tierlist FE6 Tier List - Redux

[XeKr]

1 turn isn’t that much if an extra warp improves reliability significantly. Going from 20% 3 turns to 75% 3 turns is like 11 turns on average (not saying those are the exact numbers here). Like I would consider a triple warp with Lalum/another bosskiller or something if it actually helped that much. Like you were talking about 20I in the other topic. I don’t really have recent familiarity with Ilia but that chapter seems insane enough (calced to be 1.62% with base Perceval) to warrant something like that (or what you suggested with clearing half the map). 

This strat has some contingencies and transition options which helps it, certainly. I think we may be able to estimate some rough numbers (with some worst case built in for the likely stuff, like Lilina is hit, Miledy needs healing) to see if the extra warp is worth it. Also as you said we do need to be mindful of uses to have enough for the big skips (through walls and such). 
 

Edited April 13, 2017 by XeKr

[Irysa]

2 Warp 20I isn't quite as bad as dondon's 0% as long as you have a SM with the benchmarks, a mount (that is going to get Berserked) is warped with them to the throne room and either drops them to EP the boss or drops them back a little to EP the boss themselves before the SM goes for the finish. I didn't formally crunch numbers but typically there's a more omnipresent CoD for the SMs if they have to EP and PP but the CoS of the bosskill goes up. I'm pretty sure the CoS of that strat is in the 20%s though since I recently did it the other day and I repeated it enough times to not seem super unreliable.

You can't really keep extending your clear properly without warping in someone to Restore the unit getting Berserked in the throne room or doing the clear half the map to absorb Berserk stuff (which would save warps at least), but Staffers don't exactly cope that well with Killer Ballista/Killer Bow Sniper.

Thinking about it, if willing to blow extra warps, you'd probably do some shit like warping Saul into the boss area on T2 to Restore the mount, then mount has to kill the Sniper, SM probably Elixirs or attacks Roartz again, and you continue moving towards the throne area with everyone else, clearing enemies out along the way. If the boss goes down before you get into rescuechain seize range with Roy then you Warp him in, if not then you rescuechain him to the throne (Killer Ballista is probably out of ammo by then, and the Warp would only save 1 Turn w/o improving reliabiity).

This is like, a 2 Warp 5/6 Turn or a 3 Warp 3/4 Turn though, which is relatively demanding.

Edited April 13, 2017 by Irysa

[XeKr]

At a glance, I think Boots!Miledy can reach the Druid and proceed from there?

[Irysa]

That's a fair point, but you still effectively have to blow the same amount of Warps anyway, unless you're going to like, Warp Rutger on T2 carrying Roy which means he won't get any boss action in unless you sac Miledy's turn to drop Roy out. Roy is exposed to the Sniper/Killer Ballista if you do that too, so probably you're killing the Sniper instead. This further drags it out and makes it comparable to the alternatives. Also Warping Miledy down is somewhat less than ideal now that I think about it since that Knight near the start has a Horsekiller, so Miledy is one of the ideal units along with Niime to choke the entrance.

Basically that idea, or the "clear everything out" strat eliminate the chance of even getting a possible 2 turn, so I'm inclined towards a strat that has restore as a backup unless the warp differences are favorable.

I made this to quick reply but I'm going to compile a lost of all warps I think are probably mandatory/expected.

Mandatory is bold, italicised is workable round.

14x - If Roy is bulky enough and your 10 mov flier can ORKO the boss you can do a 3 turn without Warp but it's not great on the reliability front, so you're probably using at least 1 Warp here. You can also take a 3 Turn 1 Warp if you do something like have the Flier carry Rutger to the throne so they can both try to go at him. Either way 1 Warp is expected for saving about 2 turns and improving reliability (unless you don't have Saul)

16 - If you're going faster you need like 3 Warps to save a bunch of turns, but I'm not sure if an efficiency would actually do this.

16x - 1 Warp is possible but it's pretty unreliable and even trained Roy has issues surviving so it's more reasonable to expect at least 2 Warps, with a 3rd to possibly improve reliability.

17S - I think Warps only save like 1 turn per use on this map since you basically give your Flier more reach. If you don't have a flier or don't have a delphi shield on the flier, then you need at least 1 Warp to not have to spend an extra bunch of turns going through the forest/up and down for the bridge.

18S - You probably want at least 2 Warps to try to clear quickly but the Rescue Staff is pretty good at saving Turns on this map instead. I feel like it's very difficult to improve reliability considerably here and still go at a fast pace due to issues outlined in the other thread though, so you might be more inclined to scratch Warps entirely, idk, would have to actually run the map.

19S - 14 Mag Saul means 2 Warps, and you can't really augment reliability any more than that unless you start taking a much longer period of time to try to set up from the side.

20S - At least 2 Warps unless Roy has enough bulk to survive in the corner of the room.

20XS - 1 Warp. If Roy is bulky enough to survive Wyverns on the Forest or something you could make it 0 Warp for 2 turns I think, or if he can't Niime can pull off a Nostank and Rescue Staff Roy in instead.

17I - You can get by without Warp but it does speed things up if you have spares, 1 Warp probably saves like 1 or 2 turns and subsequent ones probably save 1 each.

18I - Same as above really.

19I - At least 1 Warp so you don't have to slog through the chokepoint. Bosskill isn't that unreliable either.

20I - At least 2 Warps, more to improve reliability.

20IX - At least 1 Warp, probably 2 Warps, unless you can meet certain criteria for getting away with 1.

21 - At least 2 Warps. They each save a significant amount of turns each, I don't think subsequent ones do much besides improve reliability.

21X - You can get away with 1 Warp if you aim for the 2 turn but at least 1, probably 2 Warps otherwise.

22 - At least 2 Warps.

23 - At least 1 Warp and 1 Rescue, but each subsequent Warp improves reliability quite a bit (since you get to hide more people further back away from Ballista/Wyverns)

24 - Probably 1 Warp, mostly just for turn saving since the last throne traversal distance is massive, so it can be passed up on if you ran out before.

You basically only seem to have a buffer of like, 1 or 2 Warps/Rescues for improving reliability before you kind of can't get away with it anymore. Ilia is more leniant at least maps like 18 or 19 are more reasonable for progressing through normally which can save Warp uses for when you really want them. This also helps to improve the case for training Roy, since the bulkier he gets the more Warps you can dedicate to trying to make bosskills more reliable.

Edited April 13, 2017 by Irysa

[XeKr]

What about something lame/awesome like 2 use Javelin/Spear that breaks into Armorslayer? >_>. It’s basically the fast strat but can be extended and maybe saves a warp due to dealing with Berserk.

I accept your warp use allocation plus which are likely mandatory or just useful, and think it will be a useful post to refer to later. Though I will again emphasize that any strat that is significantly more reliable, or with transition/extension options instead of death, tends to have much better expected turncounts. You are correct the marginal value of any extra warp should still account for turncount/reliability tradeoff elsewhere.

fwiw I don’t actually mind time-efficiency arguments here. Or rather, I don’t currently care to discuss them in detail. I think they generally do not differ substantially from classic expected turncount arguments, aside from technical/ui/player execution/etc stuff. 

Another thing to note: I think there is a natural tendency for FE players who started with classic to be averse to resets in general. Because we tried so hard to save every character from permadeath so that reset=failure=bad was ingrained. obv no data on this but seems apparent from my impression of the community and personal experience. So that adds some bias toward the more reliable strats (but I’m also okay if the accepted criteria mostly ignores this in favor of the statistics and unbiased practicality).

I’ve favored training Roy to some extent for a while, and this might be relevant for certain early characters. Trained Allance do mean a slightly weaker Roy, and unlike Rutger/Shanna’s role, you can still easily get a 3rd paladin in Noah (though Str/Spd/longterm is weaker)

[Irysa]

The 2 use Javelin thing could work but it makes the reliability of even killing him pretty sus since the hitrates become notably worse, and that Druid also has Nos. I was originally thinking of a Brave Lance to try to get him before he even counters, and it would still let you get the KO if you hit the next 2 (assuming you quad which Miledy should)

I'm not disputing the fact that the higher reliability clears with more warp can result in large disparities in better expected TCs, but merely that there is effectively a hard limit on how many "reliability" Warps one can afford to use. Within that buffer I'd be comfortable assuming more warps or possibly a slower strat depending on the map, but I'm pretty skeptical that you can afford to start taking Warps out of maps like 19S, 16X, 21X, etc, especially when we consider that it's not like the bosskills in slower clears actually become more reliable, we're just able to try to keep going without resetting. So it's quite likely in fact for them to take an extra bunch of turns due to misses and the like, which may skew it further in favor of the originally faster 2/3 turns.

I'm not averse to trying to avoid resets, but I think maybe it would also be useful to consider that longer more complicated strategies tend to involve resets in order to construct them to a point of acceptable reliability; ie, we realise we needed a different loadout for x unit to deal with y because we didn't forsee the enemy moving or attacking a particular target, etc. The shorter strategies are simpler generally, which limits the amount of resets one probably needs to do in the planning stage. I realise this is getting really abstract so I'm not saying it has to be tiering policy, but I'm interested in what you think of that.

Training Roy might be easier if you neglect Alance but Chapter 8 is one of the best places to get Shanna extra EXP before she promotes, and Noah has to soak a lot of it to promote for Chapter 9. It's also not like you can easily just shovel the EXP Alance aren't taking into Shanna in the earlier maps like 3 either, because like most units at that point she has to be handled with a degree of care. If we want to talk benefits of Noah, I was thinking that, given that ideally one funnels exp into one Cav rather than two in order to reach higher levels before promotion (12 to 14), if you've dumped a lot of EXP into one Cav and they're statistically poor, it probably takes less EXP to get Noah to level 11 or so than it will to get the other Cav to an acceptable level. This is sort of a similar argument to the Rutger/Fir thing admittedly, but that is what got me wondering about how we could apply that kind of logic elsewhere, to see if it makes sense in other contexts or if the tenants are shaky.

 

EDIT: also I goofed and thought gates were 2 def in FE6 fsr so Monke and Randy need +1 str on those calcs I did the other day for SM "reliable" ORKOs, jesus. I don't think either SM is really getting either of them down in one round realistically without Bartre route's Energy Ring.

Edited April 14, 2017 by Irysa

[XeKr]

On 4/14/2017 at 2:25 PM, Irysa said:

The 2 use Javelin thing could work but it makes the reliability of even killing him pretty sus since the hitrates become notably worse, and that Druid also has Nos. I was originally thinking of a Brave Lance to try to get him before he even counters, and it would still let you get the KO if you hit the next 2 (assuming you quad which Miledy should)

I'm not disputing the fact that the higher reliability clears with more warp can result in large disparities in better expected TCs, but merely that there is effectively a hard limit on how many "reliability" Warps one can afford to use. Within that buffer I'd be comfortable assuming more warps or possibly a slower strat depending on the map, but I'm pretty skeptical that you can afford to start taking Warps out of maps like 19S, 16X, 21X, etc, especially when we consider that it's not like the bosskills in slower clears actually become more reliable, we're just able to try to keep going without resetting. So it's quite likely in fact for them to take an extra bunch of turns due to misses and the like, which may skew it further in favor of the originally faster 2/3 turns.

I'm not averse to trying to avoid resets, but I think maybe it would also be useful to consider that longer more complicated strategies tend to involve resets in order to construct them to a point of acceptable reliability; ie, we realise we needed a different loadout for x unit to deal with y because we didn't forsee the enemy moving or attacking a particular target, etc. The shorter strategies are simpler generally, which limits the amount of resets one probably needs to do in the planning stage. I realise this is getting really abstract so I'm not saying it has to be tiering policy, but I'm interested in what you think of that.

Ah, didn’t see fewod having enemy equipment (?) so no idea what that Druid had. edit: and I wrongly assumed the hm stat topic on this forum didn't distinguish properly between the Druids (who in what location has which weapon). It does say the Berserk Druid has Nos, so my mistake. 

