Character Class Optimizer Calculator

[Owns]

I have been working on a builds 'optimizer' type calculator that can systematically and automatically suggest the best class for each character. It started as I was planning my team for a NG Maddening run, found myself kind of enjoying the process until I ended up where I am, with this calculator lol. So I thought I'd share some of the results to get some feedback and thoughts on my calculator. I don't know if something like this already exists so if you know of one please do let me know!

MINOR SPOILER ALERT: All non-dlc classes are listed, including unique story-related classes that only get revealed as the story progresses.

Let me start by explaining what this calculator does (or at least tries to do) and how it works. First it looks at the growth stats of a character and adds them to the growth stat of each class. In the sample images below, these are the values in the main part of the table listed under the stats. For each level of classes (Beginner, Intermediate, etc.) it highlights in yellow which class results in the highest growth for each particular stat, just an interesting visual aid. So far, this is purely raw data:

char growth (stat) + class growth (stat)

Next, I assign a rating for each class for that character. Here I've come up with a rating made up of two parts:

1. It evaluates how closely the growths for a class align with the character's natural growths. This part takes into consideration that some characters can be more flexible than others.

2. A value is also given to the character's strengths / weaknesses in skills that are certification requirements for each class.

The two parts are combined to gain the final Class Rating. Without getting into the specifics of how the metric is calculated, it can be thought of as:

[class-char fit] + [skills value] = [Class Rating]

Obviously this method will not be perfect. Firstly it is based almost entirely on stats. Meaning it ignores valuable non-stat considerations such as potential hero's relics, character unique abilities, crests, etc. It also does not take into consideration the base stats of a character. One nice thing about including a value for strength/weakness in skills however is that when a character is strong in a particular skill it usually means they will also unlock useful unique abilities down that skill line, or that there might be a useful relic they will be able to use. So I like to think that some 'wisdom' finds its way into the [skills value] part of the Class Rating.

That's about it. Below are some sample results from each house. If this calculator does its job well, then the higher Class Rating value, the better that class should be (theoretically) for the character. The way to understand the class rating is not that it is finding the best class for that character. Rather it is trying to identify the class that best 'fits' the character's strong suits. This is done by considering the organically strong stat growths a character has as well as the skills they are strong in or have a budding talent in. It says nothing however about how good the class itself is, what benefits outside of growth stats it might bring, or how the class might fit in a builds progression for the character.

Inevitably there will be a few eye-brow raising results. I don't think a calculator can ever supersede common sense and experience, but I think it does a not so terrible job as it stands. Oh I should also add that Class Rating values don't compare across characters. Meaning if one character has a higher rating for a specific class than another character, it doesn't necessarily mean they are better in that class. It could mean they are a better "fit" for the class, but not necessarily that they will perform better.

Thoughts in general? Any suggestions? Feedback? Do you know of a similar calculator / optimizer out there already? Is there an interest towards getting an improved version of something like this?

One thing I think can improve in the Class Rating is to include consideration for base stats as well, just haven't figured out yet how to best incorporate that. I think that alone would fix some of the oddities. Thank you for your time.

Clarifications:

1. This calculator does not consider the progression path of the character. Masters classes won't consider what you chose for the Advanced or previous classes, etc. Another area where common sense is needed.

2. Comparisons of classes are done only within the same class level. So the rating of say, a Paladin, is not compared to say a Wyvern Lord. Each class level's ratings are ranked separately. This is another area where common sense has to intervene.

If you want to CTRL+F jump to a character's results, here's the list in this sample: Byleth, Edelgard, Hubert, Caspar, Ferdinand, Petra, Bernadetta, Dimitri, Sylvain, Felix, Annette, Claude, Lorenz, Raphael, Ignatz, Lysithea, Marianne, Hilda, Leonie.

Byleth:

 

Edelgard:

 

 

Hubert:

 

 

Ferdinand:

 

Caspar:

 

 

Petra:

 

 

Bernadetta:

 

 

Dimitri:

 

Sylvain:

 

 

Felix:

 

Annette:

 

 

Claude:

 

Lorenz:

 

 

Raphael:

 

 

Ignatz:

 

 

Lysithea:

 

Marianne:

 

 

Hilda:

 

Leonie:

 

 

Edited August 2, 2020 by Owns
Edited class rating explanation. Added images. Removed link to Reddit post. Removed some images.

[eclipse]

. . .so you joined just to share your calculator?

[Owns]

 

36 minutes ago, eclipse said:

. . .so you joined just to share your calculator?

"Just"?, no . . . I joined in search of others who might share my interest. And also because I thought this forum might have a different culture from what I might find on Reddit so I decided to post on both sites. Diversity is a good thing imo.

