ItJustWorks Library of Stats and Formulas

ItJustWorks is a Blue Archive JP club filled with enthusiasts for the deep theorycrafting elements of the game. This document aims to consolidate the stats and formulas driving Blue Archive behind the scenes. Many of the content in this document has been discovered and proofed by ourselves. Credit is given where it is not our work. Some things in this document have not been fully validated; there will be disclaimers if this is the case.

Science/method is (sometimes) documented in Appendix.

Thank you to everyone in ItJustWorks and all branch clubs for contributing to the findings.

This document is lead and written by Causew#5130. DM me for any feedback.

Additional/Special credits to:

Zettai Keiki#0429 (Toir)
Stokesia#0932
Noxy#8299
HerztInYou#8459 (Seventh)
sirlogan155#5214 (ZeroTwo)
degendemolisher#8671
Senovit#8237
Ahri#1337
Midokuni#0001

Special shout out to @pachimi on twitter, who asked me to copy this doc for the JP audience. Check out his version here. Pachimi-sensei’s document can also serve as a source of truth for the formulae, and it is recommended to cross-reference between our docs.

Also note that most of the stuff in this document doesn’t actually matter and will not affect your decision making while playing the game. The purpose here is mainly comprehensive documentation, and to satisfy curiosity.

To do list:

Document buff stacking

Basic Knowledge

Before diving into the hidden formulas of the game, there is some basic knowledge a player must know beforehand. Unlike the formulas, this knowledge typically has real influence on your decision making when attempting to build a unit or team.

Armour Matchups

Each student has an attack type and defence type. Damage is modified depending on the matchup between the attacker and defender.

(Table is out of date in regards to purple. Will update when it becomes relevant in Global)

Where:

Weak: 2x damage
Normal: 1x damage
Resist: 0.5x damage

Mood

There is a damage modifier for the mood of a student given the terrain they are deployed in.

Damage is not the same as the attack stat.

Mood	Damage Multiplier	Block Rate Behind Cover	Chance to Ignore Block
(SS)	1.3x	75%	75%
(S)	1.2x	60%	60%
(A)	1.1x	45%	45%
(B)	1.0x	30%	30%
(C)	0.9x	15%	15%
(D)	0.8x	0%	0%

Special Bonus

A portion of a Specialist student’s stats are shared with all strikers.

Stat	Portion Shared
Attack	10%
HP	10%
Defence	5%
Healing Stat	5%

Bond Bonus

Increasing your bond level with a student gives the student extra stats. If a unit has variants, each variant will contribute additionally to the unit.

For example, Swimsuit Mashiro at Bond16 adds 75 Attack and 230 HP. Regular Mashiro at Bond18 adds 83 Attack and 454 HP. This means both Mashiros will receive a bonus of (75+83) Attack and (230+454) HP, which equals 158 Attack and 684 HP.

Base Stats

The stats under ‘Basic Info’ are NOT a student’s base stats. A student’s base stats are visible from the Level Up page. Only HP, ATK, DEF, and Healing grow as you level and unlock mystic. Formula for stats per level below (and also in the appendix).

Stats per level formula:

Given:

X = level of student
Stat[x] = stat value of level X

Stat[1] = value of stat at lv1
Stat[100] = value of stat at lv100

Equipment and Unique Equipment

Equips give stats (shocking news). These stats can be either flat stats and/or % multipliers. Stats from Unique Equipment (UE) work the same way.

The Student Stats Page

Under the ‘Basic Info’ tab of a student, there is an option to view detailed stats for the student. It is important to note that the stats on the unit page are not the final values in battle. It also does not account for any skills. However, they still serve as a decent rough estimate for how well your student will perform.

Definitions

In-depth details regarding these stats are in the next section/linked in the first column.

