FE9 Tier list v3

[GreatEclipse]

I consider that a feature, not a bug. I do believe Mia can contribute more towards completing the game than Haar even though Haar is considerably better in their shared availability. Mia has many chapters, including several routs, during which she can contribute and Haar cannot. It's not an easy comparison (many chapters of okay performance vs. a few chapters of great performance), but it's a hell of a lot more meaningful than presuming that Mia will rarely be used because Titania, Oscar, and Marcia are unilaterally stomping the game.

You may have no problem accepting that, and I applauded you for it. However, it is unlikely that this will be the case for many other contributors. Rolf > Bastion should not be controversial so long as Rolf creates a (small) gross positive for the entirety of his existence, and is not losing badly for as long as both exist. You could argue he slows you down if you try to raise him, but you have no more reason to assume he is fighting alongside a team full of awesome mounted units than you do one full of Armour's and Mage's, so the most you could say is that Bastion is better whenever the (apparently randomly selected) team in place before he joined was full of high-move units.

1) Titania is obviously more valuable than Soren no matter what the rest of the team is. However, your general point it valid. How valuable Jill is depends substantially on whether Marcia is deployed and trained. If Marcia is always assumed to be trained and in play, Oscar > Jill seems a likely conclusion. If Marica is never in play, Jill > Oscar is even more likely. If Marcia's likelihood of being in play and/or trained is not assumed, Oscar vs. Jill is an unusual and challenging comparison. Seeing as tier lists should generate meaningful discussion, the relevant question is whether efficiency runs where Marcia isn't trained are meaningful. I believe they are.

The problem is you have no way of knowing what the chances are that any particular unit will be in play if you assume everyone is equally likely to be used, save assuming it is selected at random by an RNG. Obviously, this is a highly restrictive and unrealistic assumption, certainly not when the primary goal is archiving a low turn count. Remember that I said both standards were logically consistent, not that they accurately represent the game as it is actually played.

2) Concerning the first nine chapters, where all units have free deployment, I'm not sure if it makes sense to consider runs where they aren't used. I'm not sure, therefore, whether it is within the scope of this tier list to consider runs where Titania is not used in chapters 1-9. I'd be willing to submit that units with free deployment are assumed to be used (no matter what unit is being considered).

Assuming that units with free deployment receive some kind of use is reasonable, but that they are used in a way independent of more long-term plans, is not. Would you have the list assume completely optimal strategies for the first 9 maps every time, with Titania necessarily receiving a huge portion of the kills, and the player only allowed to select his units and battle plans freely from that point onward?

Point taken. If we can't assume that Marcia will be trained and used, Soren might have either 2 or 8-9 turns of combat in C12. If this were any other FE game, this would be a bigger problem. As it is, plentiful Bexp can negate the importance of an extra level or two or Cexp here and there.

I would think it would come out to more than that if the optimal units are capable of saving that many turns. Even then, other games do exist, and the issue will need to be dealt with if this standard is to be relevant in any other context.

This doesn't make any sense. The tier list's criteria is efficient play. At a bare minimum, this must constrain how we assume chapters are played. Given a set of deployed units, the tier list player will use those players to efficiently complete the chapter and game. There's obviously some room for debating the applied meaning of the phrase "efficiently complete", but it certainly excludes taking additional time/turns or sacrificing reliability for gains that don't play for themselves later in either time/turns or reliability.

You are correct that the list must constrain how chapters are assumed to be played, it's just that one constraint (using the best units for the job) is being relaxed while another (using said units in the way that produces the fastest clear) is not, despite the fact that both are equally counter-productive to achieving the explicit goal of the list. You will likely respond that relaxing the first goal serves a worthy purpose, by allowing more units to enter consideration and broadening the scope for discussion, and I would agree with you. However, an almost identical argument could be made for relaxing the second condition as well. Allowing unconventional yet perfectly viable strategies that do not produce the lowest turn count possible with the team under consideration would add even more factors to the discussion list, as units that can make a larger contribution with more non-orthodox approaches would receive a boost to their relative value.

