All Jobs Damage Simulator And Gear Sets

I think you are completely missing the point here. You are not simulating this to see dps in actual 1min fight. You are simulating 20h of fighting to notice marginal differences between sets. There is too many random variables in 1min fight to draw any conclusion. For example you can simulate 1min fight and have 20%DA set proc DA 15 times in 50 rounds and 30%DA proc 13 times in 50rounds. Does that mean 20%DA is better than 30%DA? obviously not. That's why you do 20hours test to exclude the effect of randomness. It doesn't suppose to simulate one 20h fight, but 1200 one minute fights.

I would rather run a fight with set HP a few thousand time until I get a comfortable with the variance. Or at least have the option. Sometimes you want damage in a vacuum but other times you don’t.

If crits are simmed instead of averaged then the amount of runs you would need greatly increase.

What? Why? Could you explain, using math, how any of what you're saying works?

A simulator that rolls a random function 10,000,000 times and checks it against your critical hit rate is functionally equivalent to taking an average.

If you believe there's some difference here which is statistically significant, I would love to hear about how that works.

I get that you're trying to talk about randomness and variability, but when you're running a 24 hour test with literally millions of auto-attacks, I think the law of large numbers takes over

This is a tool you should use and draw your own conclusion. It's not suppose to completely simulate Odyssey C run including travel time between mobs, getting debuffed and human mistakes. There is some level of your own thought process, that will connect sim results with specific scenarios in game.
Ofc you could make a much bigger tool, that will even use AI to make decisions, throw debuffs and everything else, but I doubt anyone will take time to make one, unless in few years, when you can automate it with AI :P

Fair enough, then I stand corrected. I'll sit this one out then lol. I don't actually have experience using the sims anyways.

You also basically said what I said, so it’s confusing. Yeah when you are simulating something like crits you need more iterations to come closer to a conclusion. But we are simming things like crits, multiattack, zanshin etc. with each variable multiplying the required amount of iterations. Average crits and damage would just reduce the amount significantly.


Yeah I’ve been using the tool liberally, I just think that like Time to Kill (TTK) is a good thing to sim especially when we are talking about like empy am3 where we may see swings in tp sets.
Not really looking to sim running time etc. it already isn’t simming like DRG jumps or SAM meditate.

This is what excels are for imo. Everything there is based on averages.

Oh I love monte carloing sims, just the sample size starts to get exponentially larger with more and more factos.

I used to have a dnc sim for ffxiv and I would have to run a 6minute fight 10,000 times to get consistent results.

I love this calculator. Over the past few days, I have been refining my TP and WS sets with the Null equipment, and they are getting used more than I anticipated. It seems the following is true about the Loop and Shawl.

- If accuracy is an issue, Loop and Shawl will undoubtedly come in, but even when it isn't, they allow for alternate ways to build sets. For example, in a DRK hybrid dt-tp set using full Sakpata, you can combine Null Loop with Moonlight Ring, which allows you to use the Shawl and get better TP flow. No need for a cape with PDT or a Defending Ring.

For lighter jobs like THF COR and RDM (with /nin), Null Shawl will generally perform better than a Dual Wield cape, as you can combine it with Reiki Yotai and Eabani/Suppa to cover that. This frees up a mantle to be augmented with other things. That is, unless you are fighting without capped magic haste.

- Weapon skills can also benefit from this, especially on jobs with accuracy lower than other jobs. DRK is known for this, so even having one Null piece in a WS set opens up possibilities.

Of course, this all depends on target and setup. Accuracy is hardly ever an issue these days. I'm still trying to find the sweet spot on just how much is enough. I don't use Gearswap, so it's not like I can have 7 or 8 TP sets for niche situations.

If anyone can give feedback on how much accuracy you are looking for in TP and WS sets across your jobs using this calculator, please let me know. Obviously, it is going to change from person-to-person, but it would be cool to share experiences. Thanks.

I don't think accuracy is an issue for most jobs, especially for DRK and especially in party. If you solo on DRK you will generally use Apocalypse or Foenaria. Both having lots of accuracy. If you are soloing with Caladbolg for some reason, you can still use absorb Dex and Acc with Liberator, but it would be weird to use Caladbold if you have accuracy issue. With Apocalypse, even without buffs you are capped on Catastrophe with standard damage oriented WS set against Apex cog (1551 evasion) and that's before absorbs. Same for TP set, but I guess Loop is an option if you really need accuracy in TP set. Shawl is bis back piece for TP on DRK regardless if you need accuracy or not.

Jobs that struggle with accuracy are jobs that use TP bonus offhands and missing native high accuracy like BRD or NIN.


For THF it's kinda meh. Thf has 50%+ TA, so DA efficiency is halved at best. On top of that THF has massive TA damage (+20%) on Ambu cape. For RDM and COR Shawl is imo bis (even with RDM also having massive TA rate), but I would go /dnc on both with Suppa+Eabani and haste samba, not Reiki.