Stat	Description
HP	Health points determine how much damage your student can take before she falls.
ATK	Attack is used as the baseline stat for damage actions. The attack stat is not the same as damage.
DEF	Defence reduces the amount of damage taken by the student.
Healing	Healing is used as the baseline stat for shield or healing actions.
Accuracy	Accuracy determines how likely your student will successfully land an attack on an enemy
Evasion	Evasion determines how likely your student will evade an incoming attack
Crit	Crit (chance) determines how likely your student will land a critical hit
Crit DMG	Crit DMG determines how much bonus damage your student will deal upon a critical hit
Stability	Stability determines the damage variance of a student. Having a higher stability value increases the damage floor.
Range of Fire	Range of Fire determines how close your student needs to be to shoot a target
Crowd Control Power	CC Power determines how likely your student will successfully land CC (Fear, Taunt, Stun)
Crowd Control RES	CC Res determines how likely your student will evade incoming CC
Cost Recovery	Cost Recovery determines how much Cost(SP bar) the student helps generate

Stats Formula

The General rule for the calculation of all stats is:

Where flats include but not necessarily limited to:

Base stats
Bond stats
Flat Equip Bonuses
Special Bonus
Limit Break Stats
Flat Skill Bonuses
Flat debuffs
Flat resistance values

And where %s include but not necessarily limited to:

Equip multipliers
Skill multipliers
Debuff multipliers
Stage/event bonuses

HP

Follows the general rule for calculation.

ATK

Follows the general rule for calculation.

DEF

Follows the general rule for calculation. The final number is converted to a multiplier which affects the damage taken by a unit. The formula for converting DEF to damage taken is:

Damage reduction is just the opposite of damage taken (1 - Damage Taken)

For formula simplicity, we typically use a damage taken multiplier

Defence reduction is capped at 80%. Source

DEF Penetration

Def Penetration (aka DEF Ignore) is like DEF debuffs, but entirely localised to the unit dealing damage. Because it is a different stat, things get a bit tricky when considering DEF Penetration.

Firstly, calculate the target’s DEF using the general rule.

Then, calculate Final DEF by subtracting flat DEF Pen.

Finally, % DEF Pen is multiplicative on Final Def, which will give you the Damage Taken Value.

That is:

Note: There is no unit with both flat DefPen and DefPen%, so we are unable to truly confirm the order of operations. For existing units with either stat, this will work.

Healing

Follows the general rule for calculation.

Accuracy/Evasion

Both stats follow the general rule for calculation. Discovered by anons on JP Wikiru, the formula for the relationship between Accuracy and Evasion is:

This means that when accuracy is higher than evasion, the chance to hit is always 100%.

Crit (Chance)

The current hypothesis is that it actually breaks the general rule for calculation. However the part that breaks only affects a tiny handful of students with near indetectable differences even in large samples - so for the most part we can calculate Crit Value according to the general rule. If you’re curious about the actual rule, see the work done in the Google Sheet linked in the Crit section of the appendix.

Crit Value and Crit Res are calculated separately before interacting with each other.

i.e.

Disclaimer: Probability is difficult to prove. However we are confident this is as close to the truth as possible.

Crit Chance is capped at 80%. Crit value is basically uncapped, but a crit chance value higher than 80% will be “rounded down” to meet this cap. It has been found that there is no hard cap to crit chance. Sometimes, you may see players mention a “soft cap”. It should be made clear that there is no actual cap coded into the game. Rather, this “soft cap” comes from the formula itself. The returns from CritValue dramatically diminishes as it increases. You need 2666 FinalCrit to have 80% crit chance - this is not possible under regular circumstances.

Crit Damage

Follows the general rule for calculation. Crit damage and Crit damage res are calculated separately before interacting with each other.

I.e.

Meaning a Crit Damage value of 1500 is equal to 0.15 (aka 15%) crit damage multiplier.

Stability

Stability is a static value between every character, with exception to those who have stability increasing effects on their Unique Equipment passives. The stability value directly converts into a % value that dictates the character’s damage floor. The range between the damage floor and damage ceiling will henceforth be known as ‘Damage Variance’.

Discovered by JP Twitter,, the formula for the relationship between stability and damage floor is:

As far as we can tell, Damage Variance is distributed evenly between Damage Ceiling and Damage Floor.

