Children growth and initial stats

[nocturnal YL]

...again.

From Hrs-Game:

子世代の初期値と成長率
初期値は、両親がレベルアップで成長した合計値の1/4が、それぞれ対応するステータスに加算される。つまり両親が強ければ強いほど子供も強くなる傾向にある。
成長率は、元々設定されている子供の成長率に加えて母親の成長率1/2が加算される。

​

"Children's initial stats and growth"

"1/4 of both parents' level up gain total are added to the corresponding stats to make the initial values. Therefore, if the parents are strong, the child will also tend to be strong."

"Growth rates are determined by the child's pre-determined growths added by 1/2 of the mother's growths."

I don't know how accurate this piece of info is. It's better than Awakening's if you ask me.

If I'm not mistaken...

Children initial stat = (+parents level up gain total / 4)

I think this means that the children will get their expected stats, plus a quarter of what the parents got (level up gains, i.e. excluding their initial personal stats).

But it's only a quarter. This should explain why I saw that Midoriko with a lower speed than both Aqua (mother) and Suzukaze.

Growth = Child growth + Mother growth / 2

Great news to me, since this means I don't have to worry as much about the parents dragging down the children's final growths. What's shown as the children's growths are basically the guaranteed minimum, in that case.

Again, I have no way to verify it. I suppose I can (to a limited extent), but I don't have time for that right now.

I wish they have father-independent cap modifiers too. Lutz having a low Luck cap isn't like him at all.

EDIT - Line spacing. This forum software seems to hate old browsers.

Edited August 4, 2015 by nocturnal YL

[astrophys]

Something seems a bit extreme about the growths implied by a base + 1/2 growth formula…. I mean, for some kids like Sylla, that would push her up to 92.5% magic growths (Orochi as the mother) if that's all that there was, which would become 112.5% magic growth as a Spellcaster.. surely there'd have to be some kind of averaging factor…

They aren't going to be just making them automatically have 3/2 the total growths of typical characters…. unless they then divide out a 3/2 or something like that (effectively, making 2/3 of the growths of the child and 1/3 of the growths of the mother or something like that).

I don't have the game, so I cannot verify whether this is true or not.

[nocturnal YL]

Now that I think about it, it's possible that that site may have a different definition of "children's pre-determined growths". It could have been half of what's in the game data, making their growths and their mothers' both making up half of the final growths. Divide a constant by 2 and it's still a constant. I do doubt if this is the case, though.

I'll just leave it at that, since I don't have the time needed to verifiy anything. At least not right now.

(I tried to look at some of my children units' stats, but I can't tell either way.)

Edited August 4, 2015 by nocturnal YL

[astrophys]

Now that I think of it, if there are people out there who have been hacking their games in various manners, is it possible that one of them might have come across the code that effects the growths…

If they have, maybe one of them could come forward and tell us what the built in formula is…

[XeKr]

Given the statbooster tests, there has to be a cap somewhere if that's the formula, as some of us raised stats by 4 or more and didn't see consistent changes.

Broadly speaking, I was also thinking that mother's (other parent's) growths impact the children more. Simple addition however does seem high as the children already have solid personal growths but they didn't feel that impressive in game (as +20%ish in various stats would suggest).

someone go through the other topic's stats with these formula

Edited August 4, 2015 by XeKr

[astrophys]

Well, although its no guarantee of how it may be implemented in this game, let's take a moment to consider implementations in past games to see what the precedents so far have been:

Genealogy of the Holy War: Main parent’s growths rates + (Secondary parent’s growths rates / 2) + Holy blood bonus

Awakening: (Main Parent's Growths/3) + (Secondary parent's growth rates/3) + (Personal Growth Rates/3)

The claim at the first post, Growth = Child growth + Mother growth / 2, seems a bit implausible because it would be too dramatically high and probably would have been easily noticed (Orochi!Syalla would get something like +2 magic in level ups abut 1/8 of the time as an Exorcist and would always gain magic).

