to add, and possibly the two most important (infinite range, full battlefield AoE).

Um...yeah, it is hard to put into numbers just how far past the balance line this skillset is as a cohesive whole.

metroid composite

  • m_ACac
  • Administrator
  • Denizen
  • *
  • Posts: 4381
    • View Profile
Re: Theorycrafting! (Because I like competitive metagames too much)
« Reply #993 on: November 03, 2023, 05:51:06 PM »
FFT

So actually, I started wondering since a lot of mathskill overtuned numbers were about 6x as good as they should be, if Q values were just divided by 6 when cast through math skill, would that be alright?  It's a question worth thinking about, cause something so formulaic as "divide by 6" is something that could probably be dropped right into a balance hack.

Holy Q value divided by 6 gets a value of 8.  If my above numbers hold this is fine-ish.

Raise Q value divided by 6 gets 30.  MA would raise that to 40.  Faith would translate that into about a 20% hitrate.  Yeah...not remotely concerned about that.  Low likelyhood revival in FFT, even if it's AoE and instant, is probably not a balance concern.

Sleep, likewise, not overly concerned.

So here's the one sketchy one.  Frog/Petrify, Q value gets reduced to 20.  MA would raise that to 30.  Faith would translate that to about a 12% hitrate.  If there's a feather mantle that lowers to about 8% hitrate.  12.5% hitrate was my rough target for "this would be fine if landing status roughly equivalent to Nameless Dance status".  But always landing frog or petrify as your status is obviously quite a bit better.  To be clear, this is not necessarily the fault of the divide by 6 concept, more a fault of the formula which takes the value 20 and adds MA.

Still...I mean, maybe that's ok?  As demonstrated by some of the above thought exercises, one of the valuable places to put a toned down version of this skillset is on an MP using caster, who plans to drop some summons or meteors first, and then fall back on a no-MP no charge time skillset.  A 12% chance to Frog or Petrify is high power only if used right from turn 1 when all the enemies are alive.  But that also happens to be when you would like to be casting spells like summons and meteors.  It's not going to be great on turn 3 when you've already dropped two meteors and are now out of MP but there's three enemies left.

And obviously there's no real concerns about stuff like Pray Faith with a 17% hitrate.

So...maybe dividing all the Q values by 6 when cast through mathskill actually does work out to a balanced-ish skillset.

metroid composite

  • m_ACac
  • Administrator
  • Denizen
  • *
  • Posts: 4381
    • View Profile
Re: Theorycrafting! (Because I like competitive metagames too much)
« Reply #994 on: November 03, 2023, 08:57:42 PM »
Starcraft Brood War

So...here's a game where Queens and Guardians were used lategame vs Protoss, and Protoss made dark archons to feedback the queens:

https://www.youtube.com/watch?v=1G0Z2BgtCgQ

Dark Holy Elf

  • DL
  • Denizen
  • *
  • Posts: 8161
  • Well-behaved women seldom make history
    • View Profile
Re: Theorycrafting! (Because I like competitive metagames too much)
« Reply #995 on: November 11, 2023, 04:07:54 PM »
Just posting to say I am enjoying the FFT overtunedness analysis.

I'd say Phoenix Down is overtuned, though not sure how to quantify how much. You could probably use Revive as a comparative baseline though there's an argument that Revive is undertuned of course.

Erwin Schrödinger will kill you like a cat in a box.
Maybe.

metroid composite

  • m_ACac
  • Administrator
  • Denizen
  • *
  • Posts: 4381
    • View Profile
Re: Theorycrafting! (Because I like competitive metagames too much)
« Reply #996 on: November 11, 2023, 06:09:06 PM »
Hmmm, I feel like revive would probably be acceptably-tuned as a stand-alone ability if it was like...90 JP instead of 500 JP.

Like...yeah: if that was the case (and also Phoenix Down wasn't still so obviously better than it) you might set punch art as a secondary with just revive every once in a while, but not like...to the exclusion of using other skillsets.  If you don't have martial arts set mind you, revive might end up being a 75% hitrate (and zero vert, of course) or even as low as 72% on a mage.  So...that's pretty inaccurate.  And there's still the issue of 0 vert.

But...still, it's a degree of insurance for longer fights.  It doesn't cost MP, so like a build that burns MP very fast like a summoner or a short charge meteor spamming time mage might prefer to set punch art over white magic for raise.

I do think even small reductions in hitrate represent a substantial power level downgrade for revival, of course.  (Part of why mathskill revival, despite being on paper the most overtuned aspect of mathskill, is kinda just no longer a concern power-level wise if you multiply the Q value by 1/6).

SnowFire

  • DL
  • Denizen
  • *
  • Posts: 4964
    • View Profile
Re: Theorycrafting! (Because I like competitive metagames too much)
« Reply #997 on: November 12, 2023, 04:58:52 AM »
The topic of Fire Emblem drafts came up in Discord awhile back, and specifically FE3H.  One issue with drafts is that they really mostly make sense "competitively", i.e., if you're trying to measure against something.  You get weird results if one person is drafting their faves and someone else is playing to win.  Furthermore, for FE Three Houses in particular, the game is friggin' long.  So it's a little awkward - you can do a short draft, but then it's probably a Normal mode warp-a-thon.  But if you do Hard or Maddening and ban Warp, you have something that takes so long that you fully expect some Did Not Finishes mixed in from real life / boredom / etc.

Anyway.  The REAL fun is the drafting part, IMO.  Playing it out is just a long formality.  So…  why not draft against myself as a sample?  There were two funny YouTube videos from someone who is clearly far more of a FE6 fanatic than most people on the planet where he drafted against himself and also actually ran the game 4 times:


https://www.youtube.com/watch?v=-SIO6xqhPNA
https://www.youtube.com/watch?v=jhwWMSdV9sE 

So…  why not draft 3H against myself?  Except skipping the actually playing part.  Maybe.