Additionally, it should be noted that the +0.2 constant is technically a variable in the game's code. This constant (we’ll call it the stability constant) can be buffed/debuffed independently from the stability value to control damage variance linearly. For example, Binah’s 10% stability debuff actually changes the stability constant from 0.2 to 0.1 rather than debuff actual stability (for the better, because a 10% debuff on stability is disastrous).

All students have +0.2 by default, but abilities that have no damage variance will max this out to 1 (Dress Hina). Bosses will also have their own stability constant.

Range of Fire

Appears to follow the general rule for calculation

Nothing else to say.

You can know that the pickup radius for Serina’s box is 500 units. You can use that as your measuring stick if you want to do range tests yourself.

Crowd Control Power/ Crowd Control Res

Follows the general rule for calculation.

That is, final CCpwr and CCres is:

According to JP Twitter, the formula for the interaction between final CC Power and Res is:

Unless specified, the base CC chance for most skills is 100%

For content where CC Power affects duration (Insane HOD), simply take CC power as a % over 100 and multiply that on the base duration. Using the same final CCpwr and CCres stats as above, CC duration is:

Cost Recovery

Each student has a fixed cost recovery rate of 700. This number represents the amount of cost they generate per second. This means in a team of 6 students, the cost generated by the team is 4200 per second. Any cost generation increase skills follow the general rule for stat calculation (eg. Cherino as flat addition, Himari as % increase).

One bar of cost is equivalent to 10,000 points. Which means in a team of 6 students generating 4200 cost per second, one bar is recovered in ~2.4 seconds. To put it formally:

Discussion around this can be a little confusing sometimes, because Nexon calls the cost units (10k per bar) ‘Cost’, and also the bars themselves ‘Cost’.

Other Stats and Effects

There are other stats, buffs, and debuffs not visible on the stats page.

Damage Reduction

The amount of damage taken by an entity can be directly buffed or debuffed. This is not the same as an ATK/DEF buff/debuff. It is denoted by a broken shield icon, where red means the target takes reduced damage and blue means the target takes increased damage. Regardless of whether they are increases or decreases, all instances of damage reduction are additive to each other. The DEF to damage reduction is NOT part of this equation and does not contribute to the final damage taken value in the same sum of totals.

Damage Amplification

The amount of damage dealt by an entity can be directly buffed or debuffed. This is not the same as an ATK/DEF buff/debuff, and is not the same as Damage Reduction. It is denoted by a fiery sword icon, where red means the target deals increased damage, and blue means the target deals reduced damage. Regardless of whether they are increases or decreases, all instances of damage amplification are additive to each other.

Level Difference Penalty

ONLY WHEN AN ATTACKER IS LOWER LEVEL THAN THE DEFENDER, damage dealt is reduced. The damage taken after applying this penalty is written as:

To clarify again - there is no damage bonus if the attacker is higher level than the defender.

If your student is lower level than the enemy, you have a penalty against the enemy.

If the enemy is lower level than your students, then the enemy has a penalty against you.

This Lv Difference modifier is a separate variable to the damage reduction mod, and it has its own place in the damage formula.

The penalty also caps at 60%. So whether you are 30 levels under or 40 levels under, the penalty will be 60% either way.

Recovery Rate

The amount of healing received by an entity can be directly buffed or debuffed. This is not the same as a healing STAT buff/debuff. It is denoted by a cross and wreath, where red means increased healing received and blue means decreased healing received. This works similarly to damage taken.

Block Rate

Yuuka’s UE gives her bonus block rate as a flat value… This implies that part of the calculations for block rate require a section that processes this flat value.

Uncovered by decompiling the code, the formula for block rate is:

Note that the 3000 is actually “Cover Block Rate”, but the base block rate of all cover in the game is set to 3000. If at any point this changes then this part of the formula will change accordingly.