I also think sometimes those who decry turncount metrics and prefer safe turtling don’t realize certain unreliable factors in the latter. For example ambush reinforcements, or bosses with existent crit, or other things late in the map that are still rng heavy or otherwise extend chapters (and are therefore very punishing because you already took so many turns/time to get there). 

However, the general benefit to the slower strats is that the player simply doesn’t overextend their units and so there are contingencies even for cases of bad rng (instead of purely trying to account for the average-ish outcomes or bruteforcing favorable ones). Maybe the 8 turn clear or whatever isn’t actually near 100% reliable but the chance of reset is near 0%, even if it may drag on. The strat is overall more complicated in requiring more actions, but probably just required some adaptation, not resets. Because of the aforementioned, it doesn’t always work this cleanly, but if minimizing resets supercedes other factors, then these strats become more important.

Though I do feel that players who can devise the reliablefast clears should also easily be able to beat maps the leisurely, somewhat trivial way. Many do, and have done for years, in casual runs. There are lots of ways to beat maps slowly, using whatever team with lots of exp/gold/etc, and how “easily” is way too subjective. The turncount constraint is objective and forces optimization; that’s the point. It’s when we go too far in the other direction, and accept strats with low success chance and just the simple need to bruteforce reset through, that it becomes questionable if this is actually being more strategic/tactical and efficient (and relevant to character viability).

I generally don’t think we should account for planning resets, since it’s too variable based on individual player knowledge, intuition, and even stuff like theorycraft style. Some may find it’s easier to just try the strat instead of calculating everything perfectly for the first attempt, while others might prefer to think through the possibilities with fewod/paint before going ingame. The player can calculate the battle forecast so no resets should be needed for that kind of planning of who is attacking whom with what weapon. In terms of AI, generally people who figured it out originally had to do many resets, but it’s somewhat predictable to the aware (Roy is favored, no counters favored, last deployment slot, etc), or random beyond our manipulation (strats should account for this, reliability is lower if it relies on enemies moving a certain way). To my understanding, there’s still some other mysterious factors, unless specifically delving into the code, which also introduces some uncertainty in proper treatment of information available to the player in/out of the game. Presumably though, informed players might know about certain unknowns on the initial map overview, such as those hidden in fow or reinforcements, or stuff like support lists/chest contents, as it’s fully documented on this site or in memory. Regardless in the end, some strats are derived from scratch while others are adaptations of previous ones, and I think when something becomes “known” we shouldn’t really have to reinvent the wheel. All that is a really long way of saying that we simply can’t self-consistently account for how much brainpower/skill/planning it took to figure out various strats and how much we expect the tier player to exhibit.

“who is the tier player (player the tier list is applicable for)” arguments are pretty old and well-argued. I think it’s preferable for the list to reflect the highest known level of play, the most efficient strats (including positioning, resource allocation, etc) for various team compositions. As otherwise we’ll get into territory of players deliberately making unoptimal and inefficient decisions due to ignorance or stubbornness. Knowingly making “bad” plays feelsbad? And I guarantee no one can agree on what decisions are intuitive or not for average or veteran or casual or whatever players (the curse of knowledge makes this difficult in general). Unlike efficiency, we can’t point to something objective and say the numbers show this is faster and more reliable on average. Again, minimizing that kind of arbitrariness is arguably the whole point.
 

On 4/14/2017 at 2:25 PM, Irysa said:

Training Roy might be easier if you neglect Alance but Chapter 8 is one of the best places to get Shanna extra EXP before she promotes, and Noah has to soak a lot of it to promote for Chapter 9. It's also not like you can easily just shovel the EXP Alance aren't taking into Shanna in the earlier maps like 3 either, because like most units at that point she has to be handled with a degree of care. If we want to talk benefits of Noah, I was thinking that, given that ideally one funnels exp into one Cav rather than two in order to reach higher levels before promotion (12 to 14), if you've dumped a lot of EXP into one Cav and they're statistically poor, it probably takes less EXP to get Noah to level 11 or so than it will to get the other Cav to an acceptable level. This is sort of a similar argument to the Rutger/Fir thing admittedly, but that is what got me wondering about how we could apply that kind of logic elsewhere, to see if it makes sense in other contexts or if the tenants are shaky.

 

EDIT: also I goofed and thought gates were 2 def in FE6 fsr so Monke and Randy need +1 str on those calcs I did the other day for SM "reliable" ORKOs, jesus. I don't think either SM is really getting either of them down in one round realistically without Bartre route's Energy Ring.

I was mainly just talking about the overall notion that the exp pool is limited. Taking one of Allance to level 12-14 for solid benchmarks is simply less exp available for Shanna/Roy/etc than taking Noah to 10-11 for just a competent Paladin. It’s a minor factor that’s nevertheless relevant when trying to assess Allance’s utility. In their absence, some exp is freed up for others, and you still have an option for a 3rd Paladin, so it’s not like they’re super essential to efficient clears in the Isles. ofc given 8 move and solid Steel onerounds and good-ish durability, they do have some significant return for that investment. We should just keep in mind the actual costs and value in the on-off analysis.

The relevance I see to the Rutger/Fir thing is we shouldn’t just rate Noah like 2/10 because he’s outclassed/superfluous and not in more optimal teams (Noah never gets a crest because he takes it away from the optimal unit, probably Allen?). RNG is perhaps a factor, but level 14ish Allance is going to outstat level 11 Noah in a great majority of contexts. Later promo Noah (probably not worth considering the strongest part is in the Isles) is also lacking compared to the other of Allance/Perceval/even Treck.

Monke is lame. I went Bartre route and know I had a fairly high level Fir and some Wo Dao uses left (for AS). Specifically do remember Rutger not doing well, tho obvs I used Fir significantly midgame (Rutger still got some bosskills). I don’t really know how reasonable the numbers are as I went for most items (except very minor stealables), which was turns to train and reasons to need more combat units. It seems 24 AS is needed to double the intial Brave Bow, then Elixir/Physic spam is an option. Might be a map to not use staves on, it’s not as large as others and no walls/etc (river a minor terrain issue), though I think I used at least 1 Rescue to speed things along.

Edited April 18, 2017 by XeKr

[Irysa]

20 hours ago, XeKr said:

I also think sometimes those who decry turncount metrics and prefer safe turtling don’t realize certain unreliable factors in the latter. For example ambush reinforcements, or bosses with existent crit, or other things late in the map that are still rng heavy or otherwise extend chapters (and are therefore very punishing because you already took so many turns/time to get there). 

It's funny you mention this because I finished that run of FE6 I was doing yesterday and Miledy and Rutger both got sub 5% crit by Chapter 24 Manaketes, ending the clears 5 turns in. Fir/Shanna hype? (tho Shanna prob needs a robe and obviously neither of them can take 2 Manaketes at a time like Miledy can). Biases Percival a little too since he has high enough luck to generally avoid crits from them as well!

Speaking of endgame there, more nice uses for Bolting in C21 - Roy can provide Support as Lilina chips one of the starting Wyvern Lords so that as long as Niime dodges it (or you gave her two robes/1 robe 1 draco/she procced Hp/rolled down on Strength), she can KO it with Nosferatu, saving a Sleep Staff use or w/e (also has a decent chance to crit him too). She can also compensate for things like Wyverns rolling high on AS so that base Niime + Body Ring doesn't double, or Niime missing an attack, via finishing off at 10 range from the top of the map, with Dancer support allowing for multiple attacks. That can free up a Speedwing (since Speedwing + Body Ring Niime reliable doubles all Wyverns). This is assuming the "Yodel warps Percival + Niime into the middle of the map T1" strat which is pretty reliable overall (I used Durandal Percival to improve reliability in KOing the Wyverns that attack him). Armorslayer Miledy once again proved to be pretty great as probably the best unit choice possible to engage in Murdock's room, her CoD being by far the lowest of the choices due to high defence and no horseslayer weakness, along with a more accurate 4HKO. Though, getting to the secret shop in 4 turns was difficult without having the 10 mov flier do it, but I was able to barely squeak to it with Shanna. Well, in theory anyway, I ended up having to Warp her to it on the last turn due to a movement error I made, but MIledy killed the boss herself anyway so I didn't need a second warp on Percival to attempt to finish.

Back to Lilina though; she also doesnt have a whole lot of difficulty helping out at the top left of C22 as she can hugely dent/ORKO the General on EP (Javelin means he's not going to OHKO), and she also has a realistic chance of 2HKOing the Brave Sword Hero safely there (20/22 Mag + Forblaze, not super unrealistic), or at least helping out, with a nice big Heal if neccessary on the supporting unit. Given the player typically wants to concentrate their strongest units with high mov/boots use on the right side, this is a pretty nice niche for a second string infantry unit to have, as Heroes are a real bitch of an enemy, especially when they have Brave Swords. Unfortunately I didn't end up testing the "reliable" throne room strat as Rutger decided to doublecrit with Durandal against Zephiel, allowing for the instant clear. May as well outline that though - Roy opens door, Rutger attacks Zephiel, if he gets Zephiel into KO range for Niime + Shin they go in to attempt to finish then Roy danced to seize (or if you really want you stall another turn anyway regardless but I think Zephiel might get Physic'd if he drops too low), if he doesn't then Apocalypse Niime is danced to double Sleep the Heroes whilst the rest of the gang ko as many enemies as they can and Yodel Saint Staffs. If you REALLY want higher reliability Rutger doesn't have to attack Zephiel immediately but the pressure this puts on the LHS to get away from the reinforcements is pretty testing so I think it's better to go in hard unless you somehow have enough good units on the left to try to get further in the corridor to avoid the double Hero, Sniper+Druid reinforcement combo.

20 hours ago, XeKr said:

Though I do feel that players who can devise the reliablefast clears should also easily be able to beat maps the leisurely, somewhat trivial way. Many do, and have done for years, in casual runs. There are lots of ways to beat maps slowly, using whatever team with lots of exp/gold/etc, and how “easily” is way too subjective. The turncount constraint is objective and forces optimization; that’s the point. It’s when we go too far in the other direction, and accept strats with low success chance and just the simple need to bruteforce reset through, that it becomes questionable if this is actually being more strategic/tactical and efficient (and relevant to character viability).

Sure, I wasn't disputing that taking it slower is typically simpler (my last run of this game wasn't really that fast and I didn't really ever struggle), but in many late mid to lategame maps in this game IMO there tends to be a distinct area of "extra effort" inbetween the triviality of slowing down quite a bit and bruteforcing a very fast clear. In a casual clear of a map it's all well and good to make sure nobody dies on any given turn and steadily progress forward, but trying to compute the actual worthwhile-ness of having slight chances of a reset and how many turns your "risky" action saves is often pretty vague without having to stop and crunch numbers, and trying to compute multiple parts of them together in an exhaustive manner I find often find just biases me towards either end of the spectrum. That is, "I'll just keep taking a few extra turns here and there so I don't face the prospect of a reset" or "I'll just force this really quick clear" (as long as it's like, 3 turns max). This isn't really part of expected TC, and in the earlier portions of this game (and the vast majority of other games frankly) this is not a concern, but I do think that some of FE6 is in a relatively unique position given enemy stats relative to yours + accuracy and ORKO concerns for you and enemies.