Looking at the precedent of Genealogy, I might postulate the following, although I don't have the game and this is just wild speculation:

Growth Rates (pure guesswork) = (2/3)[Child's Growth + {Secondary Parent's Growth/2}] = (2/3) Child's Growth + (1/3) Secondary's Growths

Besides taking inspiration from Genealogy with one character having twice the wait from the other, the idea of assigning the kid double the weight of the parent is from the idea that the kid is inevitably tied to their mandatory parent…, so perhaps they're using it as a stand in for the father's effect on the kid and the kid's own natural aptitudes combined…

Again, this is pure speculation on my part and I have no evidence to back me up, so don't take this very seriously…. its just a thought, a wild guess.

[VincentASM]

I suspect the information is from the guidebooks, but I don't have any of them on hand to confirm.

I doubt it's from data-mining, since IIRC nocturnal YL mentioned that Hrs-Game uses SF's data-mined growth rates.

Unless they discovered the formula afterwards and are just citing SF for convenience. Either way, formulae are harder to decipher than simple values like growth rates, so it's unlikely they found it from there.

[astrophys]

That makes sense…. although I'm still skeptical of it actually being child's growths + 1/2 parental growths…. wouldn't that be too outrageously high? Wouldn't people have noticed by now if certain characters were getting +2 stat level ups?

(Unless I'm mistaken [in which case I apologize for my error], no character can naturally reach 100% with class + personal growths, so any +2 in a given stat level ups would have been an immediate giveaway that something was up with the growths.)

That's why I'm guessing there would have to be some sort of weighting.

But who knows; I could be wrong.

[lysander]

Using awakening's formula for base stats and dividing by 4 instead of 3 seemed closer to likely than what that site was saying--not dividing out the child's defaults and only using the nebulous concept of "gained stats" is 100% wrong--but still didn't quite add up on the stats I've seen. Like the stat booster tests show, there must be something else going on there.

For growths, are they not stored? Should be as easy to see as any other character's growths I would think, but dunno.

[VincentASM]

Huh, I can't find the explanation about growth inheritance on the site; is it on a different page?

Also, I assume it's meant to be

Growth = (Child growth + Mother growth) / 2

That would make it more in line with Awakening's. Always seemed a bit redundant to me to have (Child + Both Parents) / 3, when two of those things are already constant...

Edited August 4, 2015 by VincentASM

[astrophys]

BOMDAS? Is that an alternately named acronym carrying a similar idea to PEMDAS (Parenthesis, Exponents, Multiplication/Division, Addition/Subtraction)?

Perhaps I could look up a few videos of the kids child sealing and see if the growth rates match what would be expected from such growth rate hypotheses and promotion bonuses.

[lysander]

Doesn't it not round until the very end? If they wanted to store growths and base stats as integers, that'd be why they're dividing a constant by a constant (I assume you meant the child_whatever/3 part).

Edited August 4, 2015 by lysander

[VincentASM]

BOMDAS? Is that an alternately named acronym carrying a similar idea to PEMDAS (Parenthesis, Exponents, Multiplication/Division, Addition/Subtraction)?

Perhaps I could look up a few videos of the kids child sealing and see if the growth rates match what would be expected from such growth rate hypotheses and promotion bonuses.

I was referring to BODMAS (brackets over division, multiplication, addition and subtraction). But it wasn't really necessary, so I edited it out of my post XD

Another solution is to have somebody with the right tools run a modified ROM with hacked growth rates. It should be easily possible since growth rates aren't encrypted, like they were in Awakening and New Mystery.

[astrophys]

That's true… you could set one parent's growth rate to something so large that it would still end up the overwhelmingly dominant term in the formula and test it that way.

EDIT/ADDENDUM:

I visited a video and checked Luna!Matoi leveling up

Child Seal Gains (to 18 promoted) - Standard Pegasus->Falcon Promotion Gains:

8 HP, 7 Str, 3 Mag, 7 Skl, 10 Spd, 10 Lck, 7 Def, 8 Res

If the formula is average of their growths, expectation is:

8.0875 HP, 7.22 Str, 3.4 Mag, 7.225 Skl, 9.775 Spd, 9.775 Lck, 6.8 Def, 7.65 Res

If the formula is 2/3 child, 1/3 mother, expectation is:

8.5 HP, 7.36 Str, 3.68 Mag, 7.65 Skl, 9.63 Spd, 10.2 Lck, 6.51 Def, 7.36 Res

It seems a bit hard to properly differentiate at the moment…

I'll try others….