In either case, you are no longer judging units entirely with low-turning in mind, hence the unsuccessful attempts to draw a line between the varying degrees of "efficiency" must again be confronted.

Well, the tier list assumptions could be set up in such a way that all deployment slots will be filled, but the unit that fills each deployment slot is random (with the obvious exception of the unit being tiered).

Again, units are not selected at random when the game is actually being played, whether the team is an optimal one or not, and being forced onto the field tells you nothing about how much they will be used or in what way.

Edited July 8, 2012 by GreatEclipse

[Snowy_One]

The problem is you have no way of knowing what the chances are that any particular unit will be in play if you assume everyone is equally likely to be used, save assuming it is selected at random by an RNG. Obviously, this is a highly restrictive and unrealistic assumption, certainly not when the primary goal is archiving a low turn count. Remember that I said both standard were logically consistent, not that they accurately represent the game as it is actually played.

Probably doesn't help but waaaay back when what I did when on the tier lists was assume that players team was filled with a load of generic units unless specifically mentioned otherwise. The team would consist of 1 Ike, 1 X unit, 1-5 fliers/mounted units, 1-2 sages, and 1 healer with foot soldiers filling in any missing slots at a 1-1 ratio. The team would follow obvious paths and plans, try to net as much reward as possible, and would come it at the BEXP turncount limit (highest possible turncount without losing BEXP) and the measure how well the unit did from there. Obviously things are different here and now, I know that well enough, but maybe something similar is needed to measure potential. Otherwise the list is going to end up being simply a singular 'optimal' team with any units not on said team being ranked by how little they slow said team down.

[Wolfric]

.. Wow.... This tier list is crushing my mind...

Is there... Some sort of post with the information that leaded to this Tier List summarized?

[-Cynthia-]

Units with high mobility and availability are favored because they allow for faster turncounts with good reliability. That's the short version.

[Lord Raven]

I'll write a full summary (based on my perception) soon. It'll provoke a lot of shit with aku chi but that's okay..

[Paperblade]

The problem the second standard (Formally known as the gross system) ran into was that units who gave a mediocre performance from chapter 1 all the way to final come out better than pre-promotes who join around 3/4th of the way in and are amazing for the whole of their existence, because to account for what could have been accomplished had said mediocre growth unit been replaced with someone else would be to invoke the first standard (Known as the net system). Someone would need to be so bad that trying to use them actively worked against you in absolute terms for availability to not be an unconditional benefit, and that is unlikely to be the case for anyone in this game at least.

How does the idea that each unit has an equal chance of being used (which is technically from a probability perspective) require us to the use the gross system?

A second, much less discussed, problem is that if everyone is equally likely to be in play, then no good reason exists to assume that the lowest turn strategies are being used, because those are premised on the best units being in play. This does not result in units being impossible to tier, because you can still put Titania above Soren for having the potential to help achieve a lower turn count, even if it is up in the air whether that particular turn count with be achieved. However, when comparing Soren to, say, Ilyana, you no reason to assume (Or not to assume) that the preceding chapters were a Titania stomp, because the player, despite the apparent assumption that he/she desires clearing in fewer turns than more whenever possible, is under no obligation to give Titania even a single kill for the entirety of the game. This has profound implications for how fast units in Soren's position can be expected to grow, and it creates a huge number of other problems as well. For example, if an optimal team can clear X chapter in 6 turns with optimal tactics, and a sub-optimal team can clear that same chapter in 8 with equally optimal tactics, and both outcomes are considered acceptable by the list, then why should an 8 turn clear with the optimal team, using sub-optimal tactics, not also be considered an acceptable outcome? Attempts have been made to draw a distinction between "Normal Efficiency" and "Maximum Efficiency", but any dividing line that is drawn is sure to be completely arbitrary.

To be honest I do not see the problem with this.

[Lord Raven]

That changes no arguments at all, though

[GreatEclipse]

How does the idea that each unit has an equal chance of being used (which is technically from a probability perspective) require us to the use the gross system?