There's a lot of 3H draft rulesets out there you can find (and also a lot of seemingly abandoned drafts).  In general, I'm assuming some sort of soft LTC is the criteria (rather than real-time), combined with…  well, not quite Ironman, but an expectation that you are not doing dopey LTC things like resetting until you get that turn 1 crit.  If you're Divine Pulsing due to a mistake, whatever, but DP'ing as a core strategy for crit / key dodge farming is no bueno. 

Lords/Warp/Stride all make the game go faster and thus be more about feeding XP to boss-slayers, since everyone else isn't getting tons of combat.  As such, I'm gonna restrict this as much as possible to make it closer to normal play rather than specialized LTC stratz.  I think the draft ends up the most interesting on Silver Snow with Warp / Stride banned.  Silver Snow means no Lord hard-carry (without having to arbitrarily ban Lords) which makes a lot of the draft pointless since you'd be building a carry Lord no matter what (although you still have Byleth in SS, of course).  If Warp isn't banned, you end up with a massive rush on Linhardt / Lysithea and a consolation prize of Manuela which is weird.  Stride is most defensible, but you can still get pretty far with flyers and a Dancer even without it.  There's still some wackiness here - Vantage / 100 Crit / Retribution strategies start looking pretty good, most notably - but I think this is reasonably close to "normal" play.

One weird rule I saw in two drafts posted on serenes which doesn't appear to be very common, but I liked, is "Paralogues are free up to 10 turns / One auxiliary/quest battle per month is free up to 10 turns" (see https://forums.serenesforest.net/index.php?/topic/96042-fe16-azure-moon-draft-every-last-one-of-them/ ).  Basically it means that characters' Paralogues are actually relevant rather than just skipping almost all of them (maybe you still do the one-roundable ones like Ignatz / Raphael).  I'd also assume that for White Clouds paralogues where you've drafted half of the pair but have access to the Paralogue, you only get the benefit of the part you drafted (e.g. Caduceus vs. Spear of Assal if you only drafted one of Seteth / Flayn and do An Ocean View).  I might be a bit harsher on this than 10 turns…  maybe something like 6 turns are free, turns 7-17 are at half-cost (i.e. a 10-turn clear costs 4/2 = 2 turns), and any turns after that cost full price.  But whatever, point is, doing the Paralogues can now make sense (especially for Maddening).

Finally, one last point.  Silver Snow Reunion at Dawn is dumb, especially on Maddening.  I don't like having to obligate people to have a Byleth build that can solo-survive.  While one solution is simply to have a 3 person draft rather than a 4 person draft, I went with the Black Eagles all having two entries, meaning two separate players can draft 'em.  Should make the Eagles a little more viable and has the nice aspect of making the early game less rocky too.

Per notes on other Silver Snow drafts, Edelgard / Hubert are free, Prologue is free, C1 is "pick one character to be free", and C13 Seteth is free.

Anyway, my self-draft, along with reasoning.  I'm creatively calling our drafters Player 1, 2, 3, and 4, or P1/P2/P3/P4 for short, with P1 drafting first.

--
P1: Petra
P2: Petra
P3: Dorothea
P4: Dorothea, Ferdinand
P3: Ferdinand

First things first: making Reunion at Dawn reasonable.  If you don't draft for this early then it's your own fault IMO.  Plus, getting some Eagles will speed up C2 when non-Wolves recruitment isn't really possible yet.  The left-side joinees in SS Reunion at Dawn are Petra, Dorothea, and Caspar.  Petra is the obvious power pick (in a turn count draft, you're probably making her a Wyvern, act shocked), with Dorothea your next best option.  This will ensure you at least have 3 capable units at the start in Byleth / Seteth / First Pick.

The next option for Players 4 & 3 in snake order is the secondary pick.  Now, for Frue LTC strats, I know that Bernie gets some hype, because you can do some goofy Pass + Pegasus Knight + getting her low + huge Vengeance hits.  But, per above, I'm trying to go for more reliable strats.  Ferdinand is an Eagle who'll be around for C1/C2 and isn't that much worse a carry than Petra who'll make a nice dodgetanky Wyvern who can still KO the boss.  I think he's safer than a Bernie pick.

--
P2: Constance
P1: Yuri, Hapi
P2: Lorenz

The Wolves join really fast (can speed up C2) and have some amazing loot in their Paralogues, which per above are doable without tanking the turn count (although.. Yuri & Constance's doesn't mess around!).  Constance gets you Nuvelle Fliers and, importantly for P2 (which doesn't have Dorothea), the possibility of late-game Bolting sniping of bosses.  With Warp / Stride banned, "double dance Dorothea / Constance to smack a boss twice with long-range magic" becomes one of the main ways to speed some maps up.

P1 can take Yuri & Hapi.  Yuri offers the mighty Fetters of Dromi, which is good for all the usual reasons, plus.. probably ending up a Sniper in an LTC scenario, honestly.  Not worth trying to slog through skill ranks for Mortal Savant or something.  Hapi's just one of the strongest remaining mages who can offer Physic support and do goofy monster games, who also offers an incredibly good battalion in her Paralogue in Timotheos Magi Corps.

Having taken a powerful but frail mage in Constance, P2 can snipe Lorenz.  Not for him, of course, but extra range on boss Bolting snipes or just in general from Thrysus is a thing.

--
P3: Ingrid
P4: Shamir, Sylvain
P3: Leonie
P2: Felix
P1: Flayn, Balthus

P3..  definitely getting debatable at this point, but I'm a fan of Ingrid.  One, she has good out-of-house growths in enemy Pegasus Knight and a reasonable training scheme, so you can grab her later without needing to level her.  Two, you get her excellent Galatea battalion from her Paralogue now.