The biggest issue with Shawl is capping PDT w/ it.

In a lot of fights, you don't need PDT though. It just more-so doesn't replace the need for a TP JSE cape.

50 meva > 10 PDT in majority of content that you care for defensive stats.

Yeah 100%, was just stating that it doesn't completely replace ambuscade cape for PDT situations though you could probably find some reasonable work around at minimal loss if you wanted to reclaim inventory space.

Yep, that's why I recommend combinations like Null Loop and Moonlight Ring, or Loricate +1 and Crepuscular Pebble to reach the PDT cap. Let's you use Null Shawl and other better pieces.

Also, it should go without saying that any jobs using Mythic or Ergon AM3s will probably still prefer an Ambu cape with Store TP.


Agreed. I mean, the calculator lets us figure it out for any situation. We'll probably start seeing more jobs like BRD and NIN offhanding these weapons now, as they are so much easier to gear for, and the calculator will let them know how much is needed.

For BRD and NIN Shawl is a nice upgrade, but Loop not really. BRD neck is way too good with 43 accuracy, 7stp and 3%QA and same for NIN with 36accuracy, 7stp and +25% daken. Null belt is nice for them if you need accuracy and survivability at the same time. Meva can be pushed quite high with addition of Null back and waist.

Are BLU traits supported?

I've updated how the code calculates enemy defense when applying defense reduction spells and abilities based on some recent in-game testing. I've also updated the BG Wiki page for Defense Down to clarify how defense reduction stacks.

The previous version of the code assumed that defense reduction spells and abilities stacked multiplicatively, which was an old assumption of mine to prevent defense values from dropping below 1, since I was unable to find any data which suggested otherwise.
This change also included updating defense-ignoring weapon skills (ie Blade: Kamu), which I assume stack additively with defense reducing spells and abilities to a minimum of 1 defense, but this is not easy to test.

I recruited Wotasu (from the WHM guide and the SMN guide) to cast some enfeebles on me in Brenner. Our testing suggested that the defense reduction effects of Dia, Geo-Frailty, Box Step, and Defense Down (Shell Crusher in this case) stack additively to a minimum of 1 defense. For completeness, here are the data from our testing.
I've also added Armor Break (-25% Enemy Defense) and ensured that Angon, Armor Break, and Swooping Frenzy can not be activated together.

how do we add additional enemies/targets, tried editing enemies.py but they dont show up

I'm not sure if adding new enemies was technically possible before now if you were using the executable version of the code.


For simplicity, I've just updated the enemies.py file to make it function the same as the gear.py file for modifications. Simply make a copy of one of the existing enemies and adjust the copy's Name, Location and stats. The new enemy will appear in the Enemy list of the GUI the next time you open it.

You'll likely need to download the updated code from GitHub to do this, though. See https://github.com/IzaKastra/wsdist_beta, or sign in and check the Actions page for the GitHub-generated executable to be safe if using the executable file instead of the .py files.


Edit: The enemies.py file was updated based on the pull request from Austar from many months ago. He also corrected the AGI stat for the enemies already in the code based on known enemy stat tables using their Job and Level. The current version of the enemies.py file now also includes these corrected AGI stats compared to the previous version.

First off let me say amazing work and thanks for doing this. I long wanted to move away from spreadsheets but a combination of being too lazy to do it and spreadsheets made it easy for others to check my work.

I've only played with it a little bit, but I'm getting reasonable looking numbers, my main interest isn't to find the "best" but improve my existing gear and target effort/reward upgrades. (or perhaps save some inventory/gil for things which are effectively side grades)

I'd suggest getting rid of print() statements to stdout, its slow, really slow and when you do thousands of lines that way the slow down is noticeable. I opened a file handle for the simulation and it completes almost instantly instead of taking several seconds to do.

In fact the ability to write a summary file might be nice (could list all relevant details (job/level/gear/enemy/buff/stats and results) then you can just write a file, change something write a different file and diff them. (np++ compare extension) (for that matter could probably use xlsxwriter or openpyxl to create a spreadsheet directly and highlight changes)

For some stylistic commentary you'd probably be better off using https://docs.python.org/3/library/importlib.html#importlib.import_module instead of modifying the path and doing import * there after (certainly true from a security point of view although I'm not sure that it matters for this, although I suspect the real value would be protecting against people who don't really understand python doing dumb things). Even better than that I'd suggest migrating the enemies and the gear to a sqlite db and they can be downloaded independently from the main code (i'd also suggest a 'hidden' option so for iterating the scope can be limited (whether its getting rid of lesser gear or gear that seems unattainable is an exercise left for the user, but at least iterations/runtime can be limited). Users could use sqlitebrowser to add rows to the table.