The block and ignore rates for each mood being:

Mood

Block Rate

Ignore Rate

(SS)

7500

(75%)

7500

(75%)

(S)

6000

(60%)

6000

(60%)

(A)

4500

(45%)

4500

(45%)

(B)

3000

(30%)

3000

(30%)

(C)

1500

(15%)

1500

(15%)

(D)

(0%)

The final block rates can be summarised in the following table (If there is no bonus block)

Attack Speed

Attack Speed affects the delay between actions in the “action queue”, and the animation speed for some actions.

Discovered by JP Tweeters, the animation speed/delay formula can be written as:

The action queue follows an ordered priority in executing actions:

Moving forward if there is no enemy in sighting range
Normal skill usage if off cooldown
Moving into shooting range if there is an enemy in sighting range but no enemy in shooting range
Take cover available within shooting range of enemy

Taking cover can sometimes occur after a complete cycle of step5

Firing at the enemy which consists of

Reload if magazine is out of ammo
Mounting (for mounted MGs)
Aiming
Shooting a burst
Cooldown after shooting (a.k.a. attack delay)

Again, attack speed affects the delay between each action in the action queue, and the animation speed for all actions under ‘5. Firing at the enemy’.

High priority actions do not cancel lower priority actions. All actions must finish before executing the next.

Worth noting that reloading(or rather the shot that follows a reload), forced movement (Serina, Fuuka, etc), and EX skills reset the action queue.

Tactical Supports (TS) and Summons Stats Calculation

Summons are additional entities that are called to the field by skills. These include:

Tactical Supports (Vehicles)
Pets (Peroro-sama, Turrets)
Cover (Shizuko, Shimiko)

The summoned entity is treated separately to the student who summoned it.
The summon does not receive the SUB skill buffs from Specialists, as these buffs are applied only once at the beginning of the battle.
Summons will not be forcibly moved by skills such as Serina’s health kit.

The summon will have their own stats completely individual to the student who summoned it. These stats will scale to the student’s level. Summons additionally receive a buff to its stats the moment it is summoned. This summon buff can be identified from the skill description for the summon:

In this example, Peroro-sama will be summoned with its own base stats in addition to 160% of Hifumi’s max HP.

Formally, the summoned buff is simply noted as:

StudentFinalStat being a students stats after all modifiers, such as equipment and skills

The summon further inherits equipment bonuses from its student. The final stats of the summon is written as:

Equipment bonuses are calculated multiple times. Once during the calculation of the students stats (consequently increasing the summon buff) and again during the calculation of the summons final stats.
Since equips increase many stats besides those indicated by the skill, this can mean summons such as Utaha’s turret can crit because of the crit inheritance from Utaha’s watch.

Buff Duration

The “Buff Duration'' stat affects the duration of buffs. The stat itself follows the general rule for calculation. As for how much the buff is extended by, simply take the final stat as a % over 10000, i.e.:

Calculating Damage

In general, the Damage Ceiling Formula is the product of the source stat multiplied by all relevant multipliers

Every instance of damage (hits) is calculated separately. The above formula represents the highest possible damage any hit can deal. In most cases, this damage ceiling is processed again by stability, ultimately generating a random damage value between the damage floor dictated by said stability value, and the ceiling.

There are a few caveats to these formulii but it will at the least provide a reasonable estimate for a student’s damage.

There are some details, additional rules, and variables depending on what you are attempting to calculate. These rules can be generally split amongst the following groups:

Auto Attack damage
Regular skill damage
Sub skill damage
Misc. skill damage

Auto Attack Damage

Auto attack damage is the most basic form of damage, and it follows the general formula. The most important thing to remember is the division of the number of shots fired per auto attack.

For example, Pina shoots 5 bullets per salvo (auto attack). Each bullet will deal about a 5th of the Pina’s attack and multipliers before accounting for defence. Notice how each bullet has their own individual crit and damage variance calculations.

Regular Skill Damage

Regular (EX/Basic) skill damage is mostly simple and follows the general formula… however it gets a little bit tricky when you attempt to account for hit count. Most skills in this game do not specify how many hits they deal, so you’ll have to figure this out on your own. It doesn’t help that the wording for some skills can be very confusing. Here’s some examples:

Nonomi’s skill simply says it deals damage. The skill is split into two hits and you need to divide her skill damage by the hit count.

