【Granblue】 Zeus Testing for Guild War - Warlock vs Berserk + Kuroneko, Kengou, Glory

2019-02-02   #granblue 

With light-favored ancient battlegrounds on the horizon, I’d like to go for the top 1,000 ranking as efficiently as possible. But it is unclear to me what the best light configuration is. So I will do some testing and calculations.

For these tests, I care about 3 jobs that standout as most viable: Warlock, Berserk, Kuronekodoushi, and Kengou. I only care about testing HELL guild war configurations.

General Notes

  • These tests are for myself primarily, but maybe somebody else can use the information.
  • Rank 177 with all jobs max level.
  • For limited weapons, I have 3 Eden, 2 Certificate, and 2 Ganbanteinn.
  • 4凸 THE・STAR, 4凸 Thor, 2x Shiva
  • I am married to 5☆ Song-san, 5☆ Funf, and Io-chan.
  • The tests reflect how I play, which is at 60 FPS in high quality, only reloading long animations like 3-4 chain bursts and enemy triggers.
  • I do not have +99 marks on all of the weapons or summons, please see my grid images to understand my plus marks.

Testing Methods

I will start by testing jobs individually, to find the best configuration among each job. Then I will compare the pros and cons of those configurations. There will be a few testing phases:

  • Motocal Results (add certificate 4凸, no akasha since motocal doesnt handle it well (replace with xeno), add ring values when possible)
  • Grid: Average out 2-3 separate 3 minute trial battle DPS measurements
  • Optimizations: Determine the best mainhand and character composition
  • A single 60 second test on trial battle
  • A real-world fight against Dark Angel Olivia (110mil HP is close to HELL95)
  • A real-world fight against Lich (200mil HP is close to HELL100)

Of course, we do not know what the boss of guild war will do. My goal with these tests is to find the optimal baseline and then work around that depending on the fight’s mechanics.


For all of my time spent playing Granblue, Warlock is by far my most played job. It is the most offensive staff job due to the ability 「Chaser」. It also inflicts reliable blind which is good for konshin builds. Before testing anything, my preconceived notion is that Warlock will remain the best job for me. So let’s test it first.

Motocal Results

Data link

The buff conditions are:

  • 90% Ele Atk (30 Omega MH, 50 Katana stacks, 10 Io-chan passive)
  • 30% DA (Omega MH)

This calc has a major problem that light runs into often. Motocal sucks at handling crit. If you add a flat 50% atk modifier here, it removes an Eden and adds the Omega Bow (I believe in general motocal overvalues DATA weapons and underavlues ougi weapons). Also it should be noted that in reality, light will have a lot more buffs rolling than this. Regardless, I will test this configuration, as well as a 2 Eden configuration due to these interesting results.

Zeus x Zeus vs Zeus x Luci:

Trial Battle Tests

** My grid **

For starters I will test this grid and configuration as a baseline, as it’s what I use the most. Medic is taken as its only purpose is to stack Ouryuu Katana elemental atk to 50% on turn 1 (funf 1-2-3, io2, dj4). The animation in a couple frames faster than Clear All.

Io-chan’s DATA was complete garbage in the third test due to single attack RNG, but it didn’t seem to affect the final result much (I do think I took an extra turn compared to previous tests, though).

Note: I discarded one test where I did three 4-chain bursts which tanked my damage down to 196.96mil. It is clear that 4 chain bursts suck for DPS. Desyncing Io-chan and Song-san usually alleviates this issue as you can get reliable 2-3 chain bursts that way.

Average Trial Damage: 212.3mil

** Motocal Grid **

  • Test 1: 202.8mil
  • Test 2: 194.87mil
  • Test 3: 219.35mil (full TAs on shiva turns, lucky song-san DATA)

Average Trial Damage: 205.67mil

One thing worth noting is that there is an Akasha weapon here, whereas the motocal grid includes no Akasha weapon and instead has a Xeno weapon. Do not worry, Song-san caps in this configuration. She is lagging behind due to the lack of an Omega Bow (her DATA sucks) and not getting the bonus damage from Akasha weapon. It’s clear that this configuration might be strong if you swap Song-san for Clarisse, which I will try later.

** Motocal 2 Eden Grid **

Average Trial Damage: 210.81mil

Before starting these tests, I would have never considered dropping Eden below 3. However, it makes sense, as it allows us to bring more cap up and DATA weapons. This only works with Song-san in the party, as we rely on her Crit buff more closely here. Allowing configurations like this is why we call Song-san the best buffer in light.