Chapter 23 is to me the pinnacle of this because with 3 Warps and Apocalypse Niime the map becomes a total joke - Dragonshield Percival (or Alan, but he got speed screwed in my run so Brunya doubled him, tbh Miledy could probably do it too), can be dance warped to reach the right forest tile near the gate and depending on how durable they are only have to dodge like 1 attack from the Manakete and the boss whilst the rest of the enemies can't scratch or have nearly 0 hitrates, and with a weak enough weapon (like a hand axe) he wont clear the way so they just try to surrounding him. Your unit's position triggers the AI to stop blocking the entranceway (I think they're tied to the Manakete) so you have a clear path next turn to Warp a bosskiller to kill Brunya (who is weighed down by Bolting so easier to hit and doesn't counter), then Warp Roy. This strategy doesn't even require a Saint Staff use and you only need to kill the 2 enemies near the start + the boss. Add a Sleep Staff on the Killer Ballista and I'm confident it has something like a 40 to 50% CoS for a 2 turn, wheras actually trying to slow down to make sure the throne area is clear so that the bosskiller can continously engage whilst still going at a reasonable pace is pretty taxing, especially due to all the Killer Weapons and likely requires multiple stall turns or Saint Staff uses. I think if one is willing to blow an extra Rescue Staff or position Shin more precariously you can also have him try to finish off if Rutger/Percival fail to kill with Durandal, (but he'd need a lot of Strength to pull it off).

20 hours ago, XeKr said:

I generally don’t think we should account for planning resets, since it’s too variable based on individual player knowledge, intuition, and even stuff like theorycraft style.

Yeah fair enough. Granted though, even experienced players still learn things here and there about the enemy AI and the way they'll chose to move and attack even today, so that's not something you can act like one can generally handle as long as they are vets.

 

20 hours ago, XeKr said:

I was mainly just talking about the overall notion that the exp pool is limited. Taking one of Allance to level 12-14 for solid benchmarks is simply less exp available for Shanna/Roy/etc than taking Noah to 10-11 for just a competent Paladin. It’s a minor factor that’s nevertheless relevant when trying to assess Allance’s utility. In their absence, some exp is freed up for others, and you still have an option for a 3rd Paladin, so it’s not like they’re super essential to efficient clears in the Isles. ofc given 8 move and solid Steel onerounds and good-ish durability, they do have some significant return for that investment. We should just keep in mind the actual costs and value in the on-off analysis.

Even if the EXP pool is limited, there are a limited amount of units that can take that EXP early on, and they are definitely high up on the pecking order. A larger limiting factor on their EXP gain is more Marcus's Axe Rank, since not only is Hammer Marcus nifty for flexability, but if you opted to get the Halberd in Chapter 3 then he makes good use of it in Chapter 7. However Marcus tends to ORKO Soldiers if he uses Axes, so that limits the EXP available for other units in various Chapters.

 

20 hours ago, XeKr said:

It seems 24 AS is needed to double the intial Brave Bow, then Elixir/Physic spam is an option.

On reset he actually has 25 base speed so 21 AS with the weapons he wields. You could rig it down if you wanted to I guess assuming we're playing on the unfixed RNG patch or are allowing suspend abuse.

Edited April 18, 2017 by Irysa

[XeKr]

On Sages/Lilina: My impression is they can “chip things at passable accuracy” which was basically my rough assessment of Wolt. Sure Bolting has range but directly costs Boots lategame, and the sages needed heavy investment to get to this point instead of freely contributing.

My other qualm for Lilina in particular regards her durability though this might have to do with the fact I was also training Fir. Basically if Fir/Paladin/Rutger/whoever is killing some of the melee dudes on enemy phase, usually it’s the Hand Axe ones that survive to player phase and I think, “I could have Lilina finish this bandit at 80-90% disp hit, but if she misses I’m reseting 40% of the time. Doesn’t seem worth?”

I feel like I can be easily convinced on this matter, as some of the things described seem fairly impressive, but I figured I would clarify my initial thoughts on the payoff of sages. Basically just 2 things: if it’s not that hard to train (minor turncount/reliability impact. I guess I’ve hyped Fir but she has a semblance of an enemy phase and theoretically nice payoff in the ideal case) and if they’re improving later clears by a significant amount, statistically (that can obviously mean a lot of things, but broadly speaking just sizably enough to return that investment cost).

On 4/18/2017 at 6:22 AM, Irysa said:

Sure, I wasn't disputing that taking it slower is typically simpler (my last run of this game wasn't really that fast and I didn't really ever struggle), but in many late mid to lategame maps in this game IMO there tends to be a distinct area of "extra effort" inbetween the triviality of slowing down quite a bit and bruteforcing a very fast clear. In a casual clear of a map it's all well and good to make sure nobody dies on any given turn and steadily progress forward, but trying to compute the actual worthwhile-ness of having slight chances of a reset and how many turns your "risky" action saves is often pretty vague without having to stop and crunch numbers, and trying to compute multiple parts of them together in an exhaustive manner I find often find just biases me towards either end of the spectrum. That is, "I'll just keep taking a few extra turns here and there so I don't face the prospect of a reset" or "I'll just force this really quick clear" (as long as it's like, 3 turns max). This isn't really part of expected TC, and in the earlier portions of this game (and the vast majority of other games frankly) this is not a concern, but I do think that some of FE6 is in a relatively unique position given enemy stats relative to yours + accuracy and ORKO concerns for you and enemies.

Chapter 23 is to me the pinnacle of this because with 3 Warps and Apocalypse Niime the map becomes a total joke - Dragonshield Percival (or Alan, but he got speed screwed in my run so Brunya doubled him, tbh Miledy could probably do it too), can be dance warped to reach the right forest tile near the gate and depending on how durable they are only have to dodge like 1 attack from the Manakete and the boss whilst the rest of the enemies can't scratch or have nearly 0 hitrates, and with a weak enough weapon (like a hand axe) he wont clear the way so they just try to surrounding him. Your unit's position triggers the AI to stop blocking the entranceway (I think they're tied to the Manakete) so you have a clear path next turn to Warp a bosskiller to kill Brunya (who is weighed down by Bolting so easier to hit and doesn't counter), then Warp Roy. This strategy doesn't even require a Saint Staff use and you only need to kill the 2 enemies near the start + the boss. Add a Sleep Staff on the Killer Ballista and I'm confident it has something like a 40 to 50% CoS for a 2 turn, wheras actually trying to slow down to make sure the throne area is clear so that the bosskiller can continously engage whilst still going at a reasonable pace is pretty taxing, especially due to all the Killer Weapons and likely requires multiple stall turns or Saint Staff uses. I think if one is willing to blow an extra Rescue Staff or position Shin more precariously you can also have him try to finish off if Rutger/Percival fail to kill with Durandal, (but he'd need a lot of Strength to pull it off).

Expected turncount is just something we can refer to, if we want to look at something objective, to explain why something vague like “taking a few extra turns” doesn’t mean taking 20 more and grinding on reinforcements, or “I’ll just force this really quick clear” doesn’t mean resetting until a large bunch of criticals/low hit rates proc (a few might be okay). It’s a metric that gives a sense of the turncount/reliability tradeoff.

Expected turncount doesn’t actually tell us the “best”, “optimal”, “easiest”, or “most efficient” strat, or what takes the least effort, or how we should be playing (we should technically avoid such assertions, but it’s not worth it to overload every post with nuance and disclaimers. I still try, but sometimes fail, to avoid making such claims). Ultimately, that’s just too subjective. Maybe you value time playing the game a lot, maybe you value not reseting a lot, maybe efficiency in FE means something completely different to you, we’ve touched on a few things already and there are countless others. All it tells us is (approximately) what is faster (in a turncount sense) and more reliable (in a chance of success/death/reset sense) on average, numerically. That’s it. People interested in discussion can then interpret that information in the lens of a tier list. 

I think you’re getting at philosophical matters outside the territory of this metric (not to say they aren’t interesting to discuss sometimes). Please clarify if I’m misunderstanding.

As for C23, I actually find the 50% 2 turn to be a pretty good strat. I think it would be challenging to find a 4 turn with nearly 100% reliability (and comparable expected turncount). I actually see this as a clear case where the metric is working, as in why we might prefer this 50% 2 turn over a 70% 5 turn or a 90% 10 turn or whatever.
 

On 4/18/2017 at 6:22 AM, Irysa said:

On reset he actually has 25 base speed so 21 AS with the weapons he wields. You could rig it down if you wanted to I guess assuming we're playing on the unfixed RNG patch or are allowing suspend abuse.

Is there a better source for the most consistent boss stats (should we note them?). iirc sf/fewod aren't always consistent about it, though maybe some have been updated for the initial rn seed on reset stats.

Some initial numbers. Rutger needs 9 procs or more at 50% + 15 base + 1 promo to hit 25 Spd. This assumes he’s never ramming the cap, as he’s unlikely to cap Spd given typical promotion level (around level 11 it's 6.25% to ram Spd).

11/12 (18 levels): 59.27%
11/13 (19 levels): 67.62%
11/14 (20 levels): 74.83%
11/15 (21 levels): 80.83%

Fir needs 3 procs or more at 55% + 22 at ??/1 to hit 25 Spd. This assumes a later promoted Fir with capped Spd (high probability given promotion level. I'll quote myself from another post "In 13 levels (when she caps “on average”) she actually has a 64.37% chance of capping speed. 16/19 levels is 87.59/96.58%")

??/7: 74.47%
??/8: 84.71
??/9: 91.15%
??/10: 95.02%

Given the exp formula, what kind of levels and level gap are we thinking is actually reasonable here? (edit: also see question at bottom of this post)

________________________________________________________________________________

As an aside, I want to take a closer look at a particular situation, Rutger or Deke vs. Henning since apparently Deke is getting significant credit for cases with the first Hero Crest. We are pretty confident that case is suboptimal, but how much so? (Note the following is largely proof-of-concept)

I haphazardly threw together something from some old code of mine and dondon’s arena simulation that calculates the battle forecast from hp, str/mag, skl, AS, luck, def/res, weapon hit, weapon crit (also include supports and class crit), weapon mt (include supports), terrain defense (include supports), and terrain avoid (include supports), of the player and enemy units, then runs a battle between them by rolling rns as appropriate on player/enemy phases. Things like throne healing and conditional vulnerary usage can be accounted for with minor additions to the code at the appropriate location. 

My simulations of 10000 battles suggest that for 12/2ish Deke with 37 hp, 14, str, 16 skl, 14 spd, 8 luck, 12 def and a Killing Edge vs. Steel Blade Henning (using stats on this site, easy to correct if they are off on reset):

Deke: 57 hit, 8 dmg, 31 crit
Henning: 73 hit, 11 dmg, 2 crit, heals 10% per enemy phase up to max health

Deke wins an estimated 24.69% of the time if fighting straight up. The battle lasts around 4.22 rounds on average (each phase is a round). The numbers don’t seem to change significantly unless Deke is cleanly 3hko’d (36 hp, 11 def or something), though it’s easy to play around with stats above/below average.

Deke wins 76.10% of the time using a strategy where he uses a Vulnerary if Henning can 2hko Deke’s remaining Hp. The battle lasts 11.83 rounds on average. Deke doesn’t win 100% of the time because of crit and the fact that Henning outdamages 10 hp per turn healing. Deke can retreat to heal and prevent resets, though this will add lots of turns. It’s also possible that another strat in ferrying a staffbot (though Shanna also needs to ferry Roy) or using the C7 Elixir or chipping with another Light Brand user results in better chances. We can also simulate those cases, though I don’t know which is most reasonable at this time. Will look into this a bit more. 

For 11/2ish Rutger with 37 hp, 13 str, 21 skl, 20 as, 6 luck, 10 def and a Killing Edge vs. Steel Blade Henning: 

Rutger: 65 hit, 7 dmg, 63 crit, doubles
Henning: 63 hit, 13 dmg, 4 crit, heals 10% per enemy phase up to max health

Rutger wins 87.38% of the time. The battle lasts 2.07 rounds on average. He wins 81.67% of the time in 2.44 rounds if he’s at 12 Str. 

Rutger wins 93.77% of the time using a strategy where he uses a Vulnerary if Henning can 2hko Rutger’s remaining Hp. The battle lasts 2.52 rounds on average. Rutger doesn’t win 100% of the time due to enemy crit (and Henning outdamaging vulneraries, though it’s very unlikely Rutger won’t crit enough in time).