Mutsuki’s skill says her damage is dealt to enemies within 3 areas. There are 3 hits, but each hit deals a unique instance of the damage multiplier. This skill does not need to be divided by hit count.

Chise’s skill says her damage is continuously inflicted to enemies over time. There are multiple hits, but each hit deals the advertised damage. This skill does not need to be divided by hit count.

Iori’s skill explicitly says each shot deals its advertised damage. This skill does not need to be divided by hit count.

Kinda confusing. NAT isn’t always consistent with their wording for skills and it especially doesn’t help that NEXON sometimes mistranslates them. You’re gonna need to find out for yourself whether skills are split by hits or not.

Sub/Misc. Skill Damage

Voodoo magic.

In Blue Archive, there are many sub skills that deal additional damage and work in a unique manner. This additional damage usually follows the generic formula… The problem is that the source stat and skill multiplier variable can be obscure, to the point where the skill description can be misleading or straight lying to you. In some cases, there can be multiple source stats which contribute additionally to the other source stats… this probably doesn’t make any sense right now, so it’s best to do this by case.

Note: An extra instance of additional damage which originates from a specific source is sometimes called “echo” damage.

Attack Scaling: Shun and Iori sub skills

The additional damage of Shun’s sub skill scales off her ATK stat. This is easy enough to understand. Just plug in the source stat and multiplier into the formula and you’ll do that much damage.

You can see in the screenshot below that the AA and the echo deal near identical damage (as expected from the 96% multiplier).

However, don’t forget that each instance of damage has individual calculations for crit and damage variance, so sometimes you can observe really extreme differences.

Important: Originally we thought Shun scaled off damage dealt… but remember Shun can only AA which means her damage dealt is equivalent to her attack stat, oops. Shun’s Ex skill increases her AA damage, but it will not increase the echo damage.

We can demonstrate the ATK scaling further with Iori’s sub skill:

Iori’s sub skill has identical wording to Shun’s, and it also scales off her ATK stat. Remember that for SRs with only one attack, the damage dealt by an AA is basically just the ATK stat. This looks good on paper:

1777*0.43 is 765. The damage checks out.

But because its not damage scaling, it’s actually not as strong as you think it is in practice:

On this screenshot an Iori EX hit a crit for 59K, but the echo damage is only 2k because it scales off Ioris ATK stat, not the 59k damage dealt.

Mixed Scaling: Izumi and Junko sub skills

The Gehenna Gourmet Club is the origin of ItJustWorks. These multi-scaling skills operate on some extremely obscure logic. Their in-game descriptions are as follows:

First things first, these are not scaling multipliers. Meaning, this is not a single multiplier that increases or decreases based on HP. There are two numbers and each number represents a constant multiplier applied to different source stats. The number on the left is the multiplier applied to HP%, and the number on the right is the multiplier applied to ATK.

After each source stat is processed by the multiplier, the sources are added together and then processed again by mood, armour, and defence for the final damage outcome.

Helped by Pachimi-sensei, we were able to uncover how the final damage is determined:

Junko:

The result of has a minimum value of 0

Izumi:

The result of has a maximum value of 1

This means that even at 0% potential (Junko at Max HP, Izumi at 0HP), these sub skills will always deal additional attack scaling damage. They will then do even more damage according to their current HP% value.

Incredibly convoluted and you might need to read it a couple of times to make sense of it. Don’t think too hard though - just know that they’re good skills and they generate nice numbers on the damage graphs.

Calculating Healing

Healing essentially shares the same formula as damage, but with far fewer variables you need to worry about.

Skills with multiple healing sources, such as Eimi’s, are simply tacked on additionally with each source multiplied by recovery modifiers.

Appendix

Proof of general stats formula

Shun is a useful measuring stick because her EX skill allows her to target the unit with the highest attack. Knowing this we were able to figure out the stats formula before Unique Equipment is a factor.

The test was done with Iori and S.Tsurugi, who have similar levels of ATK at certain levels.