[eclipse]

3 minutes ago, Owns said:

 

"Just"?, no . . . I joined in search of others who might share my interest. And also because I thought this forum might have a different culture from what I might find on Reddit so I decided to post on both sites. Diversity is a good thing imo.

I agree that diversity is good.  But you've linked a reddit thread, instead of the calculator itself.  From someone that isn't you, what do your actions look like?

I'll leave it alone, but I'm not clicking on that link.

Edited May 18, 2020 by eclipse

[Cysx]

Ok, I won't lie, this is not necessarily how I go about picking classes in 3H in particular and many would tell you that growths matter relatively little there compared to other FE games. With that being said, it's a handy tool to have for cross referencing and making decisions faster since these things are still pretty big factors, obviously. So in that regard it's pretty nice.

As you said, the logical next step would be to take class modifiers into account, as they're relevant here and I figure could be included organically? Character bases I don't think are necessarily a priority in comparison, although that also matters a bunch. Finally class bases play some role for certain characters/builds(armor knight to raise def being the big prominent example), though that's a very minor omission, since they tend to only take effect for classes the character probably shouldn't be in in the first place(with exceptions, like Lorenz).

Also a smaller thing which might be a result of me misunderstanding the intent but, maybe don't make the ratings of both intermediate and advanced ratings a shade of orange. That's all I've got in terms of visuals, though.

Thanks for sharing!

[Shanty Pete's 1st Mate]

It's clear that a lot of work went into this, so that's quite cool! For my part, I don't really pick classes based on the how the character and class growth rates "stack up". Well, I once actually did a "counter-pick", and made Fortress Knight Petra, so she'd have good defenses while keeping an acceptable speed growth, haha. Depending on how in-depth you want to get, you could incorporate factors like spell lists, or exclusive combat arts (i.e. a unit with 4 reason spells would score "better" than one with 3 spells, in classes with a reason-linked ability).

Thanks for sharing!

[Owns]

22 hours ago, Cysx said:

the logical next step would be to take class modifiers into account,

This is  a great suggestion, and makes a lot of sense. Thank you. I've already tested it out and it changes the results somewhat. Serves as a tie breaker in some cases.

The thing with this kind of calculator is that the more complicated it gets, the more sensitive it will be to the factors I put into it. For example, even in its current simple form, I have to choose a weighting for Skill requirements. Depending on how I choose that weight it can change the results very little or somewhat significantly. For the most part, the top few ratings remain the same top few choices, what mostly changes is their ranking (ie. which one gets top spot, 2nd, 3rd etc). I guess this is some level of robustness but at the same time reveals why a calculator like this is ultimately subjective, and trying to find the 'perfect' formula is a hopeless endeavor 😅. I still find interesting and informative to look at some data in this way nonetheless. Probably the most objectively useful part is the main table showing the combined growth rates, but I doubt that's something new to people asides from maybe the presentation.

I am surprised to hear that people don't put too much emphasis on class growths when selecting their builds. Firstly it does seem that the calculator's top few class picks, based on growths, align with builds recommended in the community. Even some builds that for the casual gamer aren't obvious (at least they weren't to me) like Sniper Felix, Wyvern Lord Petra, or Wyvern Lord Ferdinand are revealed as a good choice by this calculator. But I also say this because in some guides/comments I've read, I'm sure I've seen emphasis placed on growths particularly loss of speed from Cavalier, etc. The trophy winning guide on gamefaqs comes to mind.

Thank you for the comments.

[paladin21]

14 hours ago, Shanty Pete's 1st Mate said:

It's clear that a lot of work went into this, so that's quite cool! For my part, I don't really pick classes based on the how the character and class growth rates "stack up". Well, I once actually did a "counter-pick", and made Fortress Knight Petra, so she'd have good defenses while keeping an acceptable speed growth, haha. Depending on how in-depth you want to get, you could incorporate factors like spell lists, or exclusive combat arts (i.e. a unit with 4 reason spells would score "better" than one with 3 spells, in classes with a reason-linked ability).

It's not just the amount of spells (or arts), but their quality, and character proficiencies should also factor in here.

As an example you have Manuela, with a heavy utility white magic spells but not much attacking or healing, and Bolting as her most interesting choice, but with difficulty to learn it due to her bane in Reason.

Then you have Annette, who has been named a beast as a Wyvern by quite a few people due to her Axe growth (to get use of the Bolt Axe) and her Lightning Axe art (her relic doesn't hurt either, but that's another story, the important thing is getting to use the Bolt Axe).However, she's ranked higher as a Mortal Savant due to her being able to use magic in this build. There was a very heated debate on this matter in another thread.

