All Jobs Damage Simulator And Gear Sets

Sir, have you tried um, running the application that this thread is based on?

does the sim now include Alabaster and Murky equips?

Yes, I've just now pushed an update which adds R30 Murky and Alabaster gear, Archduke's Sword/Shield, and Blurred Knife.

You should only need to update the gear.py, item_list.txt, and icons32/ files to begin testing these.

Thank you ^.^ We are lucky to have you :D

Sorry to bring this up again but how do I check nuking using sim again?

I've been using it for months (it's great) for different jobs but haven't been able to see calculations for magic related stuff, (weather it's the damage of single nukes, MB or continuous casting DPS).

I noticed there is no option for drg main to add spirit bond/link JA haste in the sim, could you please add it Please, Much apperciated ^.^ , Thank you.

It already assumes you have wyvern + max spirit link bonus.

if self.main_job == "drg": # Bonus stats for having a fully leveled Wyvern pet:
self.stats["Wyvern Bonus Attack%"] = True # This represents the +20% attack that will be applied later for having a wyvern out. Additive bonus with smite, berserk, chaos roll, etc.
self.stats["Weapon Skill Damage Trait"] = self.stats.get("Weapon Skill Damage Trait",0) + 10
self.stats["JA Haste"] = self.stats.get("JA Haste",0) + 10
self.stats["DA"] = self.stats.get("DA",0) + 15
self.stats["Attack"] = self.stats.get("Attack",0) + 40 # +40 attack for having a fully leveled Wyvern pet with 20/20 Job Points.

Thank you very much for the confirmation :D apperciate it!

can someone make a set for fast blade II on geo?

Just because it has been a while... thank you for everything you've done with this Izanami. Glad I got to contribute some feedback, but you did all the hard work. I'm still using this and recommend it to those who have never heard of it.

Stick to black halo or judgement you will do a lot more damage.

Just stack Nyame and call it a day.

I think I found a bug in the simulator.

TL;DR I ran a test to determine if Multi-Attack (MA) is checked against the MH or the OH weapon in weaponskills where the weaponskill starts with multiple hits. I determined that MA is checked against the MH weapon when the WS starts with multiple hits. The simulation appears to do the opposite. This can lead to incorrect damage calculations for multi-hit WS, especially ftp replicating WS.

Why this matters: OH isn't always as strong as MH, especially when using a magian weapon. Checking MA against the OH instead of the MH can lead to weaker MA hits.

Background knowledge:
Fast Blade II is an ftp replicating weaponskill that hits twice with an ftp of 5.0 at 3000tp. With an offhand weapon, it should hit 3 times before multi-attack procs. However, the 3rd hit (offhand hit) will be weaker when using a magian weapon.

Context:
I have been testing FBII with capped pdif at 3000tp on THF using centovente and noticed 3 major groupings of ws damage. No TA proc (0) , 1TA proc, and 2TA procs => WS that hit 3 times, 5 times, and 7 times. I also accidentally subbed war during the last test, so I had some double attack in my set, so there were a few 4 and 6hit WS as well.

I only got 192 samples, but it looks like this:You can see 5 "peaks". 1 for each number of hits between 3 and 7 inclusively. 5 and 7 hits are most common because my set has 61% TA and only 12% DA. Please note that the mean (the peak in the middle, representing 5 hits) has numbers converging to a single center.

This is notably different from the distribution of values shown by the simulation.You can see several, almost "mirroring" peaks near the mean. Most notably, there are 2 equal peaks in the center, representing an equal chance for 2 distinct values. This is because, in the sim, the magian weapon and the main weapon have equal MA chances. If the magian weapon MAs instead of the main weapon, it leads to lower overall damage. However, this doesn't seem to actually happen in the game.

My analysis suggests that the 2nd hit of "2-hit" weapon skill should be checked first, meaning that the OH shouldn't be checked for multi attack at all for FBII. (This is probably also true for any other multi-hit ws).