Edit… its actually harder to tell than I thought…. with only 17 levels and the averaging/weighted averaging effects, plus possible random elements, its hard to tell…

ADDENDUM:

Averaging Luna, Matoi, and Subaki's growths (and accounting for Falcon) would have given 8.5 Spd, 9.06667 Lck, and 6.51 Res beyond the promotion growths. So this is more discrepant from the 10 Spd, 10 Lck, & 8 Res that we see from the 9.775 Spd, 9.775 Lck, and 7.65 Res that Luna Matoi would give.

Also, Skill is one of Luna's weak stats…

Averaging Luna and Matoi would give 7.225 Skl beyond promotion bonuses after 17 levels. We got 7 in reality.

Averaging Luna, Subaki, and Matoi would give 8.21666 Skl beyond promotion bonuses after 17 levels. But we only got 7. Of course, there could be some RNG component in play here, but if Subaki's skill growth was actually being inherited here we'd have probably gotten higher skill than we actually did. And if Subaki's growths were being inherited we'd have found lower speed on Matoi (on average a difference of more than 1 speed).

Conclusions: Subaki's growths most likely don't matter.

Edited August 5, 2015 by astrophys

[nocturnal YL]

Sounds more like it's still (child + mother + father) / 3 in that case. But yeah, we do need more data to confirm anything.

Personally, I don't really like this arrangement. Some children are supposed to be displaying opposite traits from their fathers, so linking up their stats don't make much sense here...

@Vincent -- it is on that page -- let's spell it explicitly:

http://hrs-game.main.jp/fe-if/data/gaiden-01.html 

Search for "子世代の初期値" to pinpoint it.

Edited August 5, 2015 by nocturnal YL

[charlesbarkley]

someone did a test a while ago on lutz's luck and the growths turned out to be closer to the expected value of (child +mother)/2

[VincentASM]

@Vincent -- it is on that page -- let's spell it explicitly:

http://hrs-game.main.jp/fe-if/data/gaiden-01.html 

Search for "子世代の初期値" to pinpoint it.

Thanks.

Huh, I still don't see it.

"子世代の初期値" means "2nd generation starting stats". I don't see a number 2 nearby or the kanji for "half" (for the division) anywhere either.

Above that section section is class inheritance. Meanwhile, the section underneath is support levels and the last one contains notes for the walkthrough.

Edited August 5, 2015 by VincentASM

[astrophys]

Here's a Google Translate translation of the quote that Vincent posted above…. obviously, Google translate isn't known for doing the best of jobs, but it may be useful for getting a quick, vague idea of what it says (for those of us who cannot already read Japanese) until a human translator can do a better job for us.

The quality of a child basically, "quality" of children is the same thing as his father.

It inherits his father's initial branch of an army + second branch of an army + mother of the initial branch of an army. At this time, if the quality of the mother of the initial branch of an army and father are duplicated, it inherits the second branch of an army of mothers.
Harold's son Lutz exceptionally "Dragon Knight", daughter of the cool breeze Green Lake has the makings of "medicine merchant", son drizzle of Aqua can not be a "diva".
Initial value Initial value of the child generation, 1/4 of the total value that parents have grown at a level up is added to the corresponding status. In other words, there is a strong enough children also made strong trend if parents is strong. Supporting only the level up main scenario but tough raise male character everyone to support S, it is possible to raise the level of assistance in online battle My Castle. About Gaiden capture data
Reinforcements described the action start turn.
The average level of the enemy increases, depending on the scenario the degree of progress, the enemy soldiers were junior positions and when it is more than Lv.20 become senior positions.

Wherein only the basic simple information so difficulty varies depending on the timing of a challenge.

Addendum:

We already basically knew this, but definite proof that growth inheritance must occur; 25:10 is the time in question

https://www.youtube.com/watch?v=mYKpxDfQWy8

Velour and Flannel both have 0% magic growths (personal and their starting classes) and they gain no magic on promotion from their starting class.

As such, any non-zero magic gains on promotion MUST be do to an inherited growth.