To apply opportunity cost to deployment would be to imply that units who contribute most to the goal being pursued are more entitled to a unit slot than others that do not, which is contradictory to the assumption that units are being selected are random. To give an example, if Oscar contributes 84 utility to the team when selected for the 8th available slot, and Rolf contributes 16, then a list that assumes a conscience choice to maximize utility by the player must logically assume Oscar will be selected every time. The only coherent way to make both equally likely to fill that slot is to assume both are equally entitled to the resource, in which case the only fair way to decide how it will be distributed when scarce is through the lottery. Indeed, punishing Rolf for the cost of his deployment should he be selected by the hypothetical RNG would make no sense. That presupposes you could deploy someone else who was not also selected and therefore already on the field, and that would invalidate the whole point of running the lottery.

To be honest I do not see the problem with this.

I assume you mean you do not have problems with the fact that it compromises the stated objective of the list, as the objection that it does not was already responded to. In that case, it only creates a problem in deciding just how fast the player is required to clear.

Why would none of the issues raised affect placement, Lord Raven?

Edited July 11, 2012 by GreatEclipse

[Snowy_One]

To apply opportunity cost to deployment would be to imply that units who contribute most to the goal being pursued are more entitled to a unit slot than others that do not, which is contradictory to the assumption that units are being selected are random. To give an example, if Oscar contributes 84 utility to the team when selected for the 8th available slot, and Rolf contributes 16, then a list that assumes a conscience choice to maximize utility by the player must logically assume Oscar will be selected every time. The only coherent way to make both equally likely to fill that slot is to assume both are equally entitled to the resource, in which case the only fair way to decide how it will be distributed when scarce is through the lottery. Indeed, punishing Rolf for the cost of his deployment should he be selected by the hypothetical RNG would make no sense. That presupposes you could deploy someone else who was not also selected and therefore already on the field, and that would invalidate the whole point of running the lottery.

But doesn't that also ruin the point of a tier list as well? We can find the means, the manner, and the method to find the nine or so units who contribute the most utility, the lowest turncounts, and such, and then we have no reason to ever remove any of them from the team as doing so results in a sub-optimal team that can't achieve as low a turn count. Why does it matter if Soren is better than Ilyana after all if we have to kick Oscar from the team to do it? What if one of the two is definably better than the other but the rest of the 'optimal' team is simply so STRONG as to make the difference moot? OH NO! ZIHARK IS BETTER THAN MIA AND... neither of them... really help out on the list because Marcia, Jill, and Titania are gonna rapid-clear the entire chapter anyways... Might as well have Rolf be as good as them since they're all sub-optimal and the mega-mounties are gonna blast through the chapter anyways.

Edit: Just in case, what I'm getting from this is that certain units have a high-enough opportunity cost to be assumed to be always deployed. If I have misunderstood, I apologize and will delete this.

Edited July 11, 2012 by Snowy_One

[Paperblade]

To apply opportunity cost to deployment would be to imply that units who contribute most to the goal being pursued are more entitled to a unit slot than others that do not, which is contradictory to the assumption that units are being selected are random. To give an example, if Oscar contributes 84 utility to the team when selected for the 8th available slot, and Rolf contributes 16, then a list that assumes a conscience choice to maximize utility by the player must logically assume Oscar will be selected every time.

This is circular reasoning. The determining of a unit's utility is how useful they are if they are being used, but if you say "Well Oscar is more efficient than Rolf so Oscar is going to be fielded for our maximum efficiency team therefore Oscar is more efficient" you are being fallacious. Utility is an ends, not a means.

The only coherent way to make both equally likely to fill that slot is to assume both are equally entitled to the resource, in which case the only fair way to decide how it will be distributed when scarce is through the lottery.

The "lottery" in this case is a player who has no preset biases as to which character is better, and is thus equally likely to pick any combination of units for his team.

Indeed, punishing Rolf for the cost of his deployment should he be selected by the hypothetical RNG would make no sense. That presupposes you could deploy someone else who was not also selected and therefore already on the field, and that would invalidate the whole point of running the lottery.