Mainhand Testing

My most common mainhand up until a few days ago was the GW dagger. I’d like to test it out using a 2 eden, 2 certificate, 1 ganbanteinn, 2 omega, 1 akasha, 1 katana configuration.

I don’t like this. It doesn’t cap reliably enough. Let’s try 3 eden, 0 ganbanteinn:

3 eden, 2 ganbanteinn, 1 certificate, no omega bow:

This works but it doesn’t seem to be worth the trouble for me, as you need to pay too much attention to when the DATA buffs are up for shiva turns. GW Dagger MH is still respectable, though.

Regardless of all this, it’s clear Song-san lags behind in damage almost all of the time. If I don’t need her buff, maybe I shouldn’t use her. The primary candidates to replace her are Kubira, Metera, Silva, and Clarisse. Let’s test them all with a 3 eden configuration.

Warlock Character Testing


I press all of her abilities. Counter damage won’t be calculated here, so we can assume a real-world scenario would be a little higher. By the way, I haven’t gotten all of her Abi cap up nodes yet, and my ring on her is 15% ougi cap up.

It’s the highest damage we’ve seen yet, and Kubira is not even married. As expected from a Divine General. However, a lot of this is due to her party-wide seraphic buff effect. This gets stronger over time. I noticed in my test that the configuration was a lot weaker than Song-san for the first minute and a half or two. But in the second half, it caught up. This leads me to believe that Kubira is best in long fights when her buff can max out and all her added damage can add up - her autos are actually quite weak.

Kubira seems like a good HELL100 character.


I press abi1. Metera doesn’t work with my grid, because she doesn’t cap autos. She looks great until her chaser drops and you see her single attacking for 350k. It’s possible she gains value in an Akasha Bow team which I will test later.


Silva has been supporting the team form the back line this whole time. What if we put her up front for 5 turns, allow her to burst, and then move her back? I’ll be using my original grid configuration, and after turn 5 or 6 Silva will tag out for her Song-san.

I tried a few different timings for stacking katana, but in the end, the momentum lost from switching in a character who has no buffs is too great in this buff-heavy build. Silva’s end of turn damage tries to make up for it after the switch, but it’s not enough. Also, her ougi isn’t even that much better than a juttenshu’s anyways. Switching her out for Io-chan instead gave me 185mil, as that only extrapolates the buff momentum issue.


Finally, testing the on-spec ougi nuker.

I press abi1 and abi2. I tested her 3 times as she performed way worse than I expected, so I wanted to make sure I wasn’t doing anything wrong. In the third test I allowed her to get an additional buffed ougi, which did make her the most damaging member of the party, but it wasn’t enough overall. Maybe if she were married to me it would be another story.

Omake test: Io-chan positioning

I’m going to retest my first configuration with Io-chan in slot 2 instead of slot 4. Djeeta-Iochan-Songsan-Funf.

I’ve always thought this, but Io-chan being in slot 2 or slot 4 doesn’t seem to really matter. You either get slightly better uptime on her buffs for all 4 members, or considerably better uptime for 2 members, but in the end, it always seems to balance out.

I’ll keep io-chan in slot 4 since it’s what I’m used to.


Normally I don’t reload, but I do play in 60fps. Let’s try in lite mode with reloading, too.

My no reload tests with this configuration were 207-214mil. So far, this is the biggest surprise in testing yet. I lose damage reloading. This is because I live in California, USA so I do not have Japanese ping.

It’s important to note that the trial battle monster does not attack, so I am not skipping mob animations when I reload. It’s possible that if a monster has a long attack animation, reloading is considerably better. But this is still useful information - if a boss is paralyzed, it seems I shouldn’t reload except on ougi turn.

60 second test

I’ll change my summon configuration from 4凸 Thor, Shiva, Shiva, THE・STAR to 4凸 Thor, 4凸 Lucifer, Haruto・Maruto, THE・STAR and do a 60-seconds test.

As expected, the build doesn’t have time to make use of its buffs or even make use of 4凸 Thor, so in a burst rush, it looks a little weak.

DAngel・Olivia test

Lich test

Lich has higher defense than I thought.

Omake test: Godstone x Elemental

A 3 minute Lucifer test to see how bad it is if I don’t get support summon Zeus. Compare to 212mil.