P4 - well one of the drafts banned sniping the DK in C6 from outside his room, which is a dominating thing a Shamir pick can do.  The "psuedo-Ironman" ruleset suggests that this isn't reliable, but she's still solid anyway as a Sniper ready to go with little effort.  Sylvain is another strong pickup - can either be in C2 for fast training, or in C6 for Cavalier growths.  Can switch over to a mage-y build if the rest of the draft requires it, too.

For P3, Leonie's a still outstanding power pick on just good-stats grounds, and also can come with enemy Cavalier growths for extra tankiness.  For P2, Felix is notably worse out of house due to his Sword / Fist training regime, but you can just make an effort to recruit him quickly.  Player 1 already has Hapi, and Lorenz is already taken, so Flayn suddenly becomes more interesting so as to grab Caduceus to set up long-rang magical sniping with.  Also grab Balthus as another instant recruit who can maybe be Rescue'd back by Flayn if he ends up going Grappler?  Who knows.  P1's already drafted a bunch of Wolves, too, so hey, more Support bonuses.

--
P2: Linhardt
P3: Linhardt   
P4: Marianne   Seteth

P2 doesn't have a dedicated Physic healer yet, and there's a lot of picks from now until their next pick, so taking Linhardt seems respectful.  P3 has Dorothea, but grabbin Lin would let her build more offensively rather than being on Physic duty, so yeah.  Even without Warp, Linhardt still has a role.  Player 4 grabs Marianne on similar logic - in an LTC-ish situation with an out-of-house recruitment, she's probably on Bishop duty, although you could maybe still make her a Valkyrie or something if recruited quickly enough.  Seteth, meanwhile, is a safe pre-leveled pick in an LTC, and also can now participate in C12 some to have a few extra levels for C13 I guess?

--
P3: Raphael
P2: Lysithea
P1: Mercedes, Caspar
P2: Bernadetta
P3: Bernadetta
P4: Caspar, Hanneman

Is SF crazy, picking Raph that early?  Well…  maybe, but I'm more thinking that he has a very short Paralogue whose reward is a pretty good battalion.  And he can uh punch stuff still, sure.  Lysithea is nerfed with no Warp and out-of-house recruitment and a worthless Paralogue but let's not get too crazy, she's still a fine pick.  Strictly speaking, P1 doesn't need Mercedes since they already have Flayn, but it's not like Mercedes is terrible in general, and the Mercedes / Caspar combo is still around to maybe grab a Rafail Gem to make the final boss more reliable.  Unclear if this is really worthwhile in an LTC situation as that Paralogue is scary to do fast, but eh.  Plus, Caspar also helps at Reunion at Dawn.  Players 2 & 3 pick up a redundant Sniper in Bernadetta and another way to make C1/C2/C3 easier.  P4 shrugs and grabs Caspar for the easier Reunion at Dawn, and also Hanneman for more boss Meteor sniping fun to go with Dorothea.

--
P3: Alois
P2: Ignatz
P1: Ashe, Annette
P2: Catherine
P3: Manuela
P4: Hilda
Not picked: Cyril, Anna

Filler picks.  Reminder goes here that Catherine, Hilda, and Cyril are worse than usual due to greatly delayed join time, and Annette is worse than usual due to not having access to her Paralogue.  (And of course Manuela is worse due to Warp being banned, although P4 would have liked to wheel her so she could increase Hanneman's damage - but 'twas not to be.)

Overall draft:

Petra     Yuri       Hapi     Flayn   Balthus   Mercedes  Caspar      Ashe     Annette
Petra     Constance  Lorenz   Felix   Linhardt  Lysithea  Bernadetta  Ignatz   Catherine
Dorothea  Ferdinand  Ingrid   Leonie  Linhardt  Raphael   Bernadetta  Alois    Manuela
Dorothea  Ferdinand  Shamir   Sylvain Marianne  Seteth    Caspar      Hanneman Hilda

Nobody: Cyril, Anna                        


Looking at the overall draft...  hmm, I think they end up reasonably even.  P1 notably lacks a source of Meteor or Bolting, but does have Fetters of Dromi + Petra + Caduceus Hapi.

I suspect that for LTC conditions, I probably undervalued building some sort of Vantage / Retribution tank (because it's not a strategy I use casually).  Not Azure Moon so no Dimitri to do it, but I suspect it's a strategy that if you set up does pay off in finishing rout maps faster.  Felix can do it for P2 and Seteth for P4, and maybe a dopey Balthus build for P1?  P3 kinda doesn't have a good candidate, though I guess RAPHAEL could maybe work.

Anyway, that was my fun theorycraft for the day.  (Actually I did this two weeks ago, but didn't want to just dump some picks, so wrote up some thought processes / explanation.)  No idea how it'd go in a real draft or if I actually tried to run one of these squads-  maybe if I run SS, I'll roll a d4 and try one of these groups to see how it goes.
« Last Edit: November 12, 2023, 05:10:41 AM by SnowFire »

Dark Holy Elf

  • DL
  • Denizen
  • *
  • Posts: 8161
  • Well-behaved women seldom make history
    • View Profile
Re: Theorycrafting! (Because I like competitive metagames too much)
« Reply #998 on: November 13, 2023, 12:13:33 AM »
Neat idea! Some thoughts:

-With Warp banned but LTC the goal, Rescue skyrockets in value (it's... about 50-75% of Warp in terms of distance it allows, typically, assuming you go Dark Flier which well of course you do). I think Constance should be taken earlier. Like... if we're allowing full use of Byleth for all players, Constance should quite possibly be the first pick? P1 gets screwed over at Reunion at Dawn this way buuut you can build a super Byleth if that's not banned. If you try to avoid Byleth hardcarry (e.g. ban Byleth from killing bosses or something) then yeah Dorothea/Petra offer too much value, but I don't think Constance should be taken later than fifth even then, certainly before Ferdinand. Flayn should go sooner too, especially since she's part of a package deal with Caduceus.