I think you and I disagree on what negative utility is. There are arguments to be made that sometimes fielding a unit can be worse than just fielding no one at all, because attempting to use them causes them to get in the way. That's the negative utility I am talking about.

For example, Rolf's poor defenses and bow lock would cause him to eat enemy phase attacks that would otherwise be suicided onto melee units capable of taking the punishment and retaliating for possibly lethal damage), which hinders your goal (in this case, LTC). Even if you argue "well if you use better tactics to block him off," that is still a disadvantage because the player is being forced to be less flexible to make up for Rolf's shortcomings and his offense and advantages are not great enough to overcome this loss of flexibility.

Technically you can argue that you could have the unit sit in a corner and not do anything but that is not possible in some maps and in the others, it's effectively the same as not fielding them at all.

I assume you mean you do not have problems with the fact that it compromises the stated objective of the list, as the objection that it does not was already responded to. In that case, it only creates a problem in deciding just how fast the player is required to clear.

The objective of current tier lists don't mean anything because no one can agree on what the goal is, so they just selectively apply concepts to fit their biases. Thus I don't see a problem with it contradicting current tier lists.

[Hawk King]

Heres an easy one.

Ranulf > Largo

He joins earlier, has higher movement, and can be used for shove chains.

What significant contributions does Largo possibly have over Ranulf?

[GreatEclipse]

This is circular reasoning. The determining of a unit's utility is how useful they are if they are being used, but if you say "Well Oscar is more efficient than Rolf so Oscar is going to be fielded for our maximum efficiency team therefore Oscar is more efficient" you are being fallacious. Utility is an ends, not a means.

The (hypothetical) numbers 84 and 16 refer to gross utility, not net, and so there is no circularity. The fact that Oscar is a better candidate for the slot is already accounted for, so it is not as if I am simply fielding Oscar while leaving Rolf on the bench and then concluding, based entirely on the fact that Oscar contributed and Rolf did not, that Oscar is better. Rather, if the next-best option for the 8th unit slot has 70 utility, then you maximize the net utility of Oscar by fielding him and maximize it for Rolf by not fielding him. To field Oscar is a net benefit of 14, while fielding Rolf is -54. As such, when comparing someone in the position of Rolf to someone like Bastion, assuming Rolf is regularly fielded and trained before Bastion joins actively works against him compared to leaving him on the bench, because the contributions a superior unit could have made in his place are being subtracted from his performance every time (this is why the net system has repeatedly been abandoned, only for people to start running back to it after a brief experiment with the gross version). The only way around it is to wave the cost of deployment and assume everyone is equally entitled to a slot, which in turn requires a team selected at random for the distribution of slots to be fair.

Though I do feel kind of silly for putting so much thought drawn from real world political economy into this.

The "lottery" in this case is a player who has no preset biases as to which character is better, and is thus equally likely to pick any combination of units for his team.

Which produces the same result, from the perspective of the list, as assuming the whole team save Ike is selected by an RNG. There is no simply no criteria related to the actual performance of the units for deciding who is most likely to be used in that situation.

I think you and I disagree on what negative utility is. There are arguments to be made that sometimes fielding a unit can be worse than just fielding no one at all, because attempting to use them causes them to get in the way. That's the negative utility I am talking about.

For example, Rolf's poor defenses and bow lock would cause him to eat enemy phase attacks that would otherwise be suicided onto melee units capable of taking the punishment and retaliating for possibly lethal damage), which hinders your goal (in this case, LTC). Even if you argue "well if you use better tactics to block him off," that is still a disadvantage because the player is being forced to be less flexible to make up for Rolf's shortcomings and his offense and advantages are not great enough to overcome this loss of flexibility.

Technically you can argue that you could have the unit sit in a corner and not do anything but that is not possible in some maps and in the others, it's effectively the same as not fielding them at all.