Also for anyone editing gear.py I'd suggest (for simplicity) the following the default file is:

if you want to add head3:

works and you don't need to assign the dictionary to a variable and add the variable to the list. (i.e. 1 line edit instead of 2)

I can probably submit some of these changes to github if you like, but I don't want to put a bunch of effort into if you don't like the direction (or would rather do it yourself)

So I'm on rng and trying to put together an Apex arrow set and it's showing me empty sub when /drg instead of nusku shield? Is that just because I'm attack capped so nusku wouldn't actually up damage?

Sounds like you're running the code in Verbose mode. You can disable Verbose and Very Verbose in the Settings menu at the top of the GUI. Doing so will eliminate nearly all print statements.

I like the idea of having a quick way to directly compare all stats (and damage) between two sets. I'll look into how I would want to implement this later.

I think I'm pretty decent at Python now, but I do not know what a lot of that means. I've never used importlib, and at first glance I don't immediately recognize the benefit. Clearly its useful because it seems to be popular.

As for moving gear and enemies to an SQLite database, I don't see that realistically happening only because I think it may require a decent amount of work to the main code, which is already a mess and why the file is still in "beta". Maybe on my next full rewrite I can start in SQLite.



That sounds about right. You can equip the set it does find and compare the damage with/without Nusku Shield with the Quicklook buttons. If there is no increase in average damage with the shield, then the code would choose not to equip it in the "best" set. This is also why the code finds TP sets with INT or similar augments on ambuscade capes (already capped hit rate without DEX augment). A lot of the time it's best to lock the main and sub slots anyway.

I found a case where it gave me Pareto sub-optimal choice (from the given parameters/values)
I increased the need for more -xDT and it spit out a set with better offensive traits than the previous whie ladhering to the higher need for -xDT.

Interested in a report if if find one again ? or anything that i need to do to report it in a helpful way ?

I've already done this, in fact I have a wrapper for print so everything goes to a file, and I created a summary file (it creates a sim_<num>.txt file everytime the button is pressed and now also creates a xref summary file for every row that is different with the initial iteration. I was planning on submitting the changes via github to you but rather than using my excel idea I was just going to see if I could pop up the correct widget via tkinter and put the the data directly in there. I've only been tinkering an hr. or so a week on this so its slow going, but if you want me to submit sooner rather than later I will (save you the trouble). Everything is done but the gui part.

I also wanted to let the user save a profile per job, but I haven't gotten this far.

I can probably give you sqlite code too, to just convert your dictionaries to sqlite and back is probably only a couple dozen lines of code at most.

If you can reproduce it every time, sure. Otherwise, it would be too difficult to debug.

Sounds like a consequence of how the code swaps gear (only swapping two pieces at a time). A similar issue is the code getting stuck in a critical damage build when a weapon skill build is better only because it coincidentally started in a critical damage set.

I'll try to test it more next time i ran into it. but this is a copy paste from a convo where i noticed it:

---
Well now im pretty sure did find a pareto suboptimal suggestion

first paremeters was PDT-50 and MDT-50 =

Current best set: PDT:-50, MDT:-50
==============================================================
Best "Time to WS" "Melee TP set" set
==============================================================
main Chango R15
sub Utu Grip
ranged Empty
ammo Coiste Bodhar R30
head Hjarrandi Helm
neck Vim Torque +1 R15
ear1 Brutal Earring
ear2 Boii Earring +1
body Boii Lorica +3
hands Sakpata's Gauntlets R25
ring1 Moonlight Ring A
ring2 Moonlight Ring B
back Cichol's Mantle DEX DA
waist Ioskeha Belt +1
legs Boii Cuisses +3
feet Pummeler's Calligae +3

Avg WS Time = 5.268 s
Avg TP per round = 619.9 TP


I then loosened the defense parameters to PDT-48 and-MDT-50 and got this
Current best set: PDT:-50, MDT:-50
==============================================================
Best "Time to WS" "Melee TP set" set
==============================================================
main Chango R15
sub Utu Grip
ranged Empty
ammo Coiste Bodhar R30
head Hjarrandi Helm
neck Vim Torque +1 R15
ear1 Brutal Earring
ear2 Boii Earring +1
body Boii Lorica +3
hands Sakpata's Gauntlets R25
ring1 Chirich Ring +1 A
ring2 Chirich Ring +1 B
back Cichol's Mantle DEX DA
waist Ioskeha Belt +1
legs Tatenashi Haidate +1 R15
feet Boii Calligae +3

Avg WS Time = 5.518 s
Avg TP per round = 591.8 TP


notice 2 things
A: it still keeps a -50 PDT
B: The WS timer is worsened by almost 5%

---

Please note that not gear selection was changed. just the DT paramets and then re-click to find optimal gear combo

I'll get better info next time I run into it