-Alternatively you could also ban Rescue too. (The thread you linked might have but it instead banned DLC, and Rescue without DLC classes is way, way less potent.)

-Wrath setups are indeed cool for Rout maps, but... there actually isn't a non-paralogue Rout map post-timeskip, with the sole exception of Chapter 14, which is a bit of a tricky map to do enemy phase setups for (still a bit early for Wrath, and tricky to get a battalion down low enough for Battalion Wrath... gotta do it in Chapter 13, gross). Seteth/Alois as Heroes can thread the needle to set up both Vantage and Battalion Wrath at once, but their mobility is just so awful on 14, so much forest reducing Hero to as little as 2 move. Petra's Battalion Wrath Alert Stance+ build is probably the best method, for the two players who have her.

-Regarding Manuela as a Hanneman adjutant, note that she has the same effect on Dorothea too.

-I also do like the idea of "if you draft a character, you get X turns of their paralogue for free" (for a given value of X, 10 is pretty generous but works I suppose). I don't like extending this to aux fights, except maaaybe the merchant quest ones (and the post-timeskip one should be, like, 3 turns IMO). Other aux fights are just blatant grinding and I see no reason to give free turns there but penalize turns in story fights. That's just me though.

-If you don't allow the player to benefit from undrafted units in any way, Sylvain goes up in value compared to where you have him, I think. Knightkneeler Lance of Ruin from a Cavalier is one of the quickest reliable ways to end Chapter 6 (and is useful in other places too).

Erwin Schrödinger will kill you like a cat in a box.
Maybe.

Dark Holy Elf

  • DL
  • Denizen
  • *
  • Posts: 8161
  • Well-behaved women seldom make history
    • View Profile
Re: Theorycrafting! (Because I like competitive metagames too much)
« Reply #999 on: November 30, 2023, 04:30:26 AM »
So on another forum I was on the subject of spells in FE3H came up and I made the offhand comment that if I were to rank all the offensive spells in the game in usefulness, Aura would probably be at the bottom. Which got my brain thinking: what if I did rank all the game's attack spells?

Of course, such a ranking depends on your criteria, seeing as spells have different availability. Here are the principles I adopted:

-I'm not rating the spells based on who gets them: e.g. I won't say "Miasma is better than Wind because Lysithea is a better mage than anyone who learns Wind", even though it means that I likely actually use Miasma more than Wind.
-I am considering spells at their most common rank for accession. In the case of a tie (which applies to Luna at C/B and Dark Spikes at B/A), I'll pretend they're learned at their midpoint.
-I am rating spells based on how much I like to see them on a unit's spell list.

The last part is vague, but I realized a method which gave results I was happy with: when comparing two spells (call them #1 and #2), I would compare the following:
-Considering units who learn spell #1 but not #2, would I trade #1 for #2?
-Vice versa: would I trade spell #2 for #1?
-Considering units who learn BOTH spells, which one would I rather drop from their list, if forced to?
-Finally, considering units who learn neither spell, which one would I rather add to the list? This last one is probably the case I considered the least.

Note that for all these thought experiments, I considered spells learned at their average rank, as per above. When considering Thoron vs. Bolting, I'm talking about "learning Thoron at C" versus "learning Bolting at A".

As an addendum to the thought experiments, I only considered swaps that were actually possible according to the conventions the game has for when spells are learned. In particular:

-Everyone gets one of Fire, Thunder, Wind, Blizzard, or Miasma at D reason. So when considering a character "losing" a D rank spell in the experiments above, I would assume they still end up with a D rank spell, just the worst one (spoilers: the worst one is Blizzard). You can argue this means I've underrated D rank spells as a whole.
-Nobody learns two spells at the same time, so when considering a character getting a new spell at a certain rank, I'd implicitly either push their old spell at that rank back by half a rank (e.g. from A to A+), replace their old spell, or push the new spell back half a rank, whichever was most beneficial for the comparison.

Finally, do note that I am not ranking non-offensive faith spells at all, at least for now. Attack spells only.

Useful references:
https://serenesforest.net/three-houses/weapons-and-items/black-dark-magic/
https://serenesforest.net/three-houses/weapons-and-items/white-magic/
https://serenesforest.net/three-houses/characters/learned-spells/


The List

10 is the best, 0 is the worst. If two spells are at the same tier it means I did not end up conclusively feeling that one spell was better than another; spells are not ordered within tiers. The gaps between tiers are likely not equal.

10: Mire at D+, Thoron at C, Meteor at A, Bolting at A
9: Death at B
8: Banshee at C, Luna at C/B, Dark Spikes at B/A
7: Ragnarok at A, Excalibur at A, Hades at A
6: Wind at D, Agnea's Arrow at A+
5: Fire at D, Cutting Gale at C
4: Thunder at D, Miasma at D, Bolganone at C, Sagittae at C
3: Swarm at D+, Seraphim at B
2: Nosferatu at D+, Abraxas at A, Fimbulvetr at A
1: Blizzard at D
0: Aura at A


10: Mire at D+, Thoron at C, Meteor at A, Bolting at A

Thoron is crazy. How crazy? Well, here's the thing: if you reduced its range from 1-3 to 1-2 (a massive nerf), it... still has the highest power of any C rank spell. It's less accurate than the rest but the point is even with this massive nerf it's competitive! The 1-3 range just makes it one of the absolute best spells in the game. You can target things you otherwise couldn't, and you get a better linked attack radius too (more on that in a moment). All on a spell which is powerful for its rank and not bad on accuracy.