I did say in passing that Bastion would still be better if fielding and training Rolf was counter-productive in absolute terms, I just doubted that it would be the case most of the time. The fact that Rolf cannot counter at close range certainly makes him less flexible than he would be otherwise, but teams with multiple bow-locked units are not inherently less flexible on the whole than teams without them, from my experience. Similar to how not every country can be a net exporter, not every unit can clear out a whole room on every enemy phase, and so if you replace someone like Mia with Rolf the team as a whole is likely to get in just as many counter attacks. You need to be fighting enemies with massive attack ranges in large, open areas for it to be a major problem, and the enemies that most ofter applies to are also either weak to arrows or can be countered by them.

[bottlegnomes]

When did Ike get so high?

[Zkirsche]

Soren has 2-3 more magic than Ilyana. I don't know what your definition of bad is, but that's not much of a difference, especially with forged Thunder tomes.

I do tend to have a habit of exaggerating, but Ilyana's stats are worse and she has no utility to make up for it. The lead is existant, as there are enemies from the get-go that Soren can double that Ilyana can not, and I don't assume spell forges because they are expensive and ultimately not worth it when we should be giving our forges to Oscar/Boyd/Titania that early on (as we can only forge once per chapter, the forge choice is ultimately important and they are doritos for it).

.....Do you assume that the player is only going to be using mounted units in this game? There's no problem fielding one 6 move unit and one 7 move unit in.. any chapter. Also, tier lists don't mean too much when it comes to who we're going to train. For example, Mordecai is an upper-mid tier when he's mainly useful early on and due to his Smite skill.

The majority of the upper mid and above units are there to be trained, and everyone from high and top are definitely there to be trained. Only Mordecai and Mist are not main combatants in the top few tiers.

What significant contributions does Largo possibly have over Ranulf?

Chapter 25 is apparently the only chapter Largo can do well in. The main problem with Ranulf is that he doesn't transform until turn 3 or so, so is out of action for half the chapter.

[Hawk King]

Ranulf can use a demi band. And if he doesnt he can shove until he transforms. In many LTC clears, Ranulf is needed to shove one of your mounts. Largo cant shove any mounts.

Largo has 1 potential Chapter to make contributions. Ranulf joins earlier, can make the same contributions in Chapter 25 that Largo can make, and Ranulf can be used in shove chains.

So.... Ranulf > Largo anyone??

[aku chi]

I don't assume spell forges because they are expensive and ultimately not worth it when we should be giving our forges to Oscar/Boyd/Titania that early on (as we can only forge once per chapter, the forge choice is ultimately important and they are doritos for it).

Chapter 13 is a fine time for a Thunder forge. A Thunder forge at this point is affordable. We've already had 5 iron forge opportunities at this point. We may very well have forges for every beorc combat unit at this point. C13 is a de facto rout, so if a forge enables Soren or Ilyana to ORKO where they would otherwise not, it's helpful. A Thunder forge here is arguably more valuable than an Iron Axe forge for Titania, who can use a Steel Axe to 2HKO most units at high accuracy (the extra +2 Mt from the Iron forge is very situational). A Thunder forge helps Soren more than Ilyana, however, because it can be used to get around his pitiful Str.

Chapter 25 is apparently the only chapter Largo can do well in. The main problem with Ranulf is that he doesn't transform until turn 3 or so, so is out of action for half the chapter.

1) Largo is pretty good in C27, where he can clear out the chokepoint Generals and help get some treasure.

2) Turn 3 (turn 2 if he's attacked) is not a terrible transformation time, but I don't see why we shouldn't consider Ranulf using the Demi Band. Muarim would have to be a very high level indeed to have a notable advantage over Ranulf, so Ranulf is often the best candidate for the Demi Band.

Ranulf can use a demi band. And if he doesnt he can shove until he transforms. In many LTC clears, Ranulf is needed to shove one of your mounts. Largo cant shove any mounts.

Largo has 1 potential Chapter to make contributions. Ranulf joins earlier, can make the same contributions in Chapter 25 that Largo can make, and Ranulf can be used in shove chains.

So.... Ranulf > Largo anyone??

Ranulf cannot make the same contributions as Largo in C25, but he is more valuable in other chapters, so I can see Ranulf > Largo.