To simplify matters, we are making various other assumptions like having enough Killing Edge uses, but I think that’s mostly all reasonable and relatively lenient. I have not yet included binominals for growths in the sim (trivial to calc separately), though I think it’s straightforward to do so(?). Just makes longer, arguably messier vectors, so I’ll think about a cleaner implementation later. Arguably speaking, the separate calc plus sim is almost more informative in a practical sense, given that strategies chosen (that are being simulated here) are usually based on stats in a given playthrough or particular context (the current one of interest) rather than the whole ensemble of possible ones. 

*numbers could easily be off due to bad math and/or insufficient iterations, though it’s “trivial” to tweak as required/desired. Probabilities are estimations based on simulations, unless otherwise noted.

Some thoughts at this time
-    too many numbers ahh
-    As expected, Rutger is really important in killing Henning efficiently. Scott/Gelero/Randy/others may also be interesting. 
-    If we assign significant weight to cases where Rutger does not promote with the first Hero Crest, we are entertaining quite inefficient contexts based on this initial analysis. Something like 25% vs. 87% chance of overall success or reset, or on average 5+ extra turns (and still around 15+% less chance of success), with the caveats noted earlier. We should still keep that in mind for Deke and other characters whose performance depends on the usage of known, more optimal, options.

btw what level do we think Rutger is around C12xish? Might run some stuff vs. 20/1 Fir (her most optimistic, probably unrealistic, case). depends how good sov is.

Edited April 20, 2017 by XeKr
stuffs

[grandjackal]

Question: When is Rutger promoting? We just assuming he eats the hero crest immediately? Because chapter 8 is easy enough to assume we didn't need to do that. 11/2 sounds like sandbagging to be honest. 

Also, I feel like there's an important distinction between 11 rounds and 2. At some point, it shouldn't really matter how good Dieck is without promotion (of which, btw, he is kinda sorta meh without promo. I would honestly start using Fir or Gonzales over him). Hell, that number alone should really cement Rutger as a god simply because he is almost single handedly responsible for saving you a majority of your time. 

In fact, why is Percival above him? An efficient run isn't efficient without Rutger almost by definition. Sure, Percival is literally the best combat unit in the game aside from Miledy, and he comes without any investment like that, but unless you purposefully filled your army with mouthbreathers, having an army of asskickers shouldn't even be that special by the time he joins. Rutger on the other hand will still arguably be your best bosskiller at that time on the basis of crit and accuracy. 

[Irysa]

holy shit grandjackal is alive

12 hours ago, XeKr said:

On Sages/Lilina: My impression is they can “chip things at passable accuracy” which was basically my rough assessment of Wolt. Sure Bolting has range but directly costs Boots lategame, and the sages needed heavy investment to get to this point instead of freely contributing.

My other qualm for Lilina in particular regards her durability though this might have to do with the fact I was also training Fir. Basically if Fir/Paladin/Rutger/whoever is killing some of the melee dudes on enemy phase, usually it’s the Hand Axe ones that survive to player phase and I think, “I could have Lilina finish this bandit at 80-90% disp hit, but if she misses I’m reseting 40% of the time. Doesn’t seem worth?”

I feel like I can be easily convinced on this matter, as some of the things described seem fairly impressive, but I figured I would clarify my initial thoughts on the payoff of sages. Basically just 2 things: if it’s not that hard to train (minor turncount/reliability impact. I guess I’ve hyped Fir but she has a semblance of an enemy phase and theoretically nice payoff in the ideal case) and if they’re improving later clears by a significant amount, statistically (that can obviously mean a lot of things, but broadly speaking just sizably enough to return that investment cost).

I mean, thats a roughly 6% chance of a reset on average against a specific enemy type that she doesn't have to neccessarily take depending on how you're setting up kills for her. Fir is facing similar chances of death of dying on an enemy phase at base level since she gets 2HKO'd by basically everything.

What I'm describing is imo not "impressive" but it is nifty. Like, playing without it isn't that big a deal, and it's not like Lilina herself has perfect accuracy (pretty average with bolting tbh, Roy support helps though), and there are ways to negate the need for it. The best uses of Bolting besides the 16x Bolting Sage to me all seem to at best be reliability augmenters, and the best use of a non Bolting wielding Sage is typically for throwing a big dent in bulky promoted enemies (Paladins, Wyverns, Heroes, etc). There are a few instances of things like say, Chapter 19I where you could drop a trained Sage on the Forest over the mountain on T1 and ORKO some cavs and put the Paladin into Horseslayer range (can steal his Knight Crest too before you do it), but these are pretty rare and obviously Lilina has difficulty doubling without a Speedwing typically, wheras Lugh eventually gets there.

As for the cost to Boots, Bolting is worth 1500 when sold, which is a non trivial amount but you're using at least 1 charge of it in 16x. You also don't need to buy them any more Bolting at the shop because tbh it's not really worth factoring Bolting uses into a reliable PT past Chapter 21 I think? Like, there's no real practical application in 21x, a Sage is better placed to help on the LHS in 22 which means no trying to bolting chip Zephiel (who is really resistant anyway), 23 unless you can somehow ORKO the Killer Ballista Sniper (need 34 Atk 16 AS so 22 Mag 21 Speed, each benchmark being feasible for one but not the other Sage) there's no real initial targets of value beyond chipping the nearby Druid so someone might not have to take a counter from the Nos. 24 lol rigging bolting crits in a reliable PT, I guess if they can double with it they can typically prevent someone taking a counter, but considering boots distribution if you're taking a Sage to 24 then they are using Forblaze more anyway. I somewhat question putting Boots on a Sage too TBH...I guess Forblaze access for 24 is nice, and I will admit Rutger kinda felt pretty auxilluary due to being forced into using a Wyrmslayer, but Rutger likely needs the Mov boost to help with Brunya and Zephiel. I suppose you could cut the mount typically used to carry Fae via just having Yodel or your Dancer take her from someone else.

Anyway TL;DR, we don't need to buy more Bolting IMO. So what we're asking for is more like 6200 range for their core items (not going to crunch what all those Tomes are worth if sold), which is about 1.5 pairs of boots. This is pretty competitive with (better than?) Fir I think since she likely needs some other Boosters too.

I basically see the tweaks to reliability and the monetary cost for Sages and Fir being fairly comparable, given Fir isn't even actually going to be augmenting reliability a significant portion of the time if Rutger doesn't underperform. The main caveat here obviously being that Fir is self sufficient wheras the Sages require quite a bit more babying.

12 hours ago, XeKr said:

I think you’re getting at philosophical matters outside the territory of this metric (not to say they aren’t interesting to discuss sometimes). Please clarify if I’m misunderstanding.

Eh, kind of. I'm not really objecting to expected TC or expecting it (durr) to accomodate these factors, just more trying to add in that whilst it is a good objective metric, our application of it can be teased with a bit if the circumstances allow. Also I ran some numbers for the 23 clear, it's prob more like 30 to 40%, but I didn't account for "what if our units suck" factor either so I'm not sure it's an example of it neccessarily working (the joke being I'd have to run calcs that I'm too lazy to do or attempt to work through a more rigorous strategy...). Although you can probably say it will stretch resources to go slower too given someone like Miledy will need to spam Malte on a few of the enemy groups.

12 hours ago, XeKr said:

Is there a better source for the most consistent boss stats (should we note them?). iirc sf/fewod aren't always consistent about it, though maybe some have been updated for the initial rn seed on reset stats.

If I stop being lazy I can grab all of them personally since I have movie files for every chapter to check. I typically just use them instead of site reference these days <_<

12 hours ago, XeKr said:

Given the exp formula, what kind of levels and level gap are we thinking is actually reasonable here? (edit: also see question at bottom of this post)

By 18, if we're playing efficiently then factoring in arena use where applicable and bosskiller privilege I'd say 11/10 for Rutger is very reasonable, you can squeeze another 1 or 2 levels if you try probably. He was like 11/13 at 18I in my last run (got to 14 off Martel) without too much effort, but that did have a pretty big turn buffer so cutting it down by 1 or 2 levels intuitively seems correct. I don't know when Fir is promoting but if Rutger takes bosskills and promoted enemies then that impedes her ability to gain promoted EXP significantly; this is a good point to raise against Rutger+Fir team combos, the EXP pool is limited.

Specifically for 12x, I'd expect Rutger to be 11/5 to 7 by that point (depends on arena usage, 6 is prob reasonable).

12 hours ago, XeKr said:

Some thoughts at this time
-    too many numbers ahh
-    As expected, Rutger is really important in killing Henning efficiently. Scott/Gelero/Randy/others may also be interesting. 
-    If we assign significant weight to cases where Rutger does not promote with the first Hero Crest, we are entertaining quite inefficient contexts based on this initial analysis. Something like 25% vs. 87% chance of overall success or reset, or on average 5+ extra turns (and still around 15+% less chance of success), with the caveats noted earlier. We should still keep that in mind for Deke and other characters whose performance depends on the usage of known, more optimal, options.

Henning has +1 Speed and Def compared to SF on a reset.

Randy is not that important since MIledy handles him fairly effectively. Scott and Gelero are assholes but you do have the option to give Dieck one or two Speedwings by Gelero (although tbh I'll concede that by this point, Alance can be pretty competitive with Dieck for bosskilling).

12/1 Dieck is sandbagging, he can be higher than that. Dieck does take quite a bit longer but one has to remember that it won't take 11 rounds because you're realistically going to be having Shanna carry in more units in to help contribute (promoted Alance, Marcus/Zealot, etc). Rutger obv is a big deal and Dieck's relative contribution for the bosskill is significantly cut down due to how much worse at it he is, but he does improve at it as time goes on and he does have Durandal access that other units won't have. The biggest problem with Dieck is obviously that he doesn't have high enough speed to realistically start doubling bosses without giving him speedwings, but this is a problem shared by basically everyone who isn't Rutger, Fir or Shin (even Miledy needs help lol). Shanna easily doubles but her attack is pretty low, Lance is like the next best bet but even he takes really long time to start doubling a lot of these enemies.

Dieck basically fufills 4 of the 5 criteria for bosskilling - he's accurate, bulky enough to take hits, has above average attack, can use crit boosting/effective weaponry and can be transported to bosses pretty easily. He doesn't get the last one (reliably doubles), but his accuracy lead over someone like Alan or his Attack lead over someone like Lance, combined with their own doubling problems still makes him better at it than any non SM till MIledy and Percivcal show up.

Edited April 20, 2017 by Irysa

[XeKr]

sfblog: SoV is pretty fun, though kinda easy (on Hard/Classic). Though I never finished FE2 before so eh. Not sure how much replay value it’ll have for me tho.

On 4/19/2017 at 11:18 PM, grandjackal said:

Question: When is Rutger promoting? We just assuming he eats the hero crest immediately? Because chapter 8 is easy enough to assume we didn't need to do that. 11/2 sounds like sandbagging to be honest. 

I like promoting Rutger in C7 (convoywarp the Crest) so he can reach the boss without triggering the reinforcements early. It adds some consistency to the bosskill beyond the Marcus/Zealot teamup w/ armorslayer tradebot strat. The extra move might help a bit in C8 given how long it is. Tho Rutger probably isn’t fighting too much in C8 anyways besides maybe the boss (the promotion helps quite a bit for that fwiw). C8x is Henning. I don’t see how Rutger is getting too many more levels. Maybe Deke can get a few if he gets the C2/C3 bosskill? 14/2 doesn’t actually change much (bad math was more a factor), there’s a few numbers later in this post.

On 4/20/2017 at 7:10 AM, Irysa said:

I mean, thats a roughly 6% chance of a reset on average against a specific enemy type that she doesn't have to neccessarily take depending on how you're setting up kills for her. Fir is facing similar chances of death of dying on an enemy phase at base level since she gets 2HKO'd by basically everything.