12 hours ago, Owns said:

I am surprised to hear that people don't put too much emphasis on class growths when selecting their builds. Firstly it does seem that the calculator's top few class picks, based on growths, align with builds recommended in the community. Even some builds that for the casual gamer aren't obvious (at least they weren't to me) like Sniper Felix, Wyvern Lord Petra, or Wyvern Lord Ferdinand are revealed as a good choice by this calculator. But I also say this because in some guides/comments I've read, I'm sure I've seen emphasis placed on growths particularly loss of speed from Cavalier, etc. The trophy winning guide on gamefaqs comes to mind.

Well, you just have to look at two classes in your lists: Brawler and Brigand. In most lists, there's a difference between the two of them favoring Brawler. However, most people would go Brigand. As a Brigand, you get Death Blow, which is enough to outclass Brawler in the intermediate tier. You also get a boost when learning brawling while fighting (smaller, but a boost is a boost).

There's also the point of usability. It doesn't matter the growths if you're in a class that doesn't allow you to grow. That's the reason most people say that flying classes are broken. Not so much for the growths, but for the mobility and stat bonuses.

[Owns]

6 hours ago, paladin21 said:

Well, you just have to look at two classes in your lists: Brawler and Brigand. In most lists, there's a difference between the two of them favoring Brawler. However, most people would go Brigand. As a Brigand, you get Death Blow, which is enough to outclass Brawler in the intermediate tier. You also get a boost when learning brawling while fighting (smaller, but a boost is a boost).

Yes, you make a good point here that highlights the limitations of this approach. I have actually poorly described what the calculator tells us. Sometimes one needs to take a step back to correctly identify what the data means, even if I designed it myself 😆

 

On 5/17/2020 at 11:41 PM, Owns said:

If this calculator does its job well, then the higher Class Rating value, the better that class should be (theoretically) for the character.

This is what I said about it, but that's actually wrong. I suppose it's what I was hoping the calculator would do 😅. What the calculator actually does is simply identifies which classes are a better 'fit' to the character's natural growths, taking into consideration also the character's strengths/weaknesses/budding talents in the skill requirements of the class. This 'fit' says nothing about 1) how good the class itself is, or 2) whether it is the best route in terms of other benefits like mastery abilities or if it is in line with a specific builds path. For example, you'll find many characters with a high score for assassin, because the dex and speed stats are important for many and assassin provides the highest dex and speed growths. Similarly, Mortal Savant provides decent growths in str, mag, and luck; so any character who has naturally high growths in a couple of those areas, and especially if they have a strength/budding talent in either swords, reason, or both (and no weakness in either), will automatically get a good rating for Mortal Savant.

So I think one thing the calculator does fairly well is ruling out classes that are most likely poor choices for that character. As far as the classes that do get a good rating, it simply highlights potentially good classes that one might consider. Sometimes it's on the mark, other times it isn't. You mentioned Annette as Wyvern Lord which gets a relatively low rating here, I'm not experienced enough to comment on that, but is WL Annette going to be better than a Warlock > Dark Knight path Annette? Or Gremory? Idk, it seems an odd choice to me but what do I know (really, I don't, I'm practically a newbie). An explanation would be nice (or link to a previous discussion).

Edited May 20, 2020 by Owns

[LoneRecon400]

1 hour ago, Owns said:

Is WL Annette going to be better than a Warlock > Dark Knight path Annette? Or Gremory? Idk, it seems an odd choice to me. An explanation would be nice (or link to a previous discussion).

How about practical demonstration.

Edited May 20, 2020 by LoneRecon400

[Owns]

2 hours ago, LoneRecon400 said:

How about practical demonstration.

Well I can't argue with 1HKO + canto. Awesome. In my current maddening run, only in CH 2 so far, I've been building Annette mainly as a rally bot. Her rally has been excellent so far, but I guess a unit that can 1hko is always better than one who mainly supports. Thoughts on this?

[Cysx]

4 hours ago, Owns said:

Well I can't argue with 1HKO + canto. Awesome. In my current maddening run, only in CH 2 so far, I've been building Annette mainly as a rally bot. Her rally has been excellent so far, but I guess a unit that can 1hko is always better than one who mainly supports. Thoughts on this?

Haven't tried the build myself so I'll refrain from saying too much, but it's very easy to underestimate magic combat arts. They're among the strongest damage sources in the game and the accuracy formula favors them a ton. Those builds usually don't have an enemy phase at all though, for what it's worth on maddening.
Better or worse than a rallybot, that's arguable, but then again as you can see in his skillset, she can do both.

Edited May 20, 2020 by Cysx