Compared to Thoron, Mire is gained one battle earlier (the difference between D+ and C), is less powerful (by 6) and less accurate (by 5), but lowers defence by 5. Granted, you need to land two followup physical hits on the target before the defence lowering makes up for the low power, so the advantage only really comes up against monsters. It also has a lot of uses, but is dark magic so can't benefit from Black Tomefaire (Warlock, Dark Flier). Would characters trade for Thoron for Mire, or would Hubert trade Mire for Thoron? I think it's pretty close to a push. Both rule.

Meteor and Bolting are the other top spells. The ability to strike at range 10 is bonkers. Boss killing, ballista sniping, aggroing any formation you want, the possibilities go on. For some unearthly reason Meteor even strikes an area of effect, making it easy to finish off the neighbouring targets with weak attacks like Curved Shot. And there's an argument that casting these isn't even as powerful as their linked effect. Granted, it's only as good as your support list, but handing out +7 or +10 hit (more than that for gambits) to all your friends just by equipping it is pretty great. Excellent on a dancer, good on everyone else. Far and away superior to any other spell gained at A rank or beyond.

Of the two, Meteor is more accurate (by 15) and hits a splash, but Bolting has more power (by 2) and can be used twice as often. I don't have a strong opinion on which is better.

I have trouble comparing the siege tomes and the 3-range spells. Would Marianne etc. trade Thoron at C for Meteor/Bolting at A? Would Constance/Manuela trade Bolting A for Thoron C? I'm unsure. It's a question of early power versus later power. Interestingly I find myself thinking that Manuela should make that trade for sure, because she has trouble reaching A reason, while Constance... maybe not? So it depends. As such, all four of these spells get the top rank.


9: Death at B

3 range is really good! But getting it B instead of C is obviously worse. On top of that, Death is kinda worse than Thoron anyway: -3 power, -5 hit, +2 weight... all for +10 crit? Yeah crit isn't nearly that important on spells (note: this is not the least time I will say this). That said this is still a great spell to have. Lysithea not having it (or Mire) is a major downer; I'd trade her B rank spell for it in a heartbeat. And that says something, because I'd say Lysithea already has the best B rank spell that *isn't* named Death!

One interesting thing is that one 3 range spell devalues another: for Hubert, getting Death is not very important, because he already has Mire (he would rather have learned Dark Spikes or Luna). Similarly, Dorothea or Marianne wouldn't be very interested in trading their B rank spells for Death, even though I don't rate their B rank spells as highly. But since the large majority of mages don't have Thoron or Mire, I'm inclined to rank Death highly, because it's a game-changer for Hapi and it would be for anyone else for whom it were their first 3-range spell.


8: Banshee at C, Luna at C/B, Dark Spikes at B/A

Luna's effect, ignoring resistance, is completely unique. In practice, this means it has a might of "Target's Rsl + 1". On Maddening this tends to be a pretty big number, making it comparable to "power" spells such as Ragnarok. Compared to other power spells it has low hit (65) and low weight (but this doesn't help much since you can't double with it). It has 2 uses, the same as Hades/Agnea's Arrow. The real question is, how does it stack up in terms of power?

Well, at C rank, Lysithea (and hypothetically, other units with Reason boons) can conceivably have it in Chapter 3 (at B rank, e.g. Edelgard, you're waiting longer). It's honestly a pretty weak spell in Chapter 3; average Rsl is about 5 here, giving it a might of 6... which is actually worse than any other C rank spell, on top of its fewer uses and bad hit. This actually just further emphasizes that Luna deserves comparison with power spells, since while those aren't gained until A rank, Luna arguably doesn't start coming into its own until around the point of the game you're getting A rank spells.

By midgame (Chapter 10-16, or 10-14 CF), average enemy res is into the mid teens, so during this stage, Luna is pretty comparable to the power spells. In the last few stages (Chapter 17-22, or 16-18 CF), enemy average res climbs into the twenties, which is ahead of all of them.

Of course, comparing to average Res honestly undersells Luna. This is because low-res enemies can be killed by other spells very well, but high-res enemies? Luna can do far more. No other magic does significant damage to falcon knights, infantry mages, or bosses like Cornelia, Rhea, and Nemesis.

So overall I'm pretty comfortable saying it beats out all the power spells. Lysithea would sooner give up Hades than Luna. Would she sooner give up Dark Spikes? Tough call. Dark Spikes is gained early enough to have a power edge for a while, has +15 hit, and outperforms on cavalry forever (which helps make up for Luna's other wins). I think Luna (at any rank) is better than Dark Spikes at A, but probably worse at B? Since I'm rating by average accession time, I'm willing to give them an equal score.

The last spell I've grouped in here is Banshee, another dark spell. Banshee is a pretty generic C rank spell (better than average power, worse than average hit and weight)... except that like Thoron, it has a very cool extra effect. In its case, it reduces enemy move by 5. For most targets this reduces move to zero and frequently just lets you ignore the enemy for a turn which is quite potent. Hubert has both Banshee and Dark Spikes, and I'd certainly sooner give up Dark Spikes on him... but to be fair that's because his Dark Spikes is gained later, at A - if he had it at B my decision might be reversed. Would Hapi give up Banshee for Luna and/or Dark Spikes? I dunno; that's a tough call. Almost certainly, if she didn't have Hades... which makes me think Banshee is slightly worse than the other two spells I put in this group, but only slightly.