[Lord Raven]

When did Ike get so high?

It's a similar (but not as extreme) argument to Celice; he actually needs to be really good for efficient clears, even though he has a lot of weak points (especially early on).

[Chiki]

20 mag Ike > Titania on the tier list.

[Snowy_One]

The lead is existant, as there are enemies from the get-go that Soren can double that Ilyana can not, and I don't assume spell forges because they are expensive and ultimately not worth it when we should be giving our forges to Oscar/Boyd/Titania that early on (as we can only forge once per chapter, the forge choice is ultimately important and they are doritos for it).

Correct me if I'm wrong buuuut...

Hand axes can't be forged until chapter 18, steel weapons until 14. Actual steel weapons, however, can be bought as early as chapter 8 though and a forged iron weapon has about the same MT as a steel weapon. While it's true a steel axe has 15 weight and a steel lance 13 weight, Boyd can wield a steel axe without a AS loss as early as level 15. Oscar at level 18. Titania at level 7. They won't need the steel for the weaker and faster enemies (which iron is more useful against and more durable) so it will only be the stronger enemies that the steel would be used against. Plus, shouldn't Titania be capable of killing most enemies with a normal iron axe anyways?

Basically, is there some huge denomination of mega-tough and speedy non-boss enemies that require Oscar and Boyd to be capable of using steel-level weapons at 0 AS loss before chapter 14-16? That's the only way I can see the demand being so great as to make it so that a mage outright cannot get a forged thunder (as that's the best non-steel forge).

Edit: IMO, Soren vs. Ilyana is quite simple. There is no such thing as negative utility and Soren joins earlier and does the same things Ilyana does. So if Soren's pre-Ilyana chapters are >= 0 than the only way Ilyana can beat Soren is if she can be better for her chapters. Her primary advantages are 1) Higher DEF (moot as neither is durable to be a front-liner. Ilyana *might take one more hit, but that's about it*) 2) More STR (moot for anything other than doubling with a siege tome) 3) Arguably better supports (doesn't really matter even if they ARE better). So, three advantages, three of them moot meaning Ilyana is only *slightly* better if at all during her existence. Point, Soren.

Edited July 16, 2012 by Snowy_One

[PKL]

Ranulf can use a demi band. And if he doesnt he can shove until he transforms. In many LTC clears, Ranulf is needed to shove one of your mounts. Largo cant shove any mounts.

Largo has 1 potential Chapter to make contributions. Ranulf joins earlier, can make the same contributions in Chapter 25 that Largo can make, and Ranulf can be used in shove chains.

So.... Ranulf > Largo anyone??

I agree with this. Demi Band Ranulf can shove/smite promoted Pallies and stuff, hes pretty good.

Edited August 6, 2012 by Ghost Marcia Drafter

[Dr. Silent]

I'm curious...why is Titania at the very top in a tier all by herself? Is it a joke or have I missed something? I've always found Oscar/Kieran to be much better...except in early chapters obviously.

[Chiki]

I'm curious...why is Titania at the very top in a tier all by herself? Is it a joke or have I missed something? I've always found Oscar/Kieran to be much better...except in early chapters obviously.

Because the earlygame chapters are the most important ones.

On the topic of Titania, do people support me making a new tier list? This is a little outdated for my tastes.

Edited August 8, 2012 by Aeine

[Ether]

I'm not really an active member of the tiering community anymore, but in the past, if you wanted to make a new list, you were free to, but people may choose to discuss the old one in it's stead depending on their thoughts on the deviations you've made, or something to that effect.

[dondon151]

On the topic of Titania, do people support me making a new tier list? This is a little outdated for my tastes.

or we could just scoot marcia and jill up to top tier and call it a day

[Anouleth]

I'm curious...why is Titania at the very top in a tier all by herself? Is it a joke or have I missed something? I've always found Oscar/Kieran to be much better...except in early chapters obviously.

Oscar and Kieran are a little bit better later on in the game, but Titania is far far better in earlygame.