Prime WS Information, Testing, Discussion

I think that's the discrepancy- this post says WSC are ~26%
https://www.ffxiah.com/forum/topic/57318/prime-ws-information-testing-discussion/77/#3730491

I'm working on verifying WSC now and will post the data when it's done. Thanks for the extra set of eyes on this.

Argisto provided a link earlier in this thread(https://x.com/za_ibd/status/1809190197768437899) which I believe is what BGwiki is based on(source not listed). If you are interested in confirming it feel free to collect more data.

It might be worth collecting some 2k TP Dagda numbers too just so we don't dwell on the assumption 1k to 3k is linear for it.

I've got a pretty clear picture for sword. Although, I only tested 1k and 3k so the 2k is projected. I also tested only MND so the dex is also projected. Confidence ratio is high because I did hundreds for each condition.

Imperator:
ftp=(6.6, 13.35, 20.1) wsc=MND+27%, DEX+27%

1K no fotia +fotia neck
diff 1.049879108 1.049854407
min 11167 11333
max 11724 11898

3K no fotia +fotia neck fotia +70 mnd
diff 1.04992729 1.049845906 1.049440442
min 37134 37315 39138
max 38988 39175 41073

Unfortunately stats were a little weird between the two runs, I should have planned better but was hundreds of WS in.
Stage 4 Caliburnus ML50 RDM vs Huge Hornets (S. Gustaberg)
1k: STR+272, DEX+410, MND+314, /sam (pdif+0.1), PDL+20% gear, PDL+4% AM1, WSD+3%(gearswap was ignoring disable for ammo slot)
3k: STR+276, DEX+410, MND+310, /drk (pdif+0.3), PDL+20% gear, PDL+10% AM3, WSD+3%(gearswap was ignoring disable for ammo slot)

I think this actually makes it weaker than what's in the python simulator. I don't have access to club but is there anything else worth a thorough examination?

Thank you for your work.
We recently discovered Maru Kala info is most likely wrong, so you could try that if you have access to Varga.

What's a good way to get exactly 1k tp? Do you just mess with stp or wait on sekkanoki?

You often need to make hundreds of WSs if you are unlucky, so waiting for Sekkanoki would be painful. Just get enough TP to get 100, 200 or 250TP per hit ang get 1000, WS and requip weapon to reset to 0. You can use ranged weapons to get tp faster on jobs with low delay weapons like daggers (you can shot without engaging and from distance and get more TP for each mob).

To collect these data I waited on Sekkanoki, I knew it would be painful but I had the time and I got numbers I trusted. It was 2-3 times slower than tact+misers to get the 3k data. I tried to come up with a 1k set but it seemed much harder than 2 handers like GKT. Also setting a fixed tp/swing is extra complicated for drg/dnc/rng due to conserve TP requiring you to swap weapon after each WS to reset tp back to 0.

I even tried making an occult acumen set to get fixed tp but forgot to consider ranged weapons... that's a great idea for 1h testing

Strategy was basically put up tact roll > pull with blind > wait on sekkanoki > WS > repeat. (3k was the same except no sekkanoki)

And just to sanity check, I should do this on the level -1 mobs since the stats are known right?

There's nothing to test on with more HP? I have GEO BRD and COR so I can easily ensure attack is more than capped on anything really.

I have a 100 TP/hit set that I am using and WSing in to start collecting more data

If you can use Ullr, it is pretty easy to make a set that lands you on 250tp per shot, I found it to be pretty quick... Just switched weapons after every WS, so I could start from 0.

I think you needed 97 STP to land on 250tp

This worked for me
Volte Tiara
Anu Torque
Crepuscular Earring
Sherida Earring
----------------
Malignance Tabard
Malignance Gloves
Lehko's Ring
Chirich Ring +1
----------------
10STP Cape
Gerdr Belt +1
Malignance Tights
Malignance Boots

Do we know the values for Lorg Mor 5 AM?

It's not required but it makes the 'fact check' at the end easier, and it can prevent the issue of having buffs fall off (or cor rolls not being consistent)

I know this isn't applicable to most weapons, but any job that can use gleti's is very nice for this as you can build sets that can do anywhere from 1-18 TP/tick. Unfortunately, that's only drg, thf, dnc, blu, bst. When I did my testing for mpu gandring, that is how I did it.

For everything else, like simon said, you try to get a TP set that does either 200 or 250 TP/hit.

I ended up writing a script to do this because it's painfully slow at 100TP/hit. I am only fighting bumblebees outside Windurst so this is extra slow but I will eventually get a good sample size....

Sorry for the delay, I will update whenever I have a sample size large enough for a 5% spread at 1k, 2k and 3k TP

Aspens, I agree on the imperator 27% mods and 6.6 ftp @1k. But I am getting ~19.93 for the 3K values. Take a look at what I am using for the inputs on the 3k no fotia and let me know where we are differing?

base 172
fstr (assume capped) 27
27% DEX contribution 110
27% MND contribution 83
pdl% 1.2+1.1 = 1.3
pdif 3.25+(77/256) = 3.55
wsd% 1.03

So I am doing something like this:
min (172+27+110+83)*(3.55*1.3)*19.93*1.03 = 37,136
max (172+27+110+83)*(3.55*1.3)*19.93*1.03*1.05 = 38,993

statistically your proposed ftp is less than 1% off from my value. I calculated the ftp independently from other stats so I think it's at least 20, can be 20.2 and since I calculated WSC using the ftp I calculated, that error compounds into the stat calculation.

