Evis crit rate testingFollow

Changed thread name and tracking our total here for easy reference. Will be updating as we go.

As of 6/3/09:

100TP:
Sample: 248/696 crits total (35.6%)
Subtracting 24% base were left with 11.6%

200TP:
????

300TP:
????

11.6% bonus to the 1st hit after subtracting 24% base crit rate for 100TP Evis

*Sample still small so exact % is not conclusive yet. But there definitely is SOME significant crit bonus.*

Past the halfway point. still need ~300 more to get 3% margin of error to be more confident in our results. Need more samples! {Help me out!}
_______________________________________________
Original OP:
_______________________________________________
*This is NOT an Evis VS DE thread!*

Question came up on a Drakesbane thread about Crit rate for all the "Crit varies with TP" WSs.

Has anyone actually gone out and 1shot eviscerated bunnies to determine what if any Crit rate bonus there is at 100TP? (and 200/300 would be nice to know as well, but less important). If any previous tests exists, i would love to read them. If not, i figure its time to do one, but why waste our time if its already been done.

For the love of god, i dont want this to be another "People with Mandau/X's with Evis VS everyone else supporting DE" thread. I am only trying to find out if Crit WS have a crit rate bonus, and if it is the same bonus (or lack of) across all crit WS.

*This is NOT an Evis VS DE thread!*

Edited, Mar 16th 2009 5:24pm by Banalaty

Edited, Mar 20th 2009 2:33pm by Banalaty

Edited, Mar 20th 2009 2:34pm by Banalaty

Edited, Mar 20th 2009 3:35pm by Banalaty

Edited, Mar 20th 2009 3:37pm by Banalaty

Edited, Apr 12th 2009 5:20am by Banalaty

Edited, Apr 12th 2009 5:22am by Banalaty

Edited, May 9th 2009 4:02pm by Banalaty

Edited, May 18th 2009 4:19pm by Banalaty

Edited, May 27th 2009 12:05pm by Banalaty

Edited, May 27th 2009 12:07pm by Banalaty

Edited, May 27th 2009 6:52pm by Banalaty

Edited, May 28th 2009 4:27pm by Banalaty

Edited, May 28th 2009 4:28pm by Banalaty

Edited, Jun 3rd 2009 4:51pm by Banalaty

Well, the test would be simple enough. Go to some bunnies with a 5tp/hit dagger and get as many 100tp Evis as you can to go 1 shot baby mobs. You would have capped natural crit rate, and after enough Evis, we could easily determine the crit rate. Repeat for 200/300 TP. Compare Evis crit rate to Base crit rate (20+4%with merits on lv 1 mobs)

That would tell us if it has a crit bonus at 100, and how it scales at 200/300, but would take quite a bit of time. After the 1/2/300 crit rates are determind, a similar test at 150 TP would tell us if it is anchored or scaling.

The test is simple, just very tedious and to do it with 100% accuracy. If you used a 5tp/hit dagger and TPed on 1 shotting lv 1 mobs, you could get 100 TP every time since triple wouldnt screw it up. /sam reccommended for 60TP bonus to speed things along xD

However, I would be more than happy to just have the 100TP mods for now as that is the most relevant. Think i might try to do a few whenever i have some time and start recording some data.
__________
Edit:
Did some STP calcs and /Sam with Blau, Raja/Brutal (+21 STP) gives blau 5.9TP/hit which puts you at 100.3 TP after 17 hits. I will be using this setup to test.

Edited, Mar 16th 2009 6:46pm by Banalaty

Edit #Dos:
Realized would be different after a 60TP med. After Med, I end up with 101.3 TP. Honestly IF it is scaling TP, im going to ignore the differences between 100.3 and 101.3 TP as it would be so small it wont change the outcome to any significant degree. If its anchored it makes zero difference. All WS will be done with either 100.3 tp or 101.3 TP and used to measure 100TP Crit value.

Edited, Mar 16th 2009 6:55pm by Banalaty

lol wow thats something of a "Duh" moment with Martial knife. I have begun testing 100 TP. I eventually switched to getting a 450 delay self bow to build TP faster by sniping worms and bunnies for more TP instead of engage/smack 5.9tp/disengage repeat.

Just 40 WS right now so no point in posting the results yet, but were on the way to new knowledge!

(Sometimes i think it would be nice for SE to just release comprehensive game mechanics and save the forum people countless hours of testing :S)

Are you not parsing your regular atks also, so you know would know your base crit rate for your sample period?