Why this wasn't noticed before:
Before FBII, it didn't really matter. There wasn't any ftp replicating WS with a high enough ftp value that could also be paired with a magian weapon. H2H/2H weapons can't use magian offhands or ranged weapons, and dagger replicating ftp values are too small with too many existing hits for anyone to notice.

Here is the raw data for my test, if you want to scrutinize it.

Good morning. It would be super cool if someone could verify or refute my claim. Love this community and all the work that's put into it!

*Edit to reduce vertical space from quoting myself*

Thanks for the note, but I think you're simply misunderstanding how multi-attacks are checked for weapon skills.

For completeness and to be sure everyone is on the same page: Here is the FFXIAH post with the discussion I used to base the simulation multi-attack order off of, which I roughly confirmed through in-game testing by watching TP return from many weapon skills. Note Lili's post suggesting that only the first two attacks per attack round may multi-attack. Even if those two checks fail, none of the remaining hits are checked. Austar's post a bit further down provides the explicit order of attacks for multi-hit weapon skills:

1. First main-hand hit

2. First off-hand hit (only if dual wielding)

3. All other remaining main-hand hits (+4 main-hand hits if using a 5-hit weapon skill)

4. First main-hand multi-attack check

5. First off-hand multi-attack (only if dual wielding)

6. Second main-hand multi-attack check (only if the weapon skill has more than one natural hit and not dual wielding)

I've double-checked the code itself and found no issues with its handling of this order of operations (see lines 1848 - 2203 of actions.py).

You can confirm this yourself in-game by finding an enemy with enough HP to survive two hits and checking your TP return while dual wielding and using a weaponskill with at least two main-hand hits. The first main-hand and first off-hand hits (steps 1 and 2 above) will gain full and equal TP return, while all of their natural and multi-attack hits gain 10*(1+store_tp/100) TP. If the enemy is defeated after the second hit of the weapon skill, then you'll either receive: (2*first_hit_tp) if the off-hand was the second hit, or (1*first_hit_tp + 10) if the main-hand's second hit is applied before the off-hand, which is what you're suggesting.

So it's not that "in the sim, the magian weapon and the main weapon have equal MA chances" as you said. But that, when dual wielding, both the main-hand and the off-hand will always get one multi-attack check each. Their damage distributions should have the same shape, but be offset due to their weapon damage differences (and WSD on the first main-hit). If you're comfortable with Python, you can modify the code to plot the total damage for each weapon separately. I've done this using your equipment for apex bats below (attack capped but no extra PDL buffs).
I'm not sure if 192 samples is enough to find meaningful differences in simulation vs observed for FTP-transferring weapon skills with many peaks. But I am confident that your Gaussian kernel bandwidth is far larger than it should be to enable any comparisons in your plot. Recall that melee attacks in FFXI have a 5% randomizer applied at the end of their calculation. Your bandwidth is creating KDE distributions with widths of about 20% or more of their mean values, which causes significant blending issues.


I agree with this. I try not to check FFXIAH daily when I'm not subbed, so I may be slow to respond sometimes.

I agree with this completely. What I am contesting is what counts as the "second hit" in a WS that natively has more than 1 hit in it while dual wielding.

I agree that this is what the sim is doing. I don't agree that this is correct based on what I've seen. I think it should be as follows for any WS with more than one hit like FBII.
First main-hand hit

Second main-hand hit (only if using a WS with 2 hits)

EDIT* I don't know about the remaining order, so I am not going to assert that this order is correct, only that the MA procs are checked against the main hand twice.*
Dual wield hit

All other remaining main-hand hits (+4 main-hand hits if using a 5-hit weapon skill)

First main-hand multi-attack check

Second main-hand multi-attack check (only if the weapon skill has more than one natural hit)

First off-hand multi-attack (only if dual wielding and if the WS has only 1 hit pre-multi-attack)
*END EDIT

We can actually see that dual wield is not required for checking the 2nd MA procs in FBII in the sim.


What is the difference between the first and second statement?
If I have 60% TA, the first hit (main hand hit) has a 60% chance of adding 2 additional attacks, and the second hit (offhand hit) has a 60% change of adding 2 additional attacks. This is "equal MA chances" which is the same as saying "both the main-hand and off-hand will always get one multi-attack check each".