What I'm describing is imo not "impressive" but it is nifty. Like, playing without it isn't that big a deal, and it's not like Lilina herself has perfect accuracy (pretty average with bolting tbh, Roy support helps though), and there are ways to negate the need for it. The best uses of Bolting besides the 16x Bolting Sage to me all seem to at best be reliability augmenters, and the best use of a non Bolting wielding Sage is typically for throwing a big dent in bulky promoted enemies (Paladins, Wyverns, Heroes, etc). There are a few instances of things like say, Chapter 19I where you could drop a trained Sage on the Forest over the mountain on T1 and ORKO some cavs and put the Paladin into Horseslayer range (can steal his Knight Crest too before you do it), but these are pretty rare and obviously Lilina has difficulty doubling without a Speedwing typically, wheras Lugh eventually gets there.

As for the cost to Boots, Bolting is worth 1500 when sold, which is a non trivial amount but you're using at least 1 charge of it in 16x. You also don't need to buy them any more Bolting at the shop because tbh it's not really worth factoring Bolting uses into a reliable PT past Chapter 21 I think? Like, there's no real practical application in 21x, a Sage is better placed to help on the LHS in 22 which means no trying to bolting chip Zephiel (who is really resistant anyway), 23 unless you can somehow ORKO the Killer Ballista Sniper (need 34 Atk 16 AS so 22 Mag 21 Speed, each benchmark being feasible for one but not the other Sage) there's no real initial targets of value beyond chipping the nearby Druid so someone might not have to take a counter from the Nos. 24 lol rigging bolting crits in a reliable PT, I guess if they can double with it they can typically prevent someone taking a counter, but considering boots distribution if you're taking a Sage to 24 then they are using Forblaze more anyway. I somewhat question putting Boots on a Sage too TBH...I guess Forblaze access for 24 is nice, and I will admit Rutger kinda felt pretty auxilluary due to being forced into using a Wyrmslayer, but Rutger likely needs the Mov boost to help with Brunya and Zephiel. I suppose you could cut the mount typically used to carry Fae via just having Yodel or your Dancer take her from someone else.

Anyway TL;DR, we don't need to buy more Bolting IMO. So what we're asking for is more like 6200 range for their core items (not going to crunch what all those Tomes are worth if sold), which is about 1.5 pairs of boots. This is pretty competitive with (better than?) Fir I think since she likely needs some other Boosters too.

I basically see the tweaks to reliability and the monetary cost for Sages and Fir being fairly comparable, given Fir isn't even actually going to be augmenting reliability a significant portion of the time if Rutger doesn't underperform. The main caveat here obviously being that Fir is self sufficient wheras the Sages require quite a bit more babying.

Eh, it’s kinda some small things that add up. Fir 31 base avoid and 1.6 avoid growth per level on average. 0% chance of death if exposed to 1 69% hit axe enemy (and this is pretty important since she can at least finish 1 enemy on pp and engage 1 on ep, quite safely), 2.55% for 2 axes, 6.8 for 3 axes. After gaining like 10 avoid from levels, it’s 0.44% for 2 axes, 1.27% for 3 axes. Lilina is running into various death chances for just 1 kill and doesn’t get out of ohko range for a while, even with some levels (iirc). She can hit Mercs/Archers safely but she,and most units who weaken them for her, don’t have great accuracy on them.

There’s also forest/house/village tiles that increase avoid, which hurts Lilina more since she has less skill, no triangle, and needs someone to weaken the enemy unit anyways (and few have perfect accuracy there). She only has 10 base att with Fire (11 with Roy), obviously no doubling, and Elfire is 20 less hit. Fir can directly benefit from the tiles too since she has more negligible death chances on terrain.

tbh I wouldn’t feel great about exposing Fir to 3 axes and no terrain either, if she’s still lowlevel and it’s late in the chapter, despite only “minor” impact on expected tc.

All that said, I don’t really know if it’s worth nitpickying the details as we’re kinda handwaving the investment during the Isles. I know I started it, but I had the disclaimer about its reasonableness (or lack of) >_>. Need to put some more into practice (I think I trained too many units also).

On Bolting, an issue I see is tomes burn uses even if they miss. The reliability benefit to Bolting chip seems only worth it if it’s supplementing clears, not crucial to them. If relying on an inaccurate Bolting hit for the clear to work, the strat’s chances probably aren’t great. It’s good as a reliability buffer, which means using them but accepting some misses. Trying to do this throughout the mid/lategame will quickly run out of uses I think. Even if we don't need much (or any) after C21, the accuracy + uses thing means it's hard to hype Bolting imo.  

A side note: in general when I mention “costs” I don’t just mean monetary. Gold is a cost but it can be translated into turncount/reliability which helps us better compare it to other investments like exp or such. And unused gold/items don’t really have practical value; I see their value as related to how they are potentially improving the efficiency of clears.

I don’t think Fir deserves too much hype if she’s only improving reliability a bit. Her imagined role is predicated on being the primary mid/late bosskiller. Either Rutger is screwed or not on the team and Fir is by far the highest offense on the team, or Rutger was used but Fir took over midgame and has better stats due to the laterpromo and therefore better bosskill chances (we can just not deploy Rutger if we want the slot and only need 1 SM). Again, this depends on how costly her investment actually is, and if she actually pulls much better probabilities lategame, which is certainly not settled. I’m not really convinced myself, I just currently find it interesting to think about. 
 

On 4/20/2017 at 7:10 AM, Irysa said:

Eh, kind of. I'm not really objecting to expected TC or expecting it (durr) to accomodate these factors, just more trying to add in that whilst it is a good objective metric, our application of it can be teased with a bit if the circumstances allow. Also I ran some numbers for the 23 clear, it's prob more like 30 to 40%, but I didn't account for "what if our units suck" factor either so I'm not sure it's an example of it neccessarily working (the joke being I'd have to run calcs that I'm too lazy to do or attempt to work through a more rigorous strategy...). Although you can probably say it will stretch resources to go slower too given someone like Miledy will need to spam Malte on a few of the enemy groups.

Outside the territory of the metric doesn’t mean outside the territory of the tier list. The op is imo a bit dated, but this is still ostensibly an “efficiency” list with all that implies. No one is proposing to rank characters purely by the increase in expected turncount in their absence, weighted by the expected turncount of the clear given various team compositions, blahblah. Difficult to compute the whole model rigorously (even if approximations generally suffice), and there are still many additional factors like we’ve discussed that might impact our thinking of the characters, but aren’t directly part of the metric (as it purposely doesn’t include them).

It’s absolutely fair to scrutinize expected tc and its applications or implications; we’ve already done a lot of this. With respect to the effort thing, I see it like the intuition/knowledge thing. Once someone puts in the effort to find the strat with best expected tc (to the best of their ability), anyone can then “trivially” execute it. This is also why I sometimes note the best known strategies or highest known level of play, to cover the bases for any more fast/reliable strats out there. 

You noted the middle spectrum fastreliable strats may be difficult to devise so you feel some preference for the extremes, but you can imagine someone else simply preferring the slow strats since others require more effort (counting squares and various logistical things). Yet just disregarding the fast clears seems like a regression. Someone else who has seen how a clean warpskip works might try to just replicate the strat instead of taking lots of time to play through the chapter their own way (don’t have to constantly refer back to a reinforcement list too for safety and such). Etcetc. Ultimately, we all have our biases, and the list can’t really reflect all of them simultaneously. So this could actually be a situation where deferring to the statistics might help, as an objective compromise of sorts.

For C23, what’s limiting the reliability? The bait’s survival or the bosskiller chances or something else? What about burning a Rescue for Lalum (though that’s maybe pushing it)? Or take an extra turn or so to setup things better, with multiple combat units (we did use lots of Boots) and only use a single warp or so for Roy. I’m pretty sure that’s what I did, though I don’t recall the details.
 

On 4/20/2017 at 7:10 AM, Irysa said:

If I stop being lazy I can grab all of them personally since I have movie files for every chapter to check. I typically just use them instead of site reference these days <_<

By 18, if we're playing efficiently then factoring in arena use where applicable and bosskiller privilege I'd say 11/10 for Rutger is very reasonable, you can squeeze another 1 or 2 levels if you try probably. He was like 11/13 at 18I in my last run (got to 14 off Martel) without too much effort, but that did have a pretty big turn buffer so cutting it down by 1 or 2 levels intuitively seems correct. I don't know when Fir is promoting but if Rutger takes bosskills and promoted enemies then that impedes her ability to gain promoted EXP significantly; this is a good point to raise against Rutger+Fir team combos, the EXP pool is limited.

Is that something you can upload to share? Mediafire/dropbox/google drive/etc. 

11/10 Rutger has 40.18% of 9 Spd procs in 16 levels. 77.28% with Speedwing.

I think ideally (in theory), Rutger/Fir teams would use Rutger for early bosskills and while Fir is training, then Fir takes over around C12x and gets most of the bosskill/promoted exp. She’s lower promoted level (so better exp gain) with higher stats (from unpromoted levels) so it’s efficient to do so. Though on that note I said maybe level 20 for the full benefit but it seems pretty ridiculous, 17 possibly (it’s way less kills) but perhaps still a stretch.

On 4/20/2017 at 7:10 AM, Irysa said:

Randy is not that important since MIledy handles him fairly effectively. Scott and Gelero are assholes but you do have the option to give Dieck one or two Speedwings by Gelero (although tbh I'll concede that by this point, Alance can be pretty competitive with Dieck for bosskilling).

Randy has 18 Spd so an SM with 60 crit that doubles might be comparable to Miledy with 30 crit not doubling, even factoring in the logistics of getting over there. Scott/Gelero have quite significant crit so elongated battles have relatively poor reliability (or rather, taking extra turns doesn’t improve chances as much as usual). Might show this more rigorously later. 

Also I think there’s just the Isle Speedwing and the Desert Speedwing? And Miledy/Allen/others certainly compete for them. 

Need 19 Spd to double Gelero in 12x. 10 base + 2 Speedwing + 2 promo means 5 procs are needed. 14/7 (15 levels) Deke has 48.45% chance of that. And again, 19 Spd doesn’t guarantee a clean win either and he still has 30 less crit. 

On 4/20/2017 at 7:10 AM, Irysa said:

12/1 Dieck is sandbagging, he can be higher than that. Dieck does take quite a bit longer but one has to remember that it won't take 11 rounds because you're realistically going to be having Shanna carry in more units in to help contribute (promoted Alance, Marcus/Zealot, etc). Rutger obv is a big deal and Dieck's relative contribution for the bosskill is significantly cut down due to how much worse at it he is, but he does improve at it as time goes on and he does have Durandal access that other units won't have. The biggest problem with Dieck is obviously that he doesn't have high enough speed to realistically start doubling bosses without giving him speedwings, but this is a problem shared by basically everyone who isn't Rutger, Fir or Shin (even Miledy needs help lol). Shanna easily doubles but her attack is pretty low, Lance is like the next best bet but even he takes really long time to start doubling a lot of these enemies.

Dieck basically fufills 4 of the 5 criteria for bosskilling - he's accurate, bulky enough to take hits, has above average attack, can use crit boosting/effective weaponry and can be transported to bosses pretty easily. He doesn't get the last one (reliably doubles), but his accuracy lead over someone like Alan or his Attack lead over someone like Lance, combined with their own doubling problems still makes him better at it than any non SM till MIledy and Percivcal show up.

Made some adjustments like actually truncating the Throne healing instead of leaving a decimal, updated Henning stats, etc.