I would wiggle the WSC to match the ftp I got (20.1) so maybe 25%? 26% with the type of truncation this game does we're really in the ball park. It actually has me thinking if any of the non-prime WS have been this thoroughly tested

Do you have the numbers for a stage 5 dokoku? I’m curious if the DMG+skill increase it further. I am assuming I’ll see the 2% pdl increase be 1500-2000dmg.

Disregard the comment I failed to realize the V… I am getting 72389 average with stage 4. About 5k increase not bad.

Do you have a Silver Gun (even better +1)?

Easy to build a precise 1000 TP set with some store tp and a bronze bullet.

Might save you some time. It's how I did my Zesho Meppo testing.

Edit: Lawii had a similar point. Doing this saves a ton of time.

I appreciate all the comments on gear but I'm testing dagda (club) on GEO which has basically nothing lol

Geo can use Silver gun +1 and bronze bullet ;)

Good call on Silver Gun and Bronze Gullet.
HQ needs 50 STP / NQ needs 48 STP to land at 250

I used Soil Gorget instead of Fotia because my alt didn't have any of them.

Here's some 1k data points, I'll start working on 3k. I assume this is close enough to a 5% spread?

Stage 4 Lorg Mor
GEO/WHM
Bumblebee (level -1) in West Sarutabaruta
DMG: 227
STR: 278
MND: 377

Dagda at 1000%:
Count: 176
Min: 6449
Max: 6768
Avg: 6618
Spread: 4.95% (max/min)


Dagda at 1000% (Soil):
Count: 179
Min: 6593
Max: 6919
Avg: 6766
Spread: 4.94% (max/min)


Edit:
Dagda at 3000%:
Count: 194
Min: 22523
Max: 23640
Avg: 23113
Spread: 4.96% (max/min)

Edit2:
Dagda at 3000% (Soil):
Count: 305
Min: 22678
Max: 23808
Avg: 23257
Spread: 4.98% (max/min)

This spread leaves a room open for fTP range from 4.244942197 to 4.481697147

Got it, more data, I'll keep going ;p

Genoxd what subjob are you using and what mobs are you fighting?

Bees in West Sarutabaruta, and GEO/WHM. Will update post

On Stage 5 Opashoro testing with Oshala are we SURE the modifiers are MND/INT?

Running SMN with COR + GEO

Geo frailty
Indi fury

Crooked DRK roll 4
SAM roll 2

Oshala 3k TP:
54164
52656
54557
52385
53923
51849
53234
52389
53736

Then with Entrust Indi INT:
54993
53524
55970
54868
53271
52629
55069
54077
56603

Not really any reeeaaaallly noticable change in damage, I found the damage to be the same when entrusting STR and MND also.

Your post actually confirms it is INT. Assuming you didn't change anything else. The damage range is ~9% (max/min).

The small sample size I have in my post has a range larger than 5% and since the only thing I changed was MND between the two sets. MND must be a mod.


Entrust would only be INT +25. Considering the numbers I'm assuming you were using actual WS gear so +25 INT would be a small increase that would be hard to tell by eyeballing.

Ok boys and girls, math time!

I recently posted on the monk forms about my experiences with varga purnikawa and maru kala, and we came to the agreement that the current stat mods on the wiki are very incorrect. So I've spent many hours gathering data on maru kala to try to come upw ith the correct figures. Testing was done in south gustaburg against tunnel worms. To get even increments of tp at 1000, 2000, and 3000 I made a boost macro (no not that boost macro!) and used my ask sash to regain in even 200 tp increments, clicking off the effect as necessary to not go over the mark. I followed the same principal as I did when I tested for sarv modifiers earlier in this thread. Since monk can equip the mummu set just as ranger my base maru kala set was 5 pieces of mummu and stage 5 varga purnikawa, to ensure that I was capped on fSTR values. I performed a large enough sample size for each tp anchor point to gather data for 1k, 2k, and 3k tp first without fotia gorget, then with fotia gorget to determine the tp scaling. Afterward, I equipped a bunch of accessories that added either strength and no dex, or dex and no strength to gather data for the stat mod testing. At no point did I use any PDL or WSD equipment, so the data should be as clean as possible.


My monk is master level 42, and my Hand 2 Hand skill is 482

So, who wants to take a crack at reverse engineering these to get the mods while I do my nightly limbus runs?

The spread is way too low to calculate exact values. Melee test are way longer than ranged. Ranged has now randomizer, you hit for the same damage all the time if you are ratt capped. Melee has 5% randomizer when you are attack cap, so you need to sometimes do hundreds of WSs to pin point exact values.

What would be the minimum spread for any degree of certainty? The range is generally 4.5% to 5%, and a few of those values are closer to the 5% difference. I can keep working with some of them, but it's unrealistic to hit the 5% on every figure with the way the bell curve works. Is 4.8% reliable? Or are we looking at 4.9% minimum?