As we see here, Camilla!Flannel gained 2 magic in child sealing from Garou to its level 18 promoted variant. As Velour, Flannel, and the promotion of Garou have no magic growth, and since there is no promotion bonus, all magic gains here must be directly attributable to the effects of Camilla unless child seal growths break the rules.

Camilla has a 25% magic growth. Flannel and Velour have 0% growths…. let's see what combinations could possibly have yielded 2 magic.

Averaging father, mother, and child: 8.33% growth --> 1.4166 magic gained on average

------> I suppose its possible for the RNG to have gotten lucky, but this does not seem immediately likely; Velour would seem more likely to have gained 1 magic

Averaging mother and child: 12.5% growth --> 2.125 magic gained on average --> This fits much closer to the 2 magic that we actually saw, although more trials are needed to confirm.

Child counts twice as much as mother: 8.333% growth, numerically the same as the first case for magic --> 1.4166 magic gained

Let's look at Velour's other stats…

LOTS OF MATH in spoiler

Total bonus from promotion from level 20/0 to 20/18 via Child Seal MINUS promotion bonuses:

15 HP - (22 HP - 19HP) = 12 HP

16 Str - (11 Str - 8 Str) = 13 Str

2 Mag - (0 Mag - 0 Mag) = 2 Mag

12 Skl - (6 Skl - 4 Skl) = 10 Skl

12 Spd - (7 Spd - 6 Spd) = 11 Spd

8 Lck - (1 Lck -0 Lck ) = 7 Lck

12 Def - (7 Def -4 Def) = 9 Def

7 Res - (1 Res - 0 Res) = 6 Res

1 Move - (6 Move - 5 Move) = 0 Move

So that's what we need to account for in 17 level ups

12 Hp, 13 Str, 2 Mag, 10 Skil, 11 Spd, 7 Lck, 9 Def, 6 Res

EDIT: Corrected to 11 SPD

PREDICTIONS (via various methods):

Average of father, mother, and child:

11. 9 HP, 12.4667 Str, 1.4166 Mag, 7.0833 Skl, 9.9167 Spd, 5.95 Lck, 9.0667 Def, 5.6667 Res

Difference:

-0.1 Hp, -0.5333 Str, -0.5834 Mag, -2.91667 Skl, -1.0833 Spd, -1.05 Lck, +0.0667 Def, -0.3333 Res

Sum of Squares of Difference (many statistical techniques use least squares for quality of fit)

11.5333

Average of mother and daughter:

11.05 HP, 11.9 Str, 2.125 Mag, 8.5 Skl, 10.625 Spd, 5.95 Lck, 8.5 Def, 6.375 Res

Difference:

- 0.95 HP, -1.1 Str, +0.125 Mag, -1.5 Skl, -0.375 Spd, -1.05 Lck, -0.5 Def, +0.375 Res

Sum of Squares of Difference

6.01188

Weighted average of mother and daughter; daughter counts double:

11.333 HP, 11.9 Str, 1.41666 Mag, 8.2166 Skl, 10.2 Spd, 6.233 Lck, 8.78333 Def, 5.95 Res

Difference:

-0.667 HP, -1.1 Str, -0.58334 Mag, -1.78334 Skl, -0.8 Spd, -0.7667 Lck, -0.21667 Def, -0.05 Res

Sum of Squares of Difference:

6.45275

So, although there are still some discrepancies, it would seem (judged by least squares) that the average of the mother and the daughter is the closest to the actual result, although there's not a huge least squares difference between it and a weighted average of mother and daughter. The average of both parents and the child seems the least likely of the three models considered.

This analysis has not accounted for the "discreteness" of level ups, in that they can only be full point values.

EDIT/ADDENDUM: Elise!Kisaragi… this one is a good one because Takumi and Kisaragi have 0% personal magic growths and Elise has a 65% personal magic growth. Also, Bowman, Golden Kite Warrior, and Holy Bowman have 0% magic growths. So sure, there might be some RNG involved, but Elise is likely to have a noticeable effect on this stat>

Source: 11 min, 52 seconds into the video

https://www.youtube.com/watch?v=X5argrPMy6k

Bowman to GKW has a promotion bonus of 1 magic and the advancement of 17 more levels to level 20/18 has given it 5 more magic.