7: Ragnarok at A, Excalibur at A, Hades at A

And now we get to the rest of the power spells. Ragnarok is the middle child. Compared to it, Dark Spikes has -2 might but hits weakness on cavalry. Excalibur has +20 hit, but -4 might, and hits weakness on fliers. Hades has -15 hit, but +3 might.

I'd say power trumps hit in all these cases (it's easy to find a battalion which trades power for hit at a 1:5 ratio, the relevant gap), which would lead to Hades > Ragnarok > Excalibur, but some things complicate this. In Hades's case, it's a dark spell, so there during Advanced tier it takes a power hit compared to Ragnarok if you go Warlock or Dark Flier; it also has one less use. And in Excalibur's case, it hits weakness on fliers, so while it will miss some kills due to lower power (which enemies depends on point of the game), it gains some by killing fliers. I tend not to value flier-killing quite as highly because they're already easily killed by archers, but it's still worth mentioning.

Excalibur vs. Dark Spikes is also worth drawing attention to. Both high-rank, high-power spells strike a weakness. Excalibur has +20 hit and isn't a dark tome, Dark Spikes has +2 power (so +6 against a weakness). I tend to value cavalry weakness more than flying weakness because there are cavalry bosses who lack protection, and because the other anti-cavalry options are range 1 (i.e. Knightkneeler, Spear of Assal), while the other anti-flying options (bows) are long range. And then of course there's the fact that Dark Spikes is sometimes gained before A, so yeah, I'm comfortable saying that Dark Spikes is better.

In conclusion I think this tier is roughly equal: all worse than Dark Spikes and Luna, but all better than some other power spells still to come!


6: Agnea's Arrow at A+

Agnea's Arrow is another power spell, and it's worse than most of the rest, simply because it's gained later. It's got +1 power and -10 hit on Ragnarok (one less use as well) which was already probably a losing trade, but the accession time really makes the difference. Heck, Lorenz gets both this and Ragnarok, and since Ragnarok is a tier earlier, there's no question which one he values more. And while Constance and Dorothea certainly appreciate Agnea in their arsenal, but they'd rather have its power earlier.

That said, Agnea is still good! I think it's comparable to Wind, and that's a good thing. I'll say more on Wind in the section about basic spells, but for now, some quick thought experiments: would Dorothea or Constance give up Agnea for Wind? Lategame power vs. earlygame accuracy... for Constance I think the answer is no (Fire's not much worse than Wind), but for Dorothea maybe yes? Conversely, giving up Wind for Agnea... well, Linhardt or Flayn would do it in an instant, because they have Fire, while Annette... maybe. In all cases, +5 damage vs. Excalibur is enough to attract notice at those midgame one-shots. On the other hand, Lorenz would happily trade Agnea for Wind, since he already has Ragnarok, and +10 hit early matters more than +1 power late.


For the next two sections, I'm going to split the spells by type, rather than rank.

6: Wind at D
5: Fire at D
4: Thunder at D, Miasma at D


Most mages get one of these at D. They're kinda balanced against each other. Compared to Wind, Fire trades 1 weight and 10 hit for 1 power. Thunder makes the same trade compared to Fire. Miasma almost makes the same trade compared to Thunder, but mercifully for it, it doesn't lose accuracy.

On the whole I think Wind > Fire > Thunder is an easy call to make. The power differences do matter - Thunder vs. Wind is often the difference between needing a finisher hit from steel vs. a finisher hit from iron, and there are a few foes like Chapter 1 Dedue (for the Eagles/Deer) where you really want every bit of magical oomph you can get. But 10 hit alone is surely close to outweighing 1 power (worth noting that iron sword vs. iron lance is 1 power vs. 8-9 hit at low levels), and the weight certainly matters. Since this will be the character's lightest spell with a couple exceptions; its weight will be the limit on them doubling.

Miasma vs. Thunder is harder to call. Since Miasma doesn't lose hit, 1 power vs. 1 weight is a comparison which IMO favours Miasma in the earlygame, but favours Thunder later when you're just using the spells to try to double armours anyway most likely. And of course if you're in Advanced tier and don't have Valkyrie access, Miasma suddenly loses power by a lot.

As a side note, Wind even has 10 crit (Thunder has 5, the other two zero). You can't really do crit builds with mages (outside zany double Wrath setups, I suppose) but it's a nice bonus on an already solid spell.


5: Cutting Gale at C
4: Bolganone at C, Sagittae at C


These spells are similar; they're the baseline C rank spells. You want to have one of them: they represent a +5 power increase, give or take, over your previous spells. But these three aren't very exciting past that. Almost everyone has a C rank spell (either these, or the similar-but-better Banshee or Thoron), though Anna and Lysithea buck the trend (Lysithea with Luna, which is quite different; Anna just has nothing!).

Interestingly, they're actually quite accurate, averaging 90 hit (a few characters, like Dorothea and Marianne, even have one as their highest-hit spell). Sagittae sits in the middle (though it does have a strangely high use count at 10); Cutting Gale has +5 hit, +5 crit, and -1 weight, while Bolganone has +1 power, -5 hit, -5 crit. I'd say Cutting Gale's a touch better than the other two, but only a touch.

Comparing them to basic spells is hard. Most people have one of each, so the question becomes... would I rather trade my basic spell for Blizzard, or lose my C rank spell entirely? The former gives you bad accuracy for a couple chapters and in some cases a game-spanning AS hit, the latter gives you power woes for several chapters. I think the answer depends a lot on which spell we're talking about (Wind to Blizzard is 30 hit, ouch) and what spell you get at B rank or beyond. Of course, characters with two D rank spells wouldn't mind losing one, and characters who have two C rank spells (which could include Thoron) don't mind losing one too much either.

On the whole I think I value Wind>Blizzard too much for any of these spells to trump, and otherwise I find it hard to call. Hence my decision to put Cutting Gale (the better C spell) with Fire, and Bolganone/Sagittae with Thunder/Miasma... though this also implies the gap between my 5 and 4 tiers is very small.