I understand your skepticism over a relatively low amount of data, but I still think it has merit. Please note how the data for the "5 hits" area of the graph converges around a single center in game instead of 2 centers in sim.


Edit: I don't want to make any assertions about order of hits after the second hit. I only want to assert that the first 2 hits in FBII are main hand hits.

Edit2: I failed to qualify something in respect to dw vs not dw clearly in the MA checks. Fixed.

Looks like I misunderstood your first post and that you're challenging the currently accepted model for multi-attacks on weapon skills while dual wielding. It sounds like you're suggesting an alternative:

• All weapon skills get up to two multi-attack chances.

• Single hit weapon skills get one chance.

• Single hit weapon skills while dual wielding give the 2nd chance to the off-hand hit.

• Multi-hit weapon skills give both chances to the main-hand (leaving nothing for the off-hand).

I'm not fully convinced of this by your post and would need to see a lot more data and testing, but this is probably worth investigating since the simulations at least appear to not match the shape of your 192 samples distribution.

One question to answer is: "Does an off-hand weapon use one of the two multi-attack chances during multi-hit weapon skills?" The brute-force and easy to interpret approach to answering this is to use a few thousand 3k TP FTP-transferring weapon skills with a strong main-hand and a weak off-hand while attack capped, then compare the output to simulation. This would at least show that the simulation is wrong somewhere and allow us to tweak things til it matched the observed data.

You could probably come up with a few other tests involving comparing results with identical situations but swapping the main+sub weapons (weak main-hand, strong off-hand), intentionally minimizing accuracy, maximizing multi-attack, and/or watching damage and TP return together. I'm not subscribed right now, so any testing I do on this will have to wait until the next free login campaign if I have time for it. I'll keep this in mind and comment again if I find time to do some testing during the next free login.

I responded in other thread that all you need is single test with 100% DA and high damage sword in main and 1dmg offhand. Ideally as close to being naked as possible to lower WSC (makes weapon damage more impactful) and make 1000tp fast blade (it doesn't have ftp transfer, but at 1000 tp it has 1.0 ftp for all hits including first). Damage will either fits 4x main hit and 1x offhand hit or 3x main hit and 2x offhand hit.

Just curious - Have the Hoxne items been added to the simulator?

Very curious to what this will do for WS sets for the ammo using fTP replicating WS's

I added it locally for Howling Fist.

Select ALL (including Hoxne Ampulla / Earring, and using them of course)
- WS: Avg Damage = 55944.7, Avg TP return = 284.4TP
- TP: Avg. WS time = 2.264s, Avg TP per round = 848.0TP

Select ALL (excluding Hoxne Ampulla / Earring)
- WS: Avg Damage = 42620.6, Avg TP return = 266.5TP
- TP: Avg. WS time = 2.994s, Avg TP per round = 641.3TP

So about a 30% boost in WS damage and frequency. If WS frequency is most of your damage, which it is, even on MNK, then you're looking at a >40% increase in DPS.

Avg TP per round feels too high, but perhaps I don't know what it does. Eh . . . actually maybe it's right if it's building SAM roll in or something. So Ampulla MNK has a pretty good shot at having a 1-round build with the right buffs?

I assume you were using just the ammo and didn't change anything else? I'm wondering as we look more into this if we will end up adjusting the entire set to focus on other stats after reaching DA cap.

I already did it same day that ammo poped up. It was +53%dps (I wrote ~50%). I actually did it for most jobs and wrote it here.

I remembered that post, but I figured they was asking here for the results of this particular simulator.

Btw, if it wasn't obvious, both Hoxne Ampulla (DA+100%) and Hoxne Earring (all stats +25) get used, but Hoxne Ampulla adds like 50% DPS and Hoxne Earring adds like 1%


No, I let it do its normal optimization thing. It only does 4 iterations, so I'm not completely sure how it works. If I was implementing it, I'd probably use something like this, but the number of iterations probably wouldn't be 4, then.

The point is I was using this particular simulator ;)

Bloodpact integration could be possible, please ?