11/2 Rutger with 37 hp, 13 str, 21 skl, 20 as, 6 luck, 10 def, Killing Edge
Vs. Henning 63 hit, 6 dmg, 63 crit, doubles

14/2 Deke with 39 hp, 15, str, 17 skl, 15 as, 8 luck, 12 def, Killing Edge
Vs. Henning 57 hit, 8 dmg, 31 crit

Henning with 38 hp, 12 str, 21 skl, 14 as, 7 luck, 13 def, Steel Blade, 3 def/30 avo/10% heal Throne
Vs. Rutger 63 hit, 13 def, 4 crit
Vs. Deke 71 hit, 11 dmg, 2 crit

Fight it out
Rutger: 79.31% win%, 2.50 rounds
Deke: 27.05% win%, 4.32 rounds

Vulnerary strat
Rutger: 92.37% wins, 3.26 rounds
Deke: 84.60% wins, 10.46 rounds

I noted in the last post potentially other, better cases for Deke. A number of them, like ferrying more units, might take more turns as only Shanna can traverse the gap. Also in those cases it’s not like the bosskill suddenly becomes very reliable, given how terrible everyone’s hit rates are, it only improves the odds a bit. Plus if others are in range, Henning possibly isn’t attacking the Killing Edge user on enemy phase (Rutger might have an adv with less Def here). And Dieck can only get shared credit with the other combat units in a teamup, so the contribution is lessened still.

Again, I specifically picked Dieck because it seems commonly accepted that giving him the Crest, and not using Rutger, is a realistic situation even in efficient play. His ranking on this list seems to take this into heavy consideration, that he can carry weaker teams in less efficient contexts. 

I’m trying to get a sense of what this means in terms of suboptimal situations we are willing to entertain. ~3-4 turns slower in just 1 chapter is a lot, and it’s even more throughout the game (imagine adding the berserker bosses and various fast ones later). Characters have gone up and down lists for less. Is mount/flight even worth that much in various single chapters? I don’t particularly care if Deke moves or not, I’m just saying that if we are considering inefficient contexts to this extent, we should also think about comparable ones for other characters. Note, Deke can still be upper mid regardless because of his earlygame and other combat roles (hand axe stuff, clearing space for Roydrops or safer warps, etc), plus not like other upper mid units are doing any better (though they have other notable utility besides bosskiller). 

Those bosskiller criteria are relevant, but rather than a checklist, I think it’s more informative to simply compute/simulate the relevant bosskill chances (which obviously takes hit rate, bulk, att, weapons, spd, and their relative worth, all into account). Deke may be pretty good outside his Spd, but it turns out that enough Spd to double is really important to bosskilling (obviously).

Edited April 22, 2017 by XeKr

[Irysa]

1 hour ago, XeKr said:

I like promoting Rutger in C7 (convoywarp the Crest) so he can reach the boss without triggering the reinforcements early. It adds some consistency to the bosskill beyond the Marcus/Zealot teamup w/ armorslayer tradebot strat. The extra move might help a bit in C8 given how long it is. Tho Rutger probably isn’t fighting too much in C8 anyways besides maybe the boss (the promotion helps quite a bit for that fwiw). C8x is Henning. I don’t see how Rutger is getting too many more levels. Maybe Deke can get a few if he gets the C2/C3 bosskill? 14/2 doesn’t actually change much (bad math was more a factor), there’s a few numbers later in this post.

I agree with promoting Rutger for Devias if you're opting to dodge the reinforcements, but if you're engaging them (since they are quite EXP rich and relatively easy to handle) then you can delay it a little. Mind, I still tend to promote him at the start of 8 anyway, because it's not like Leygance goes down THAT easily to other units.

Dieck can take the C2 and 3 Bosskills pretty easily, but I tend to prefer to give them to Marcus as if he gets a speed proc (25%) it makes Chapter 4 much easier. In a world where we're not planning on using Rutger Dieck should probably take them though. C5 and 6 bosskills if they're set up correctly for him or if it just so happens he ends up finishing off due to whatever errant misses or crits are a thing too, as are the C7/8 ones in the same way they're relevant for Rutger.

1 hour ago, XeKr said:

On Bolting, the problem I see with not buying more is tomes burn uses even if they miss. The reliability benefit to Bolting chip seems only worth it if it’s supplementing clears, not crucial to them. If relying on an inaccurate Bolting hit for the clear to work, the strat’s chances probably aren’t great. It’s good as a reliability buffer, which means using them but accepting some misses. Trying to do this throughout the mid/lategame will quickly run out of uses I think.  

My point is you're not going to be buying more to use for post C21 anyway unless you have sub 4k gold and can't make up the last stretch, because it's not really that useful.. It's not a question if if you buy more or not.

I don't think there are any strats that are not just pure LTC specific that actually rely on Bolting to hit, just Bolting serving as a reliability augmentation.

Tome accuracy is a thing yeah but there arent so many uses for it that it's a huge deal. 5 seems really small but when there are only about 5 places where it'll have a more than marginal impact, it doesn't matter too much.

1 hour ago, XeKr said:

I don’t think Fir deserves too much hype if she’s only improving reliability a bit. Her imagined role is predicated on being the primary mid/late bosskiller. Either Rutger is screwed or not on the team and Fir is by far the highest offense on the team, or Rutger was used but Fir took over midgame and has better stats due to the laterpromo and therefore better bosskill chances (we can just not deploy Rutger if we want the slot and only need 1 SM). Again, this depends on how costly her investment actually is, and if she actually pulls much better probabilities lategame, which is certainly not settled. I’m not really convinced myself, I just currently find it interesting to think about. 

I think at this point someone just has to commit to actually doing an efficient playlog of the game and trying to train then utilise Fir for lategame bosskills. I don't have any plans to do something like that anytime soon.

1 hour ago, XeKr said:

For C23, what’s limiting the reliability? The bait’s survival or the bosskiller chances or something else? What about burning a Rescue for Lalum (though that’s maybe pushing it)? Or take an extra turn or so to setup things better, with multiple combat units (we did use lots of Boots) and only use a single warp or so for Roy. I’m pretty sure that’s what I did, though I don’t recall the details.

Survival is very reliable if it's Barrier/Pure Water + Dracoshield Percival, it gets worse if it's anyone else since they probably need to dodge one of the Manakete or Brunya's Bolting.

Rutger needs to score a crit with Durandal to KO Brunya, can't reasonably expect to 2HKO (she's really bulky), and whilst 75 disp hitrate+ and 20~25ish crit is solid, that's not stellar. Energy Ring Percival/Alance can ORKO with Durandal without crits but has worse Accuracy. Trying to Warp a Refresher up is problematic because Yodel's Warp range is 4 less than Niime's, Saul probably didn't get any boots, and standing further up starts to expose more frail units to Wyverns + Ballista which is not ideal. Shin to try to finish is the only reasonable way to improve the bosskill (which isn't that bad either actually, 18 Str + Murgleis will give him two chances to finish off if Rutger hit twice with 18 Str and Durandal). However you have to either position Shin more precariously (as in, he has to survive some Wyverns), or blow a Rescue AND Warp Charge to get him into position so that Yodel's Warp Range will suffice for him to finish off which is really resource intensive and might not be doable depending on your earlier strats.

I suppose a way to alleviate this problem and save at least 1 Rescue Staff use is to have Yodel eat an Energy Ring, since then he can warp Shin from the spot he kills the Manakete into Boss Range. That still costs a Boots though and 4k gold for a reliability augment seems a liiiiitle questionable, although I suppose it becomes nearly foolproof with it might be worth it? I think I'm more put off by the extra Warp tbh.

The big reliability drop was because I thought the Druid had Nosferatu for some reason, he has Flux. That combined with some further thoughts alleviate most of my concerns so I think the strat will work fine.

Main problem with trying to go slower and sending multiple combat units up there is that there are a lot of Killer Weapon enemies with nearly maxed out attack, and you have to get rid of the enemies blocking the gate unless you do the bait trigger strat. Getting through the Manakete, Nosferatu Druid, and then KOing the boss (where you probably need a someone hit again from 1-2) is relatively demanding. Miledy Maltes the Manakete, Paladin Brave Lances the Druid, Rutger Durandals, Shin Murgleis? The Wyverns around the throne love being dicks with positioning though so trying to set up a reliable 3 turn to me seems like it's going to force you to fight more groups of enemies than you really want to unless somehow you have enough boots for like, Wyrmslayer Paladins to help make sure Wyverns aren't completely clogging the area.

1 hour ago, XeKr said:

Is that something you can upload to share? Mediafire/dropbox/google drive/etc. 

PM'd you them. Don't realllllly want to publically share the movie files themselves and I'd rather just document the boss stats at some point when I feel like it.

1 hour ago, XeKr said:

Randy has 18 Spd so an SM with 60 crit that doubles might be comparable to Miledy with 30 crit not doubling, even factoring in the logistics of getting over there. Scott/Gelero have quite significant crit so elongated battles have relatively poor reliability (or rather, taking extra turns doesn’t improve chances as much as usual). Might show this more rigorously later. 

Also I think there’s just the Isle Speedwing and the Desert Speedwing? And Miledy/Allen/others certainly compete for them. 

Need 19 Spd to double Gelero in 12x. 10 base + 2 Speedwing + 2 promo means 5 procs are needed. 14/7 (15 levels) Deke has 48.45% chance of that. And again, 19 Spd doesn’t guarantee a clean win either and he still has 30 less crit. 

You still need 16 Str and 22 AS and A Swords to perform notably better than Miledy given you can only attack Randy from 1 tile. Rutger gets there at like 11/9 (less with Bartre Energy Ring/Wings), which is kinda pushing it and I'm skeptical Fir can do any better without Boosters at that point.

You're correct there's only 1 Speedwings up to that point, I got mixed up. So yeah he can only take one by Gelero, in which case he isn't doubling realistically. The more I think about it, the more my problem with promoting Dieck first is less to do with Henning and more to do with how we have to ALSO pass up on training and promoting Fir for him to be the next best choice for bosskilling. This furthers the inefficiency of using Dieck (the reliability difference of using a Swordmaster against Gelero and possibly in the future with enough effort is more than enough to make up for the monetary value of the second Hero Crest)

As for the Henning stats, it's not actually that difficult to get more units into the boss area since it's broadly more efficient to have Shanna just function as a constant Give > Drop bot across the gap whilst other units expend PP picking up units to give them to her, whilst everyone else runs to the throne as soon as they get over. But I did acknowledge the shared contribution for the bosskill means Dieck's contribution there is significantly reduced. I think you may have misunderstood the original premise of why people even entertained the idea of promoting him first though, because it's less specifically about bosskilling prowess and more about general unit strength and dominant period if he's promoted early. He basically misses this period if you promote him with the second crest and his higher base level works against him due to him not really benefitting much from the EXP flooded Isles enemies. The bosskilling thing is really more just icing on the cake that he's probably the next best choice you have against many of these bosses, for better or for worse.

That being said, I think that given he is so much worse at bosskilling, compounded with the fact he might not even get the Bosskill EXP a lot of the time due to having to rely on help from other units, he's basically going to turn into pure second string and can't really be expected to keep up with lategame bosskilling, so that does knock him down a peg I think. Granted, I don't think it affects his tier placement much because earlygame matters a lot, and you have to admit, it's at least fringe entertainable to utilise Dieck against Devias/Leygance/Henning and such, wheras promoting someone like Lott or Wade first is just completely nonsensical.

Edited April 22, 2017 by Irysa

[Colonel M]

Apologies on not responding. Ill look over everything and Ill post some thoughts within a few days.

 

I cant do it atm, but reversing Percival > Rutger. Rutger will go back above Percival later today.

[XeKr]

If it matters, generally the binomial growth calcs in my previous posts are for at least # procs, not exactly # procs, even if I forgot to note that specifically. This treatment works better for benchmarks because obviously 20 Spd doubles Gelero just as well as 19. Basically just pretend I said “at least” where appropriate. ;p

On 4/22/2017 at 2:51 PM, Irysa said:

My point is you're not going to be buying more to use for post C21 anyway unless you have sub 4k gold and can't make up the last stretch, because it's not really that useful.. It's not a question if if you buy more or not.