Because he is performing the test on a level 1 rabbit there is little doubt that his dex is 50+++ over the targets agility, so his crit rate will be capped. That means 20% + 4% meriits == 24%. The variance in crits would be the modifier at whatever tp anchor and scaling rate (if scaling is applicable) to your current tp level.

This is a test I'm interested in pursuing because it addresses something that has been on my mind for some time. There is a logical progression in realative power of most weaponskill types. There are a few exceptions I can think of such as sidewinder that a lower version is superior to the quested or higher level one, but many jobs use their quested weaponskill for everyday use. Asuran Fists is a prime example of this, although as I said there are other exceptions such as rampage etc etc.

Evisceration is capanble of opening and closing a darkness chain, which coupled with shark bite has been viewed for years as their purposes, and dancing edge to be the main damage on lower eva/defense targets. This trend is still true, but what if we've been underestimating evisceration for all this time? The adjustment that allowed it to crit was taken without much testing, so I never heard of what it did, many assumed it had no modifier at the time and it wasn't spoken much of since. Just playing with numbers here

with 136 dex and 46 chr, wsc of dancing edge is 48 and WSC of evisceration is 33. If you add this to blau at capped str you get base damage 91 for dancing edge and 76 for evisceration. When you multiply this by 5 swings each you get

--DE 455 (I'm ignoring fTP for simplicity here since its a small mod on first hit)
--Evisceration 380

The biggest change then is the difference a crit gives in high attack versus low attack scenarios. Conbsider pDIF at 1.2

DE -- 455 x 1.2 == 546 damage

Evis with 1 crit at 1.2 pDIF

( 76 x 4 x 1.2) + (76 x 2.2) == 532 damage

with 2 crits

(76 x 3 x 1.2 + (76 x 2 x 2.2) == 608

Evisceration needs to critical once to be 14 actual dfamage lower, twice it will be 62 actual damage higher


In a high pDIF now I'll give 1.75

DE

--455 x 1.75 == 796

Evisc with 1 crit -- (76 x 4 x 1.75) + (76 x 2.75) == 741
Evisc with 2 crits -- (76 x 3 x 1.75) + ( 76 x 2 x 2.75) = 817 damage

Here 1 crit is 55 damage less than DE and 2 crits is only 21 damage more.

NOTE: The fTP of the first hit of dancing edge means DE DOES get a stronger weighting in the above calculations than I gave results of. The above calcs are meant only to show a trend of how the weaponskills differ at varying pDIF ratios, the actual results are not to be used for specifics.

In a low pDIF scenario with a critical rate greater than 30-35% evisceration can theoretically outperform dancing edge, and when you tack on the earlier discussion of x's knife perhaps the stories of evisceration not sucking are not so unfounded after all. I want to see more testing to see what kind of mods we're looking at in crit increases, it warrants consideration at the least. The change probably wouldn't be much in overall performance, but you never know. And with x's, it would be a bigger change ^^.


Edited, Mar 17th 2009 1:38pm by Melphina

I know, I ignored fTP for simplicty just to show the trend. In actuality in a high pDIF scenario dancing edge's fTP mod holds more weight than low pDIF. The calculation was to show that "theoretically" evisceration and dancing edge will be similar in performance with a crit mod. I edited the above post to emphasize this point. I still think the lead goes to dancing edge sans x's, but the purpose of the discussion at hand was to evaluate evisceration and it's crit mod, so I don't want to go off on a tangent and turn it into something it was not meant to be. Pdif is not usually 1.75 so the number theorycrafting was only meant to exemplify a trend rather than be used as a reference on specifics.

At any rate, I think this warrens more testing. With no madrigal my pDIF at birds with subs is 1.0-1.02 and with madrigal its in the 1.2 ish range, so it's worth noting that with my high dex build evisc may hold a place. I'm curious to continue testing.

Edited, Mar 17th 2009 1:39pm by Melphina

Starting my batch of tests.

Its boooooring...

I figure if we can get data from more than 1 person we can get a better estimate faster.

Measuring your crit rate with a parser would require actually engaging the mobs rather than using your ranged attack, otherwise you'd be making a similar assumption about the crit rates being the same.

If I manage to get in game sometime, then I'll give this a whirl and try to help the cause. I'm shocked that this test hasn't been performed previously, quite honestly. Then again, I always took it for granted that DE > Evisceration because of the old Evisceration.

Since I can't parse (PS2 go!), I'll just record my damage values and post those as results whenever I get around to it. I have 4 crit merits, so it won't introduce any oddities in the data.