3: Swarm at D+, Seraphim at B

Swarm vs. Mire is depressing. Both D+ dark spells. Same hit, a similar debuff effect (Swarm's is speed instead of defence). Swarm gets to win weight/power by 1, Mire gets to win range by 1. This is hilariously unequal.

Swarm mostly shows up and manages to be slightly lighter than Miasma, which is nice, but also weaker and has Blizzard-level accuracy. The speed debuff is only very rarely useful (mostly against monsters). I'm pretty confident saying it's worse than Miasma; Lysithea and Hapi would much sooner give up Swarm then be stuck at 70 hit until C or even B rank. But it's not much worse because it always maintains its niche use against monsters. Just... most units would only give up an already redundant spell to get it.


Seraphim is the first faith spell on this list, and they play by different rules. First of all, the thing to know is that Faith Prowess does not boost hit as much as Reason Prowess (it boosts evade more, but this is only rarely useful because faith spells are so heavy). The gap ranges from 4 points at D+ to 10 points at A+. Additionally, because there are so few offensive faith spells, it becomes reasonable to not set Faith Prowess at all and save yourself a skill slot, at which point the hit gap versus Reason instead varies from 11 (D+) to 20 (A+).

Seraphim has 75 hit, but if you consider the math posted above its actual effective hit is in the 60's at best, i.e. not good. Fortunately its only real use is targeting monsters who usually (not always) have terrible evasion, and hitting them with 24 power. It's honestly not an amazing damage move even against them since it has trouble doubling, but it does cut through barriers nicely (particularly noteworthy since with Caduceus/Thyrsus it makes a great opener to begin a chain of uncounterable barrier-breaking attacks)... unless the barrier in question is an anti-magic one. Still, this is a pretty cool niche. Would mages trade Nosferatu for it? ... yeah, probably, in most cases. Would they trade a non-garbage D rank spell (even Miasma or Thunder) for it? Probably not. I'm inclined to say it's comparable to Swarm: it's overall better (but different) at its monster niche, but doesn't have Swarm's slight earlygame utiity, and requires investing into a faith tree which some would rather not bother.


2: Nosferatu at D+, Abraxas at A, Fimbulvetr at A

Fimbulvetr has 12 might and 65 hit. What were they thinking? As a reminder, Ragnarok is 15/80, Hades is 80/65, Agnea is 16/70. Yes, Fimbulvetr has crit, but good luck doubling anything with this (except maybe the slowest of armours, who you'd kill without a crit), and it's hard to get excited about a 15% higher chance of maybe killing something on your non-doubling chip damage.

It's worse than Agnea. Constance manages to get Fimbulvetr at B (unlike the A rank it is for everyone else) and Agnea at A+ and I'm pretty sure I still care about Agnea more (though at that point it's at least competitive). Dorothea would rather have Agnea too, the extra power is worth the wait. Lorenz wouldn't care either way. Looking at things from the other side... would people with Fimbulvetr trade it for Agnea? Interestingly, people with Fimbulvetr are usually neutral in Reason, so it would be a rather long wait for the added power. So it's actually close in those cases. But there's no case I'm willing to say Fimbulvetr is definitely better, unlike Agnea. And if you're looking to add a power spell to someone's resume, like Linhardt or Annette, they'd certainly prefer Agnea by a lot, +5 power over Excalibur has a major niche would Fimbulvetr can't really fill with its +1.

Marianne and Ingrid would trade Fimbulvetr for Fire, and Constance would rather keep Fire than Fimbulvetr, I think.

What about Thunder/Miasma? This seems more promising, but... honestly, nobody with these would trade them for Fimbulvetr, since they basically all already have a high-power spell of some sort at that rank which outclasses Fimbulvetr. Even Mercedes, who barely cares about Thunder, wouldn't make this trade. What about Seraphim? Would Marianne give up her A rank spell for Seraphim? She... loses 3 peak damage against humans (god, Fimbulvetr only beats Thoron by 3, depressing), but gains 12 damage against monsters, and breaks their barriers, and gets this spell earlier. Yikes? Seraphim feels more important to Ingrid than Fimbulvetr does too. Okay. I think I can convincingly say Fimbulvetr is the worst spell considered yet.


Let's talk about another underwhelming power spell.

Abraxas has 14 might, 90 hit (but really, compared to Reason spells you should hit it by 10-20 points or so), and is heavy with 2 uses. It's best described as worse Ragnarok. Now of course, worse Ragnarok is still useful enough, as Agnea's Arrow proved. If it were a Reason spell with 75 hit gained at A, I would consider it comparable enough to Agnea's Arrow, maybe even a bit better: lose 2 power, but get it earlier. Unfortunately, it's not a reason spell, it's an A rank faith spell. This is bad.

See, the thing is, almost everyone slinging spells wants to go to A reason and beyond, with S being the real goal. Reason Prowess has a strong effect on accuracy (up to 20 hit), and most offensive spells are reason spells. So pushing this up leads to both more accuracy and new spells, and pushing it to S gives a very nice bonus of +1 range. It's safe to say anyone training seriously in magic classes is likely to reach at least A reason, barring the odd hybrid build which sacrifices reason for weaponry (think Holy Knight Bernadetta or maybe Dark Knight Sylvain). On the other hand, because there are no great offensive faith spells, few people want to raise faith ranks just for Prowess. And if you learn Aura at A, there's a guarantee your last useful spell otherwise was at B, so that's a big extra investment for this. To top it off, unless you do go Holy Knight, there's a shortage of classes which boost white tomes: Warlock, Dark Flier, and Valkyrie all buff black spells, but not this. Gremory at least evens the playing field.