I don't think there are any strats that are not just pure LTC specific that actually rely on Bolting to hit, just Bolting serving as a reliability augmentation.

Tome accuracy is a thing yeah but there arent so many uses for it that it's a huge deal. 5 seems really small but when there are only about 5 places where it'll have a more than marginal impact, it doesn't matter too much.

I edited a bit to clarify the previous post before you posted, as to not focus on the shopping part as much. Bolting just seems like a marginal thing since it has 5 uses (including misses) and accuracy isn’t that great. It’s like 40 effective hit after the throne and you then compare skill/spd/luck. I guess we can sim the 20Ix Lilina case if we’re interested enough though I would prefer to simplify it a bit (or at least nail down in pseudocode what the best strat plus contingencies are).

On 4/22/2017 at 2:51 PM, Irysa said:

I think at this point someone just has to commit to actually doing an efficient playlog of the game and trying to train then utilise Fir for lategame bosskills. I don't have any plans to do something like that anytime soon.

actually playing the game? pfft.

Agreed. Even a non-rigged run using Rutger (with some of the insights of the recent ltcs, and posts itt, even if not the exact strats) would go a long way.

My “casual” playstyle is basically expected tc inspired, though I probably go for the more reliable strats rather than the more speedy ones (and I sometimes do random things like get all major items). It’s probably impacted by the fact I often have animations on and such so eh. I tend to go through various FE games and I recently did FE6 but I might do another Fir run (with less extraneous objectives, and probably no Sue/Lilina) when cycling back to it. fe6 is the next echoes game rite IS?

In “serious” optimized runs I tend to play resetless low turns nowadays. Kinda like a speedrun but more focus on adaptation and impromptu routing, less on execution. ofc no fixed rng (though sometimes fixed growths if it’s possible). Plus I don’t have time for big gaming sessions, so I like having breaks and time to think about the best approaches to a given situation. But it can still get pretty stressful. >_>. And FE6 in particular isn’t really amenable to this.

On 4/22/2017 at 2:51 PM, Irysa said:

You still need 16 Str and 22 AS and A Swords to perform notably better than Miledy given you can only attack Randy from 1 tile. Rutger gets there at like 11/9 (less with Bartre Energy Ring/Wings), which is kinda pushing it and I'm skeptical Fir can do any better without Boosters at that point.

What’s special about 16 Str and A swords? I’m probably missing something obvious. Anyways for reference: 

16 - 9 base - 2 promo = 5 procs at 30%
11/9 Rutger has a 48.5% chance of at least 16 Str in 15 levels, 81.3% with Energy Drop

22 - 15 base - 1 promo = 6 procs at 50%
11/9 Rutger has a 84.9% chance of at least 22 Spd in 15 levels. Assuming no cap ramming. 

Laterpromo ??/1 Fir has very high chances of 22 Spd due to cap plus 2 promo, which is further improved by a few promoted levels (too lazy to calc atm)

16 – 9 base – 3 promo = 4 procs at 25%
17/3 Fir has a 69.4% chance of at least 16 Str in 18 levels, 96.1% with Energy Drop

I’ll run the sims with different Str (and Miledy) another time.

Random updates: Adjusted things to take into account potential Braves or theoretical weapons with even more consecutive hits (and things like Adept/Astra if the nested loops want to work).  Also reduced the number of digits to maybe better represent the uncertainty in the estimated probability, and wrote some stuff to print the results for easy copypaste. Checked results more against some simple cases (with 100% hit and such), so hopefully the math is okay(?).

Side note regarding the “rounds” thing, essentially what it tells is on average how long it takes until victory (boss kill’d) or death (reset). afaict, this is not the same as the number of rounds it takes to win on average, which perhaps we are interested in if we mainly care about the successes. This latter number can be calc’d, but for some reason it runs really slowly if tracking the outcomes that way (basically need to match up specific wins with the number of rounds it took in that iteration, instead of just simply counting them up as they occur), so eh. So for now the convenient number is rounds until success or reset combined. And I think that does still tell us something, since losses remain important to note (because we need to retry then). 

So it should be “trivial” to run whatever numbers we want. Hardest thing is getting stats from the movie files or bases/growths and inputing them. >_>

On 4/22/2017 at 2:51 PM, Irysa said:

As for the Henning stats, it's not actually that difficult to get more units into the boss area since it's broadly more efficient to have Shanna just function as a constant Give > Drop bot across the gap whilst other units expend PP picking up units to give them to her, whilst everyone else runs to the throne as soon as they get over. But I did acknowledge the shared contribution for the bosskill means Dieck's contribution there is significantly reduced. I think you may have misunderstood the original premise of why people even entertained the idea of promoting him first though, because it's less specifically about bosskilling prowess and more about general unit strength and dominant period if he's promoted early. He basically misses this period if you promote him with the second crest and his higher base level works against him due to him not really benefitting much from the EXP flooded Isles enemies. The bosskilling thing is really more just icing on the cake that he's probably the next best choice you have against many of these bosses, for better or for worse.

That being said, I think that given he is so much worse at bosskilling, compounded with the fact he might not even get the Bosskill EXP a lot of the time due to having to rely on help from other units, he's basically going to turn into pure second string and can't really be expected to keep up with lategame bosskilling, so that does knock him down a peg I think. Granted, I don't think it affects his tier placement much because earlygame matters a lot, and you have to admit, it's at least fringe entertainable to utilise Dieck against Devias/Leygance/Henning and such, wheras promoting someone like Lott or Wade first is just completely nonsensical.

I did briefly mention his other combat roles and his earlygame and how his tier position can still be warranted and such. Just in a world where we weigh more efficient teams more than less efficient teams (though not exclusively considering them), Deke probably shouldn’t get much credit for cases with the first Hero Crest. How much depends on how lenient one chooses to be, though the numbers suggest we would need to be quite generous and handwave several turns (consistency in analysis is what matters, not what’s done in this specific case. And this is indeed a factor for other contexts like theory!Fir.). Obviously Lott/Wade have negligible chances at assisting certain bosskills, so those cases are so inefficient they get almost no credit at all.

The arguably cleaner way is to consider nonspecific team compositions uniformly, regardless of their efficiency, though technically this equates stuff like solo runs to optimal, synergistic team clears. However, I think that treatment is generally disliked because under the constraints of this list, we value efficient play in general, not just the relative efficiency in a character’s presence/absence (there’s some nuance to tease out there).

I don’t think there’s actually much that is contentious here, I just picked the Henning case as a proof-of-concept to demonstrate how the analysis works (and I was vaguely curious what the numbers actually looked like). If someone says Deke is a good bosskiller, or that Rutger’s bosskilling can be mostly replicated by Deke, we can provide numbers to support/challenge that claim.

Edited April 24, 2017 by XeKr

[grandjackal]

On 4/23/2017 at 11:54 PM, Colonel M said:

Apologies on not responding. Ill look over everything and Ill post some thoughts within a few days.

 

I cant do it atm, but reversing Percival > Rutger. Rutger will go back above Percival later today.

YO COLONEL IT'S BEEN FOREVER WHAT'S GOOD

Real talk though, I feel like I would want to have a nice docket of discussion. As someone who plays cart and doesn't have the luxury of savestates most of my thoughts are theoretical in nature. However, I feel like a lot of low turn count or efficient play in FE6 is more unrealistic in some sense as opposed to other games. The nature of inaccurate weapons and enemies having actual stats generally makes uncertainty a fact of life in FE6. Strats are made inconsistent on the basis that things will inevitably go wrong in any given chapter. Makes FE6 actually fairly difficult to discuss at the finer level. So, I only really have theoretics to offer. Others can take the discussions from there, even if there is nowhere. Old habits die hard.

-A god tier. It's not like in FE8 with Seth being an untouchable golden god. In FE6, it's more the idea that there is no such thing as an efficient game without these characters. Like Marcus. Marcus is hardly a god, but you aren't realistically running this game without him. Earlygame is many many many turns extra without him, nevermind the safety and dependency he offers in doing his job. Rutger. He's basically the one true boss killer in a game that demands you have one. However, this is where things get tricky. Saul could end up here on the basis of being able to use the Warp Staff as soon as you get it, or Ellen (man that's weird to think). Niime could be here, while someone like Miledy wouldn't. Miledy isn't a necessary character, she is merely the best all around character. If that makes sense. Maybe name it something other than god tier? The name would be misleading in FE6.

-Why exactly is Astor in high? Thieves are stupid to rate, but I wouldn't say he's that great. Certainly not better than important cavaliers.

-Why is there a tier difference between Echidna and Bartre? If it's about performance, Echidna's shitty luck and poor survivability makes her iffy in practice vs on paper. Bartre could be blind, the fact he can attack and not randomly explode shouldn't be ignored. Also, bows. Bows are a better accurate side arm than swords. Bows let Bartre actually do something in Sacae, for example. It's not like Echidna outperforms him in Ilia either.

If it's about routes, I have doubts on Echidna's route being better. sure, killer axes, cute. Only Echidna or Lott at the time can use them realistically (and Lott doesn't even put them to particularly good use). By the time you have other people who can use them, they become available. Even then, I could argue they are frivolties. Is spending a good portion of money to say Echidna has 20 more crit for some extra chapters more important than say an extra pair of boots later down the line? Bartre doesn't depend on you buying specialty weapons, he can mostly function on iron and whatever nifty shit you find. Secondly, while Echidna gives you killer axes and 3K extra, Bartre's route supplies you with an Energy Ring that doesn't exist in Echidna's route. We've been talking about Rutger hitting benchmark strength numbers, and an Energy Ring removes all our worries. Hell, it could let Fir be the actual replacement people claim she could be. Gonzales being 5 levels lower is basically pointless with his stats, and I would value the fact he starts near the action (and essentially a chapter earlier) to get a better head start is more meaningful. You also get Tate and Klein a chapter earlier for what it's worth, and I would argue Elphin is better than Lalum. I would take extra beef over a minor amount of speed. In the least, I would say Bartre and Echidna shouldn't be seperated by an entire tier.

-Lot being under Fae and Lugh boggles my mind. Fae is arguable in the sense that Fae is an immediate dragonslayer that quickly becomes hard to kill despite her very limited performance time. That arguably makes sense to me. Wielding slayer axes early vs the cavalier parade of chapter 4, or just being an axe dude in general in earlygame is just great. Lugh on the other hand has the contribution power of "vaguely better than Wolt". Wolt could not exist past chapter 1 and you wouldn't notice. Being the step above that is not exactly powerful performance. If we're just funneling kills into him (kills that Lot is earning honestly by himself), then how much do we need to funnel into his dork mouth to make him functional? You need a lot of speed or a lot of explosive power for western isles. Dieck is not fast enough for the isles. For Lugh to actually get to that point, he basically needs to be level 15. 14 levels in 6 chapters doesn't strike me as realistic. Hell, if we're being realistic he's probably spending all that feeding not really outperforming base Lillina. By the time he does get over his sucking problem, He's still not exactly a wonderful combat. His ass is still fairly kickable, namely. All this effort to have a basically ok lategame character vs Lot, who is a fairly ok character that didn't necessarily require funneling. 

-Honestly, Lugh shouldn't be that much higher than Lili. Arbitarily right above her and maybe Garret on the basis of existing in forced chapters where he's still relevant chip damage. Little bastard's been overrated forever. 

-Gonzales being lower mid hurts me, but requiring secret books to be the god he is is a thing in a game where you can buy boots. I just...want him to not be in lower mid. Please?

-Geese should be below Douglas. Geese is basically never competent, and requires effort to continue being never competent. Hugh costs you boots but at least he doesn't suck, and Douglas can at least not die to physical things which is a use. 

-I am the last person to talk about armor knights, but honestly Bors should be in utter shit with Barth and Wendy. Bors is in the same boat as Wolt, but worse. In fact, you're better off not using him in chapter 1 because to do Wolt's job, he has to die in the process. You are utter shit when it is better NOT to use you when I have no choice in the matter of your fielding.