Warlock Conclusions

  • Kubira and Song-san are both great options. If my Kubira were ringed, she would be a clear winner in long fights. But she isn’t, whereas Song-san is. In real-world scenarios, Kubira will take a lot of damage, but counter a lot as well. Song-san puts many debuffs onto the boss which can be very good. It’s hard for me to determine which is best, though if the fight is short, I think Song-san is best.
  • 2 Eden is actually viable, but you must pay closer attention to buffs and ougi timings, so it’s a bit of a hassle.
  • I like my original grid the most, but if I ever swap out Song-san, I should swap out the Omega Bow.
  • Double Zeus is really good, maybe even worth the summon reroll time.


With Certificate being such a strong mainhand, it’s possible that バーサーカー is at the top of the light meta now. Let’s test it out.

Motocal Results

Data link

The buff conditions are:

  • 50% Ele Atk (Katana stacks)
  • 30% DA/TA (Certificate MH)

Use all limited series weapons and an ougi cap up omega weapon. I didn’t add in stuff like Rage IV, because I’m really only using motocal to get a grid baseline here.

Trial Battle Tests

The first configuration seems clear:

  • Test 1: 189.24mil ( I tried less emphasis on chaser damage since no Akasha)
  • Test 2: 185.09mil (Shiva in Thor chaser, lost the SS because I reset it becaues it sucked)

Average Trial Damage: 187.16mil

This isn’t working. Let’s try adding in Omega Rod instead of the second ganba, to match the first warlock grid.

Average Trial Damage: 196.61mil

Berserk Character Testing

It seems for Berserk we may want to switch strategies. Let’s try with Kubira instead of Io-chan, and remove the Omega Rod again.

Average Trial Damage: 189.86mil

This doesn’t work either, although it is interesting to note that Kubira does a lot more herself with Rage IV around.

I don’t want to give up on Berserk just yet. How about with Io-chan/Kubira and auto gauph key? This should be the maximum cap breaking build.

Average Trial Damage: 197.44mil

This configuration looks best for Berserk, but it’s still not very good compared to the warlock on trial battle.

However, Berserk has a major advantage in the 40% DEF buff that certificate mainhand gives on ougi. This should not be discounted in a konshin build.

For ancient battlegrounds, though, this probably isn’t that important - you can maintain konshin easily with all the pots it gives you.

60 second test

It’s about the same as the 54mil that Warlock got.

DAngel・Olivia test

Lich test

Berserk Conclusions

  • Berserk performs worse than I thought, although it did fine in the Lich live test.
  • The DEF UP buff from Certificate might be the only selling point of using the job.
  • Higher raised cap and base damage from Rage IV isn’t worth losing 20% chasers overall.
  • I’ll still do 1 or 2 hell95/hell100 as Berserk just to test the final times, as the results are still respectable, and live battle will change a lot.


Possibly the most bursty damage job in the game, Kengou seems more viable on light due to new bow/gun characters and the release of Akasha bow. Let’s test it out.

Motocal Results

If we force in a Kaneshige, it wants us to drop Omega Rod from the grid.

Btw, if I don’t force in Kaneshige, it adds an sk15 chev sword. It’s possible though that a build with Funf and Ouryuu Katana is better than chev sword. I’ll test that too.

This setup is pretty devoid of buffers and buffs - but even without Ouyruu Katana, Zeus luci doesn’t seem the way to go. I’ll test it anyways.

Trial Battle Tests

For starters I’ll try the motocal grid, but with Xeno Sword swapped for Akasha bow.

I won’t bother continuing down this line - it can be improved.

Let’s try with an ouryuu katana instead, so a configuration like this:

It didn’t improve things very much.

Kengou Character Testing

What if we drop the katana, go back to another ganbanteinn, and never swap out Silva? (We’d lack a healer in a konshin element, but it’s ancient battlegrounds, maybe that’s fine?)

Average Trial Damage: 190.78mil

Metera does a lot of damage on an Akasha bow grid, but Silva and Song-san lag too far behind.

What if we swap Silva into Io-chan?

Average Trial Damage: 196.98mil

This seems like the best so far.

What if we drop Akasha Bow, add a Xeno Sword, and just use Song-san/Funf/Io-chan?

Selfish Djeeta steals all the damage for herself.

Kengou Mainhand

Akasha bow mainhand doesn’t work well with Io-chan because their cap up buffs do not stack. So I’ll test Akasha mainhand with a swap into Funf.