Based on only what we see on the equip screen, Shun should be targeting Tsurugi. This is true without specialist support.

But we discovered that adding support bonus to the equation changed the targeting.

Knowing this, we can work out where the support bonus is considered:

Lv.75 4* Iori Base + Bond17: 3367 + 83
Lv.75 4* STsu Base + Bond19: 3629 + 108
Support Bonus = 952
Shoes T6 = 1.25x
Hat T6 = 1.35x

IORI IS TARGETED BY SHUN EX GIVEN ABOVE STATS.

Formula 1:

Iori = (3367 + 83) * 1.35 + 952 = 5609.5
STsu = (3629 + 108) * 1.25 + 952 = 5623.25
Rejected, because Iori should have higher attack due to Shun targetting

Formula 2:

Iori = (3367 + 83 + 952) * 1.35 = 5942.7
STsu = (3629 + 108 + 952) * 1.25 = 5861.25
Not rejected, because Iori has higher attack and is confirmed to be targeted by Shun

From here we had to work out where UE fit in with the large amount of flat boosts it gave to a student. We also needed to know where %modifiers lay too for stats. Continuing to use Shun was impractical due to the large amount of effort and cost required to line up two students attack values and praying an enemy Shun will EX. Instead, we decided that it’ll be easier to send a solo Haruna to her death and add up the damage dealt by the enemy PvP team.

This process helped us define the formula:

After observing that this calculation held true for ATK, HP, HEAL, and CRIT DMG, we now hold the assumption that this formula is the null hypothesis for all stats. I.e. This formula is the default and should be used in all stat calculations until proven incorrect.

Note that this has already been partially discovered by JP players and published by JP Youtubers, but there was no proof of process. For complete comprehension we needed to do this from scratch.

Stability

Discovered by anons on JP Wikiru. Comparing the formula to our records of highest damage and lowest damage dealt per student in any situation, the formula checks out.

Accuracy/Evasion

Discovered by anons on JP Wikiru. Comparing the formula to rough data we collected this formula checks out:

Cost Recovery

Was done and proofed ages ago early in JP’s life, but noone documented it so we had to do it again. Here’s the link to the discord convo if you wanna read it yourself.

Picture Version.

Direct Link

Basically it’s just frame counting how long it takes to get 10 bars and then guessing the rough cost units per bar based on the time it takes to recharge and a recovery rate of 4200 units.

Finding the damage formula

Didn’t require much work… so we don’t have the documentation for this oops. Basically we just controlled all our variables and hit a mob we knew the defence of a few times. Not very hard.

Junko and Izumi sub skills

We first discovered the fact of multiple sources from looking into a game logic sheet provided to us by a dataminer. For privacy/security reasons we can’t show you this datamined info.

With the knowledge that we have multiple sources, it just became a matter of coming up with formulas and proving them wrong until something stuck. Went through about 6 different formulas (our calculated damage was often a severe overestimation of actual damage) before we got to our current one. We fully expect the current one to be bunk too so I’m not going to write too much about it until we figure it out for sure.

Eimi recov test

EimiHPTesting - using cashew's damage calc sheet base

Stability

From n=695 the distribution of damage appears to be even.

Data sheet (Thanks Dajin)

Where did “ItJustWorks” come from?

At the beginning of the JP server none of us knew how any of the game logic worked.

No matter how hard we tried to fit the variables we had (stats, equips, skill%), it never added up to the actual values in-game. Then, with our limited knowledge on the game, we tried to figure out Junko sub. It didn’t go well.

Borrowing units from circles became a thing afterwards, and we created a circle called ItJustWorks.

And here we are now =)

Stats per level formula:

Given:

X = level of student
Stat[x] = stat value of level X
Stat1 = value of stat at lv1
Stat100 = value of stat at lv100

Old disproven formulas

Crit:

These formulas were hypothesised by JP Wikiru anons in May, link. Probability being what it is, we haven’t put in the effort to prove any of the formulae wrong. With the release of HOD, these formulas can be proved wrong.