For all these reasons, Constance would far rather give up Abraxas than Agnea's Arrow, and Dorothea would not think of trading the reason A+ spell for the faith A spell. The comparison is not close.

Would Annette rather give up Abraxas or Wind? Joke question, I've never gotten Abraxas with her even though it's nominally her strongest spell, it's just too out of the way since she's faith neutral and learns nothing else past C. Would Lysithea rather give up Abraxas or Miasma? Again, silly question, she gets virtually nothing from Abraxas due to already having other power spells. Would anyone trade their non-Blizzard D rank spell for Abraxas? Honestly I'm having trouble seeing it. Maybe Linhardt but since he wouldn't want to delay Warp he wouldn't get this until A+, which is... ugh. So probably not him, even.

What about Abraxas vs. Seraphim? As faith spells they're easier to compare. And Seraphim feels like it wins this competition; while Abraxas does offer power for some, it's power that many people already have in other ways, and very late. Seraphim being earlier with a clear niche at monster-fighting gives it the edge.

What about Abraxas vs. Fimbulvetr? Now we're finally talking. Abraxas has more hit (under any assumptions about reason vs. faith), more power (unless Black Tomefaire enters play), but is more out of the way for most. Do I think this balances? Actually... yeah, roughly. These are our two power spells that aren't very good but fill a niche because any power spell is better than none. Which one is better will depend somewhat on the unit; Marianne would probably prefer Abraxas, but Constance would likely get more out of even an A rank Fimbulvetr than Abraxas. Annette gets very little out of either. Rarely do I feel they'd be too far apart.

To bridge to the next section, how about Abraxas vs. Nosferatu? I think some units might reasonably give up Nosferatu for Abraxas, just to have more uses of a power spell, or a power spell at all. But... certainly not everyone, Nosferatu does have a little niche of its own, and you get it almost for free, while Abraxas requires going out of your way. I'm fine with these two on the same tier.


Finally, let's talk about Nosferatu. Everyone gets this, but very few care about it much. 1 might obviously sucks, 8 weight sucks, 80 hit... is okay. Notably, earlygame is where faith spells suffer least for hit, the gap between Thunder at D+ and Nosferatu at D+ is just 4 hit. So if your D rank spell is Blizzard, Nosferatu is actually your most accurate spell for a bit! (Sorry, Marianne.) Anyway, Nosferatu's niche is that if its crappy stats are good enough to get the job done, you can heal yourself with it, either on player phase or potentially enemy phase if you're feeling spicy. And while this is definitely a niche I have made use of, I feel like it's probably a smaller niche than Swarm has... Swarm being obtained at the same rank, and having similarly poor stats (a bit less hit, better power/weight). but debuffing a monster's speed probably comes up more than the self-heal? You could argue me, but that's my kneejerk. I think Lysithea and Hapi would rather give up Nosferatu than Swarm, and plenty of units would trade Nosferatu for Swarm. Not everyone: notably Marianne and Linhardt start with this and nothing else, so it's a bit better for them. But overall, I think this is where this goes. This is one I could easily be argued on.


1: Blizzard at D

Blizzard is the worst reason spell.

My assumptions for these ratings is "everyone has a D rank spell". As such, Blizzard gets no credit for being better than an empty slot. And it's the worst D rank spell by a lot, so nobody actually wants it. Everyone with it as their solitary D rank spell would rather have another D rank spell instead.

Let's compare Blizzard to Thunder, which is already one of the weaker D rank spells otherwise. Blizzard's numbers? -1 power, -10 hit, +10 crit. As I've noted previously, crit isn't very useful on spells. Blizzard is heavier AND less powerful than the average D rank spell, and also has the lowest hit by far. All for... 5 more crit than Wind? Yeah who cares.

Shamir and Bernadetta are the only characters who learn this as well as another D rank spell. Would they rather have Blizzard, or... almost anything else? Especially for Bernie, who currently lacks a spell above 80 hit, almost anything else would be better. For Shamir, maybe you could sell me on Blizzard for a second C rank spell being a losing trade. Maybe.


0: Aura at A

If you've ever had the joy of watching Roderigue get attacked by bandits in Felix's paralogue, seeing him get doubled while having 50-some displayed hit on them in return, you may have an idea how bad Aura is.

Aura tries its best to make the other offensive faith spells look good. Compared to Abraxas it has -2 power and -20 hit. Gross. It's the least accurate spell in the game, fullstop. Compared to Fimbulvetr, it has +5 listed hit (so anywhere from -3 to -15 in reality), equal power, and... even 5 less critical, which is sad since crit might be the one thing Aura has to try to have a niche. Plus it's a faith spell, which as mentioned is a bad thing.

How about Aura vs. Blizzard? Again, Shamir and Bernadetta are the only characters who have a use for Blizzard using my assumptions (that everyone must get one D rank spell, and that Blizzard is clearly the worst). Would they trade Blizzard for Aura? ... honestly, I don't think so. The slight critfishing niche Blizzard has against slow monsters beats Aura trying to do the same with 5 more crit but 8 more weight, I think, and more to the point they just get Aura way earlier; Shamir won't ever train faith to A (bane). It's closer for Bernie but I think the answer is the same.

What about everyone else? What if we make them an offer: you get Blizzard at D+ in addition to your other spells, or you get Aura either at A, or a half-rank after your existing A rank spell (since nobody is trading their existing A rank faith spell for this). ... again, I think some characters *might* get a little value out of Blizzard for its tiny crit-fishing niche, while basically nobody is grabbing Aura. You could make a slight case for someone like Linhardt, but... on average, I think no.


So there you go!

Erwin Schrödinger will kill you like a cat in a box.
Maybe.