That’s actually garbage, but I’m not surprised. Kaneshige is very strong.

For the sake of completion I’ll test omega bow mh as well.

I am surprised this hit a normal number, but it’s not worth dropping Kaneshige for.

Omake Test: Kengou Zeus x Luci

What if we run Song-san/Silva/Metera in a Zeus Luci build?

I was able to take an extra turn from not hitting healer abilities.

60 second test

This is where I expect Kengou to shine. Let’s see how it does with a Song-san/Silva/Metera configuration. It’s the no katana grid:

As expected, Kengou does the best performance in the 60 second burst sprint.

Kengou Conclusions

  • It’s best on short bursts where mumyouzan can have a large impact and Silva’s burst damage is significant.
  • However, in a longer fight where other configurations can get their buff and summon setups rolling, Kengou will lose over time.
  • It’s a strong class for pubs and HELL90, but I won’t use it on HELL95 or HELL100 solo/duo.


So far I’ve determined that Warlock is the most interesting job to me. In the past, I valued the Quatre dagger a lot as a mainhand. But I don’t like Quatre dagger anymore. So let’s test Kuroneko, which is similar to Warlock but has perma chaser, no party chaser, more powerful ougi, and a slightly more hassle playstyle.

Trial Battle Tests

I’ll use the Warlock configuration that I determined to be best:

I won’t press Abi1. I did one test run with Black Charm on, and it did something like 150mil, so I discarded it immediately. Instead, permanent Djeeta chaser and using MP for a raised ougi cap is interesting.

Average Trial Damage: 217.85mil

Compare this against 212.3mil for the same grid with Warlock. It appears that Akasha weapons and even higher ougi caps have powered up Kuroneko again.

Kuroneko Character Testing

How does Kubira do with the Black Cat Mage? Remove Omega Bow and add another ganbanteinn.

Kubira remains viable on longer fights.

Kuroneko 60 seconds

This surprises me - it beats Kengou. I did not expect this result. The echo damage is very significant…

Kuronekodoushi Conclusions

  • It’s the strongest job for everything. But it requires more effort to play (managing MP and tashinjin healing is a pain)
  • Djeeta in the kuroneko doushi outfit is very cute.

Omake Job Test: The・Glory

I’d like to quickly do a Glory test. I don’t have the full setup for it, but I’ll try with what I have:

I didn’t really think so.

Omake Job Test: Ouryuu Spear

I’m not setup for Apsaras (I don’t have zenith on that job) so let’s try with Sage, allowing us to keep Omega Rod anyways.

It’s an OK setup, but I don’t feel compelled to use the spear or Sage on easy content. But we’ll see if Sparta gains value if Phalanx is needed in the real fight.

Table of Results

Job Configuration Result
Warlock 3 eden / 1 ganba / 2 cert 212.3mil
Warlock 3 eden / 2 ganba / 2 cert 205.67mil
Warlock 2 eden / 2 ganba / 2 cert 210.81mil
Warlock gw dagger mh 208.97mil
Warlock kubira 221.63mil
Warlock metera 192.49mil
Warlock silva 191.53mil
Warlock clarisse 195.17mil
Berserk 3 eden / 2 ganba / 2 cert 187.16mil
Berserk 3 eden / 1 ganba / 2 cert 196.61mil
Berserk kubira 197.44mil
Kengou 3 eden / 1 ganba / 2 cert 190.78mil
Kengou io-chan 196.98mil
Kengou io-chan & funf 189.67mil
Kengou akasha bow mh 153mil
Kengou omega bow mh 187.07mil
Kuroneko 3 eden / 1 ganba / 2 cert 217.85mil
Kuroneko kubira 215.27mil
The・Glory 3 eden / 2 ganba / 2 cert 172.06mil
Sage ouyruu spear 188.8mil
Warlock 60 Seconds Burst 54.91mil
Berserk 60 Seconds Burst 55.12mil
Kengou 60 Seconds Burst 62.68mil
Kuroneko 60 Seconds Burst 69.26mil