Bowman to GWK has a promotion bonus of 0 magic and the advancement of 17 more levels to level 20/18 has given it 5 more magic.

More math:

The case in which the growths worked as Father/3 + Mother/3 + Child/3 (Awakening style) or (2/3) Child + (1/3) Mother are identical for magic in this case because both Takumi and Kisaragi have equal (0%) magic growths. But if its the average of the parent and child then Kisaragi should have 32.5% Magic growth

Father/3+Mother/3+Child/3 Method: 21.666% Magic growth predicted

(2/3) Child + (1/3) Mother Method: 21.666% Magic growth predicted

Child/2 + Mother/2 Method: 32.5% Magic growth predicted

17 levels from 20/1 to 20/18 should give gains of:

Father/3+Mother/3+Child/3 Method: +3.68333 Mag predicted

(2/3) Child + (1/3) Mother Method: +3.68333 Mag predicted

Child/2 + Mother/2 Method: +5.525 Mag predicted

Actual growth: +5 Mag

So the Child/2 + Mother/2 method seems to be favored here. Of course, there could be some RNG, but this is a pattern we've been seeing repeatedly over other samples.

As for Strength:

Child/2 + Mother/2 predicts +4.675 Str as GKW and +6.375 Str as Holy Bowman… this rounds off to the +5 Str and +6 Str we actually saw.

Father/3+Mother/3+Child/3 predicts +5.38 Str as GKW and +7.0833 Str as a Holy Bowman… the GKW result works out fine, but Holy Bowman gets a full extra point of strength than we actually did.

(2/3) Child + (1/3) Mother Method would make this result even worse, as Kisaragi has a higher personal (initial) strength growth than Takumi does.

Conclusion:

[Child Personal Growths]/2 + [Mother Personal Growths]/2 is most likely, from many examples, the way in which growth inheritance works, although there might be some noise due to RNG in the statistics.

Edited August 5, 2015 by astrophys

[Jayvee94]

^ have you factored class growth rates?

[astrophys]

Yes. I have. Holy Bowman has 15% strength, 0% magic. GKW has 5% str, 0% magic. etc.

5% from elise and 40% from Kisaragi gives an average personal strength growth of 22.5%.. 37.5% class + personal as Holy Bowman results. This is what predicts that in the 17 levels from 20/1 to 20/18 Elise!Kisaragi should get 6.375 str if my hypothesis for growth inheritance is true. He does get 6 Str (and you cannot get fractional strength in the actual game).

If Takumi, Kisaragi, and Elise's growths were averaged and the Holy Bowman growth added, he'd have a personal + class growth of 41.66666%, so therefore the 17 levels would get a strength growth of 7.083333.

In this case, magic was simpler as both GKW and Holy Bowman have 0% magic growth anyways

[Jayvee94]

Have you noticed that Lutz's Luck was meant to be capped from the start? Like from when he joins

[Jayvee94]

It kinda sucks that Shigure and M! Kanna has no variable growths.

[astrophys]

Well, remember that not all of the details have been worked out yet (such as if here are any exceptions, etc.). None of my examples have worked with those characters yet.

I'm guessing that Shigure might be affected by his father's growths (making it "optional parent"/2 + child/2 rather than "mother"/2 + child/2). But I have not been able to test this yet.

[nocturnal YL]

Have you noticed that Lutz's Luck was meant to be capped from the start? Like from when he joins

I did notice this very well, back when I was trying to figure out children initial stats in another topic.

Lutz doesn't feel particularly lucky to me stat-wise. Stat cap modifiers still work like Awakening and being a dragon knight doesn't help.

It kinda sucks that Shigure and M! Kanna has no variable growths.

Even though the text says "mother", I'm fairly certain that it's just shorthand for "variable parent".

Thanks.

Huh, I still don't see it.

"子世代の初期値" means "2nd generation starting stats". I don't see a number 2 nearby or the kanji for "half" (for the division) anywhere either.

Above that section section is class inheritance. Meanwhile, the section underneath is support levels and the last one contains notes for the walkthrough.

...They updated the page and removed the part about growth.

[eharper256]

Sorry to necro this thread a bit, didn't want to start a new topic. Has there been any further news/confirmation it is indeed (Optional Parent + Child)/2 for growths?