Are you doing anything special, Aanalaty, or just using a dagger, getting up to 100 TP and going to town?

Edited, Mar 17th 2009 5:15pm by ChanchanXI

After 150 1 hit eviscerations at 100 - 104 tp, i got 56 crits.

Thats 56/150 = 37% crits

at 104 dex and with 4 crit merits.

This would give evisceration a 10~15% crit bonus.

Seeing Banalaty's test, i can see the idea that evisceration has a bonus to crit rates at eve 100 tp. Cumulative with Banalaty's this gives us 96/250 crits, or 38.4% crits.

Sample size is still too low to determine what the bonus is, but i think we can say that there is a bonus.

350 samples gives you roughly +/- 5.2% margin of error with 95% certainty. So your testing so far shows that you can be very confident that the bonus is between roughly 10% and 20%.

Given that this is a programmed value and that the value is likely to be a 'nice' number, if you can get within a 3% margin of error then it's pretty acceptable to state that as the 'real' value, barring further testing to disprove it.

To get a 3% margin of error with a 95% confidence interval you need about 1050 samples. I'd say 1000 samples is good enough (3.08% margin of error).

Let's make sure I understand evisceration correctly.. Every hit in the ws has a chance to crit, right? At 100% tp it seems to have a +15% crit boost, meaning about a 39% critical rate for each hit. Right?

Meaning a 39% critical hit rate for at least the first hit, assuming you have capped your critical hit rate for a given mob. I would suspect that the crit bonus is the same for all hits throughout the WS, but it's an assumption that would need to be explicitly stated.

An easier way to think about it would just be to add 15% to your base crit rate (without SA/TA's influence) for a given camp and figure that to be your chance to crit per hit on the WS. Even then, this was only preliminary. I think there's still a +/- 3% on that data? Something like that. But SE likes even numbers, right? >.> So it's probably about 15%.

I havn't forgotten about this. Just no time right now to sit down for hours killing more bunnies. But whenever my schedule clears up a bit, ill get back to this to get a more reliable sample size. Id really like to get this nailed down for 100/200/300% and repeat the test for drakesbane. I have been curious about this for ages.

*Assuming* that the crit bonus to the 1st hit is shared with all hits, you can just use your crit rate in your WS gear and add X% from WS. And when testing is done for 200/300% then the same, just a larger % (whatever that may be).

Hey guys, I'm sort of back (hopefully) from a very long forum hiatus. I got X's Knife in march and soon became impressed with evisceration damage. I've been keeping up with the KI forums and Treelo's evisceration threads (this one and this one) really captured my attention. I've been seeing some fairly impressive numbers, for example these unstacked WS (EV1, EV2). Those were highly buffed, but given being unstacked I feel that I've been underestimating Evisceration for a long time.

The quote above, about building TP. I didn't even realize this thread existed and was doing my own testing using the same method. I just wanted to cue you guys in on faster TP building, that might hopefully encourage some to produce (bigger) samples:

Thf/Sam
Coffinmaker/IronBullet
Rajas (no Brutal)
= an roughly even 20 TP/hit.
Go shoot 5 bunnies with this and you've got an even 100tp.


Here's what I've come up with so far:
Sample size - 46 WS's
162, 127, 212, 161, 217, 152, 143, 225, 232, 112, 236, 165, 144, 157, 136, 228, 155, 233, 126, 137, 226, 117, 125, 200, 147, 225, 116, 211, 115, 121, 152, 159, 109, 146, 227, 140, 227, 226, 235, 135, 156, 161, 128, 138, 110, 211
Here's a plot chart of the data shown in pDif
I WS'ed in X's knife and an idle TP set; base dmg of WS was 69
Hope this can be contributed to the total numbers. I'd go do more testing but I've run out of time for now.


My recorded crit rate so far:
34.78%; or +10.78% @100TP, from a sample of 46

Edited, May 9th 2009 7:58am by Shamaya

Update: (Also editing OP)
Did another 50 today bringing me to 250

94/250 from me (37.6%)
56/150 from Wyrmnax(37.3%)
16/46 from Shamaya (34.7%)

166/446 total (37.2%)

13.2% crit bonus so far. 446 sample size.