Formula 1:

Formula 2:

WITH

Additionally, Pachimi-sensei provided evidence that the denominating constant is 666.66, rather than 650, with n=10,000: crit rate count(2/23)

Using statistical tests, we can reject the old formulas

Stability (old):

Discovered by anons on JP Wikiru, the formula for the relationship between stability and damage floor is:

Finding the function for defence to dmg red (old).

The process behind finding the formula was simple. All we had to do was plot damage taken against the defence of our test unit. Perorozilla was used for our 0 DEF data point because the boss applies a flat -9999 DEF debuff to a target. Previously it was impossible to plot this line because there was no way to drop to 0 DEF.

We first plot damage taken vs Perorozilla on normal difficulty.

And to help with validation we plot the same thing on higher difficulties.

And then normalise so that DEF = 0 gives us DMG Taken = 100

You may have noticed that R2 is not perfect, and that the lines split up from each other at extreme negative values. This means that there is a bit of error in our formula. We couldn’t do much at the time though because this error is caused by stability/damage variance. If we had a variance free way to deal damage then we would be able to refine this formula and eliminate the error. We’ve discovered that Junko’s self damage actually goes through all the usual damage calculations, and is free from stability - so next Perorozilla we potentially can use her to correct this formula. Until then we’ll have to deal with the error.

Update: JP solved this ages ago. We now use their formula.

DEF (old old)

Follows the general rule for calculation. The final number is converted to a multiplier which affects the damage taken by a unit. The formula for converting DEF to damage taken is:

What the curve looks like

The limit for damage taken is 30% as DEF approaches infinity
Allow for approximately 0.5% positive error (the formula may slightly overestimate actual damage taken).
Damage reduction is just the opposite of damage taken (1 - Damage Taken).
Plans in place to eventually correct the error
Defence reduction is capped at 80%. Source
Disclaimer: JP uses Damage Reduction = DEF/(DEF+1666.66), this may be correct

Changelog

2024-07-29: Added DEF Pen
2024-05-28: Clarified that CCres also affects CC duration for relevant content
2024-04-12: Added damage amplification as a separate stat to damage reduction
2024-04-09: Added note about 60% cap to Lvl diff penalty (how was this not here already?)
2023-12-07: Fixed accuracy formula from 700 > 666.666
2023-08-19: Minor edits for clarity
2023-01-14: Turned into pageless
2022-11-17: Added buff duration
2022-10-21: Updated note on crit chance cap.
2022-09-04: Updated block rate formula
2022-07-10: Cleared up confusion on crit chance formula
2022-06-05: Fixed incorrect crit dmg formula and added calculations for summons
2022-04-17: Updated DEF:damage red formula and Izumi/Junko formula
2022-02-15: Added CCpower/res formula
2022-02-01: Updated Crit Formulas
2022-01-02: Added info about attack speed
2021-12-27: Discovered base block rate and formula for block rate adjustment with Yuuka on JP Twitter.
2021-12-18: Discovered stability and evasion formulae on JP Wikiru. Documented and confirmed findings.
2021-12-12: “Finished” first version of the doc

Table of contents

Basic Knowledge

Armour Matchups

Mood

Special Bonus

Bond Bonus

Base Stats

Stats per level formula:

Equipment and Unique Equipment

The Student Stats Page

Definitions

Stats Formula

HP

ATK

DEF

DEF Penetration

Healing

Accuracy/Evasion

Crit (Chance)

Crit Damage

Stability

Range of Fire

Crowd Control Power/ Crowd Control Res

Cost Recovery

Other Stats and Effects

Damage Reduction

Damage Amplification

Level Difference Penalty

Recovery Rate

Block Rate

Attack Speed

Tactical Supports (TS) and Summons Stats Calculation

Buff Duration

Calculating Damage

Auto Attack Damage

Regular Skill Damage

Sub/Misc. Skill Damage

Attack Scaling: Shun and Iori sub skills

Mixed Scaling: Izumi and Junko sub skills

Calculating Healing

Appendix

Stats per level formula:

DEF (old old)

Changelog