-And yes, Sophia and Wendy should swap. Sophia is literally impossible to use. I don't care that leveling to 10 and promoting lets you become a healbot, getting Sophia to 10 is an impossible task. Wendy can at least realistically attack something and A. hit it, and B. not explode. Sophia's ranged attack has no value because it is the equivelant of shouting "boo" at her enemy. Wendy sucks, but she at least vaguely resembles a unit if you look at her at the right angle and light.

[-Cynthia-]

Echidna doubles most enemies upon joining, Bartre can't double anything- it's a pretty huge offensive difference. Echidna's luck isn't great, but I didn't notice notable crit rates on her aside from Killer weapons (note that Tate has the same base luck, but this hasn't been brought up with her). 

 

[grandjackal]

5 minutes ago, -Cynthia- said:

Echidna doubles most enemies upon joining, Bartre can't double anything- it's a pretty huge offensive difference. Echidna's luck isn't great, but I didn't notice notable crit rates on her aside from Killer weapons (note that Tate has the same base luck, but this hasn't been brought up with her). 

 

-Echidna doubling is nice, but the strength gap is fairly massive. Let's take an enemy with 10 defense. Echidna with iron is packing 21 mt, and Bartre packs 30. Despite doubling, she does a grand total of 2 damage extra. 10 def is not unusual around the time you hit the bridge, and every defense over she loses 2 damage to Bartre's 1. She has to make up the difference using more expensive weapons. Granted that she can, Bartre doesn't require expensive items to be as good as he is. In fact, he can make use of generally less demanded items like steel weapons, of which you probably have some lying around from Dorothy and Ward, or other less demanded weapons.

As for luck, A. 1 more base luck than Zealot is luck issues, and B. Tate does indeed have a bad base that she recovers from with fast leveling speed and 40% growth, two things Echidna doesn't have. Echidna can't really fight promoted enemies without risking displayed crit, and even occasionally normal enemies as we start getting closer to lategame. 6 base with 20% growth is ignorance. Any random mercenary has a chance of shredding her. 

[-Cynthia-]

I think you're really overrating the cost of Killer Axes here given the amount of money we have at this stage of the game. Like are we not supposed to buy anything because it might be an 1/8 of a Boots for the last few chapters? That's like 15 chapters away from Echinda's jointime.

[Interdimensional Observer]

2 hours ago, grandjackal said:

Wendy can at least realistically attack something and A. hit it, and B. not explode.

The value of a ranged attack is that you can hit something with range 1 only and not be killed. Flux Sophia > Javelin Wendy. And more to the point, Wendy doesn't have durability. While this tier list is NM over HM, she is taking 10 damage from an HM Iron Lance Soldier in her joining chapter, if we assume -2 Str on NM, that is just barely 3HKO, not even if she gets hit with a Steel once. And if anything doubles, then there thar she dies!

Debating the worst of the worst hardly goes anywhere. I'd say Sophia is superior purely on the grounds that her long-term contribution if trained would be more unique than Wendy's. Wendy's gotta compete with far more lancers than Sophia with dark magic users or mages in general.

[Refa]

Wait, I thought this tier list WAS HM.

[grandjackal]

Well in that case, lets look at that question from a different perspective. Why does Echidna need killer axes? Why do we need Echidna to have Killer axes? Fight paladins? Bartre can just use the poleaxe, or shoot them (and not risk a crit in the process). Wyverns? Echidna would need to land a crit and a hit to kill them in the first place with her 24 mt. Hell, with the one free killer axe we get, Bartre is possibly more efficient with the one free one he gets (don't quote me on that though, those other two guys talking a storm here could prove otherwise). Pegasi? Bartre's potentially buff enough to one shot them with a melee weapon (a silver axe, of which few in the game even have the weapon rank to use, and Percival doesn't need it). 

Also, we are depending on 20s in crit to occasionally kill a scrub maybe every now and then (and probably needing to hit twice anyways). At that point, why even waste the money? Just use iron and give the kills to other people. Not like we need Echidna to go on lategame, and it doesn't cheapen her contributions.I'm just saying even if that is the advantage that gives her the edge over Bartre, I don't think she's so outperforming him as to deserve being in a different tier. 

@Interdimensional Observer I will take Wendy being plausably usable over Sophia dying when looked at funny and trying to 4HKO people at best with 74 base accuracy. Also, unique? You have to train her to 10 just to become an E rank staffbot. Hardly unique.

Also, Wendy doesn't have to do this while avoiding Purge bishops. I know arguing who the worst character is is dumb (FE10 died for a reason), but implying it's not Sophia is crazy pills.

[Xanaxian]

@Refa It is HM, but what's feasible in HM is often even more so in NM. So generally there can be some overlap if something works in the former mode.

[Refa]

Rutger and Percival are completely unnecessary on NM, for starters.

[Irysa]

On 24/04/2017 at 8:30 PM, XeKr said:

I edited a bit to clarify the previous post before you posted, as to not focus on the shopping part as much. Bolting just seems like a marginal thing since it has 5 uses (including misses) and accuracy isn’t that great. It’s like 40 effective hit after the throne and you then compare skill/spd/luck. I guess we can sim the 20Ix Lilina case if we’re interested enough though I would prefer to simplify it a bit (or at least nail down in pseudocode what the best strat plus contingencies are).

Yeah I understand. It mostly comes down to whether oneshotting one or two Wyverns, denting a few Paladins and Heroes, and Bolting chipping a boss or occaisonal enemies someone might miss or w/e adds a notable amount of reliability. I feel like it's a better payoff than most others in the lower tiers are doing for you though, easily.

On 24/04/2017 at 8:30 PM, XeKr said:

What’s special about 16 Str and A swords? I’m probably missing something obvious. Anyways for reference: 

It's the stats required to have a chance of getting him down in a single round of combat. Remember that by Miledy can be safely engaging Randy repeatedly on PP and EP and gets him down in 1 hit and 1 crit at base (or just like 4 or 5 hits depending on throne healing). You can use the Killing Edge to doublecrit him too, but the Str requirement doesn't change (I think I was getting mixed up with other bosses getting Physic'd or something) but its important to keep in mind that Rutger faces a bit of displayed crit (even on average, Goddess Icon to Rutger hype?), and is 3HKO'd, and if you do try to get him up there you're having to either have Miledy fly back to grab him (wastes time compared to having her fight Randy) or sending another flier to carry (Miledy probably had to clear the way, means you can't use that flier for grabbing items or carrying someone else like Roy) which means that I'm pretty sure the SM has to prospectively be able to make up for this efficiency drop by being considerably better at beating the boss than Miledy is.

14 hours ago, grandjackal said:

-Why exactly is Astor in high? Thieves are stupid to rate, but I wouldn't say he's that great. Certainly not better than important cavaliers.

Nets a huge amount of gold via stealing stuff over the course of the game, aka more Boots. He also has above average combat in the Isles and isn't so frail as to die to LRT in later chapters without training.

 

14 hours ago, grandjackal said:

-Why is there a tier difference between Echidna and Bartre? If it's about performance, Echidna's shitty luck and poor survivability makes her iffy in practice vs on paper. Bartre could be blind, the fact he can attack and not randomly explode shouldn't be ignored. Also, bows. Bows are a better accurate side arm than swords. Bows let Bartre actually do something in Sacae, for example. It's not like Echidna outperforms him in Ilia either.

Regarding this and other comments you were making, Rutger has pretty crap luck too man and we're not docking him for that. The reality is that most units Echidna is fighting pull relatively low hitrates on her naturally anyway, same as Rutger, and she doesn't have to engage them at 1 range either. Bartre also has much worse skill, and you're hyping up accuracy and reliability, where Bartre's only accurate weapon is an Iron Bow. It suffices vs most enemy types but do't get ahead of yourself. Bartre also starts to get doubled later in the game. I'd say Echidna does outperform him in Ilia given Echidna can ORKO enemies on EP there wheras Bartre can't. In Sacae Bartre is getting doubled and can barely hit the Nomads.

And no, Bows are not a better side weapon than Swords. Echidna can use the Light Brand, Armorslayer, Wyrmslayer, Killing Edge, all of which have good utility and give her Weapon Triangle control. Bartre has Iron/Steel/Short Bows. Chipping Wyverns is nice but we could easily just deploy Klein or Igrene instead to oneshot them, or actually stand a chance of doubling them. Bartre's accuracy is also very questionable with Steel Bows, and it doesnt let him oneshot anything.

Neither Echidna nor Bartre require us to buy extra weaponry to make good use of them, and both Echidna and Bartre like to use the Killer Axe so that logic doesn't make sense. Echidna also has an existant Res Stat so she can dodge Status Staves and stuff with a Barrier Boost, something Bartre has to gamble on.

And finally, Bartre route does get you an extra Energy Ring and Angelic Robe (and have easier bosses), but it takes longer (for both maps, assuming full recruitment, gets sigificantly worse if not assuming full recruitment), Klein and Tate get lowered stats (due to the way HM bonuses work), you have to deploy one of Lott or Wade to get useful items in the village map, and to even get the Energy Ring you have to kill the Iron Ballista, which neccessitates rigging a dodge on a flier to kill it before it kills any civilians.

14 hours ago, grandjackal said:

-Lot being under Fae and Lugh boggles my mind.

Fae has unique utility for some of the more difficult maps due to her ability to burn Siege Tomes and act as a Status lure. Lugh only really has 2 Chapters less than Lot of contributions and his chip is reliable and fairly large for that point in the game in Chapter 4, and he doesn't take counters or get OHKO'd, wheras Lot gets doubled when using the Halberd against most Cavs (borderline ORKO'd, would need to be mended or protected if he misses). Lugh is also pretty welcome in Chapter 7, the hardest in the game, because he has decent Chip on most of the bulkier enemies again.

I agree with you that Lot is a little underrated and never truly bad, just never particularly good (bar vs Soldiers), but Lugh has niches early on that make him valuable wheras Lot's biggest claim to fame early is mostly just doing good vs Soldiers and existing without being trashy. Lugh does need a significant amount of favoritism to start making bigger contributions, but Lot is going to be unpromoted for quite a bit too anyway, so he's not exactly doing great either, especially since he's not as easily rescuable by infantry units compared to Lugh. Lugh eventually gets to ORKOing mooks status, Lot basically never gets there and will always be relying on crits (which is not actually terrible mind, given it's how most units in this game get ORKOs)

14 hours ago, grandjackal said:

-Gonzales being lower mid hurts me, but requiring secret books to be the god he is is a thing in a game where you can buy boots. I just...want him to not be in lower mid. Please?

Can't be rescued after promotion except by unpromoted Shanna and Shin, both units who have better things to be doing. He also needs that promotion to stay relevant. He's garbage.

14 hours ago, grandjackal said:

-Geese should be below Douglas. Geese is basically never competent, and requires effort to continue being never competent. Hugh costs you boots but at least he doesn't suck, and Douglas can at least not die to physical things which is a use. 

Geese can break walls in Chapters 11L and 12 due to Brave Axe access (although admittedly one will have Echidna or Bartre if you were waiting around anyway at that point so 12 doesn't really count). But breaking a wall in 11L is better than pretty much anything Mr. 5 move unrescuable not that accurate does.

14 hours ago, grandjackal said:

-I am the last person to talk about armor knights, but honestly Bors should be in utter shit with Barth and Wendy. Bors is in the same boat as Wolt, but worse. In fact, you're better off not using him in chapter 1 because to do Wolt's job, he has to die in the process. You are utter shit when it is better NOT to use you when I have no choice in the matter of your fielding.

He gets a village in Ch1, can help with the Armorslayer trade chain and possible rescuedrop to the Fort in Ch2, helps get Lugh in Chapter 3. Better than Wendy or Barth by that merit.

Edited April 26, 2017 by Irysa