That is a good point, but I do believe that all the other crit weaponskills are a +% crit. However, now that you mention it, I believe they tested crit via the same method, so they would have reached the same conclusions that we're reaching, and could have been misled if the possibility that you speak of were true, Desoo. Unfortunately I can't think of how to do a test like that. Something would need to be fought for which we did not have capped crit. Something high enough level for us to uncap crit would likely survive more than one WS hit, which could make testing very very difficult. I think it seems that there is at least a decent chance we can hope that it is a straight addition as opposed to a multiplier of your otherwise base crit rate. As physicists/scientists/philosophers say, the simplest answer is more often than not the correct one. Although SE is surely not an elegant universe.

I think the difficult part would be telling when you got crits. Maybe it's easier than I imagine. I just can imagine that anything that is high enough level to where we're not already at capped crit rate with either base stats or -stats (think stumbling sandals) would quite possibly survive even more than two EV hits. It could be quite difficult if not impossible to tell how many hits crit given the net damage. For example, one hit critting at a mid-range crit-cap pDif could do possibly as much damage as 2 hits critting at a lower pDif and could also possibly do as much damage as 0 hits critting at a high pDif. Even with capped pDif, there is still pDif fluctuation. Add more hits, and it may become difficult (if not impossible?) to tell for sure in many situations just how many crits you got. At least I think this is the case; dk if I'm missing something.

For the record, taking into account the data provided I've done several calculations on the subject of evisceration versus dancing edge with x's knife considered and the data is pretty interesting. I've run scenarios with strong and weak gear, at high and low pDif, with solo and stacked weaponskills. This is 8 different scenarios, and the conclusions are pretty clear. When soloing weaponskill with x's knife, evisceration is the way to go. When stacking weaponskills, dancing edge generally wins due to the fact the first hit has the dex/agi mod and that crit makes up for the power lost from the rest of the weaponskill not criting.

This thread isn't intended to show math of DE vs Evis, so I'll make another in the coming week or two with numbers. I wanna do something new and get a blogger account and link to the math, that way my OP isn't 10 miles long. I'll just keep a summary as the op, and for those who wanna see the hard math they can click the link to the blog, which will make everything so much easier. It'll take me a while to post the numbers though (eight scenarios is a lot), but I'll get to it when I can.

Edited, May 26th 2009 12:45pm by Melphina

Took me a while to nail down a second batch of tests... Been keeping track on the last few days...

103 more WSes on bastok bees

25 crits -> this is lower than my base at 24% Dex @ 98 when doing these, so i am pretty sure i am capped. And 4 crit merits.

Last % should be 36%.

Sleepless night, waiting for NM windows to open, i decided to make myself useful and give you a nice number

Dex = 99 (yey for lifebelt with +1 dex augment...)
Crit merits = 4

97 more WSes on bastok bees - to complement my previous 103

41 crits


EDIT: 41 crits, NOT 71. As nice as that would be.


Edited, May 28th 2009 1:30pm by wyrmnax

50 more. 16 crits in total.

And that brings me to freaking 500 WSes on bees. It takes a while.

How many we need again for confirming?

EDIT: Mistake. im still on 400 -_-"


It *really* takes a while.



Edited, Jun 3rd 2009 12:12pm by wyrmnax

Has this testing died?

I am currently finishing skilling up dagger on ninja / warrior so i can get the WS, and i plan to provide info once i have done so, but if the thread is completely dead then there is little that i could do...

Sorry to bump this from 3 pages back, but I think this is an important project that shouldn't be given up on.

pDIF Caps Testing for 1H and 2H Weapons
Here's some recent information on cRatio-capped pDif values (normal min/max; crit min/max), done by Masamune on BG. His thread got severely overlooked. His sampling doesn't culminate into the revised charted pDif formulae we've been seeking, but it's still important. With some confidence, it appears that both the 1hander and 2hander pDif maxes have been raised, indeed with 2hander receiving a greater boost.

For this test, it means that we should keep the normal max and the crit min in mind when parsing data. It would be bad the crit %'s we derived from this thread were partially the result of confused analysis (due to thinking in terms of the old pDif values). Let's finish this testing ^^

Edited, Oct 20th 2009 12:10pm by Shamaya

Bump.

There is now more reason than ever to finish this up. Now that we have a crit damage boost trait, it is more important than ever to get the crit rate on Evis pinned down so we can see just how much the crit damage boost will, by extension, boost Evis. Unfortunately I now work full time and have had to cut back severely on FFXI so there is little or no time to address testing like this anymore >< However I will be happy to maintain this thread and update our results as they come in. Please assist! We need good numbers on this WS to really measure its new power accurately!

We will also need to keep up with the pdif updates Shamaya mentioned a couple posts up. If in doubt, just post the raw WS numbers and we can figure out what to do with it.