Overall Conclusions

  • The winner is: Kuroneko Doushi!
  • 3 minute trial battle test: Kuroneko 217.85mil / Warlock 212.3mil.
  • Kuroneko vs Warlock is fairly close. Kuroneko does a little more damage, but it also requires paying more attention and trying a little harder to not screw up your MP management or heal CDs after tashinjin turns.
  • 60 seconds trial battle test: Kuroneko 69.26mil / Kengou 62.68mil.
  • Kuroneko is best for quick burst play, but Kengou can also do it fine.
  • Berserk isn’t really interesting to me unless the 40% DEF Up buff from Certificate Mainhand is useful.
  • Metera underperforms for me, but this is likely because I do not have 6 Eden. If Metera caps autos, she would be really good.
  • Kubira is good on longer fights. I don’t know if it’s worth using her over Song-san for me, because I am married to Song-san, and Kubira requires more button presses.
  • I didn’t test any fist/axe builds with Mahou Senshi or Wrestler, because I don’t have Six (so Akasha Axe is a pain to make), I haven’t unlocked Mahou Senshi, and I don’t have levion sisters yet anyways. It’s possible those builds are good, I expect them to fall somewhere in the 180-190 range when optimized.
  • I have a lunatic broom but it underperformed, the grid is too crowded.

Based on this testing, in guild war, I will try to use Kuroneko Doushi and Warlock, both. Depending on whether I want to try harder or not at the time. In pubs I will use Kuroneko or Kengou. My grid will be the following, original grid I use:

Thanks for looking.

Post-GW Followup

I finished Light Ancient Battlegrounds at individual placement #112 and my guild went 4-0. Here are some thoughts.

  • For true top 10 ranking, guild feeding you meat & shiva/buffs is important. HELL95 is highly efficient, while HELL100 requires lots of shiva/thor alt spam to be worth it.
  • Mahou Senshi does not require Akasha Axe or Levion like I predicted. It works the same as the old fire Mahou Senshi build with esser from last fire GW - simply mash to the setup turn with ougi off, then do 65m+ in a single turn with ferry/thor/shiva. This is why HELL95 is so fast.
    • I personally believe Mahou Senshi is fundamentally broken for this HELL95 format, and I will make a tetrastreamer for every element that has a decent chaser setup now. However, as I like doing HELL100, I will still use HELL100 setups on day3/4.
  • The Mahou Senshi build normally would use Ferry (faster damage cap up than Io-chan, guaranteed TA, chasers). The problem with Ferry is that she is weak and does not cap easily herself. This means you need 4+ Eden, full plus marks in grid, pluses on Ferry as well. Marriage helps a lot too. My ferry was too weak.
    • Metera works similarily. She works as long as you have enough Eden and plus marks/marriage to back her up.
  • To that effect, for GW, marrying weak characters is better than marrying strong characters.
    • For example, a ring helps Ferry a lot more than it helps Funf. Funf can already cap with minimal effort, but weaker characters need help to hit cap. A ring on a strong character is only worth 5% damage, but ringing a weak character can be worth 15% if the unique ATK mod helps them hit cap faster.
  • I used Kuroneko the entire time with Io-chan/Funf/Kubira. Backline is Yaia (マジ天使) and トラちゃん. Song-san does not make the cut as she doesn’t really bring anything. I would like to ring my Kubira, as this character is clearly incredible.
  • Kuroneko is inferior to Mahou Senshi on HELL95, but I think it’s superior on HELL100 (assuming no 4-alt shiva setup). Mahou Senshi tapers off in longer fights where the 1-2 turns of setup aren’t enough to outright kill the boss.
    • We saw this day 4 when たま猫さん, the second ranker, started catching up to 星宮いちごさん、 the top ranker. 星宮いちごさん stayed on Mahou Senshi the entire time and established a giant early lead, but たま猫さん’s kuroneko really started catching up on the later days. Given a few more hours, たま猫さん may have won with neko.
  • Many people are saying Certificate is not good after this GW. Solely for GW with low latency, it’s true. If you can have more Eden, that’s better, as it enables the weaker characters with important chasers, and Light does not have strong ougi characters to make the lockout timer worth it.
    • Personally for me, I would not trade my Certificate for more Eden. GW is only 1 week a year, and the certificate are amazing in general play, especially for Akasha/Ubhl. Also with my ping, ougi is not the worst, and I rely on Io-chan ougi buffs anyway.
  • Regarding general exhaustion for top 1k, the hardest thing for me was waking up on time because I love sleep. My friend/guildmate had to call me every morning. Also, I noticed it can be easy to become completely absorbed in the game, so I set alarms to remind myself to take breaks and take care of myself. After this, it was no problem healthwise and was quite fun for most of it.
  • Overall, I am happy with this guild war. Kuroneko Doushi is my favorite class again.

artist: Kyu