For The Shinies! A Guide For Thief

Well I pretty much summed that up in that post.

It has WSD+5%, but lacks DEX. Sari has 22DEX, Shijo 15DEX and 5% crit damage.

22DEX is actually slightly stronger for Rudra than 5%WSD
Shijo only has 15dex, which would be slightly weaker than 5%WSD, but 5% crit would even this out on SA/TA.
50DEX on Twashtar obviously beats it by long margin.

I actualy dont. I made my own, but it has no options for anything, its just hardcoded with stats from bis sets etc. and having no descriptions its unusable by anyone but me XD

Bad Simon >_>

Yep, which is why I defined it a slightly inferior option to Sari (leaving out the higher acc, which is less interesting to discuss), with the big big option that you can obtain a Sandung whenever you want, whereas with Perf Sari... good luck with that.

Btw Ternion Dagger +1 for dps is about the same as Sari. So I really wait for Odyssey upgrade for this dagger.

Ternion+1 looks very promising indeed (no TH+1 though!)

For TH I actually use Gandring offhand and TH+2 DM herc :P

I've been wondering about dual weild in tp sets recently, I've been playing with pretty much what everyone else has been here AF+3 body + Reiki, but recently I've been messing around with the Adhemar Body +1, and Windbuffet belt +1. I dont have spreadsheets and things to actually see the numbers on which is better, but judging by the gain/loss it seems like a decent alternative.

My main question is is it worth giving up being dual capped and going one over for the crit? I know the difference in tp gain is pretty minimal, just curious why I've never seen this option brought up.

7% TA and 4 store TP is imo better than 6%TA and 2%QA. Especially that THF has big damage buffs to TA hits.
Then 36 or 37% DW will have the same TP return for many dagger combinations, so you lose nothing. Af+3 body has also higher meva/eva/vit/agi/mdb/hp. Attack advantage on Adhemar probably events out crit damage on Af+3, unless superbuffed, than crit damage is much better.

I think you guys are significantly overestimating how much stronger twashter/centovente is over twashter/taming sari. I've played around with both combinations and centovente really doesn't impress me. If you're geared hard enough twashter/taming's unstacked eviscerations are a force to be reckoned with. I've said it before and I'll say it again. By default my twashter/taming's unstacked eviscerations are stronger than my rudra's. It takes a LOT of buffs to make rudra's surpass evisceration, and when it does the difference is negligable. I've collected some numbers to emphasize what I'm talking about.

The following data was collected in Moh Gates on Apex Raptors. I only summoned a few whm trusts for cure support, and none of them use dia. Raptors do not use any defense or attack altering abilities, so they make a great target for generic data gathering. Weaponskills were fired off as soon as they were available, so everything in this parse is as close to 1000 TP as it can possibly be.


Twashter/Taming Sari Evisceration and Rudra's Storms at 1K TP


Twashter/Centovente Rudra's Storm at 1K TP



The takeaway is that centovente's rudra's are strongest, but the difference between centovente offhand's rudra's and sari offhand's evisceration is surprisingly small. Centovente raises my rudra's damage by roughly 25%, but with sari offhand my evisceration is already roughly 15-20% stronger than my base rudra's.

Now, I know that buffs do play a big part in the game and I'm aware this is a lone scenario. I'm not saying these numbers are absolute, just throwing them out there to support a point. Also, there's one other thing that people don't ever seem to take into consideration when thinking about centovente offhand, and that's that weaponskills have a 2 second animation delay just like activating job abilities do. When I'm in a full buff scenario with maximum haste buffs and full delay reduction the difference in time it takes me to go from 1000 to 2000 TP is literally a second or two. The amount of time you're able to stay engaged to a target is a bigger limiting factor. Getting a Twashter/Taming build to 2000 TP is hardly any more trivial than getting a twashter/cento build to 1000 TP, and with triple attack overflow and the way rudra's scaling drops off beytond 2K the line blurs. A Twashter/cento build will NOT get off twice as many weaponskills as a twashter/taming build, and the weaponskills will NOT be twice as strong.

but i thought aeneas was the best dagger. what would its best offhand be? and i wish i knew how to pm the person that built the community thief guide to update for malignance gear, etc.

None is realy forgetting anything. My numbers include 2 swc ja delay after ws.

If you are using evisceration, then I dont see the point to use twashtar main. Tauret/twashtar is way better for that.

Oh and if your pdif is so low as you show on this example, then Evisceration naturally will be stronger, because crit +1 pdif makes BIG difference then, but then Tauret is even better because its huge boost to crit rate melee to 1000 tp will rival am3 from Twashtar.

The reason why you get such result is because your pdif is incredibly low in the test. It's a scenario that already favors evis.

Aeneas is not the best dagger, but if you wanna use Aeneas the best OH is Twashtar.
In this scenario you would be spamming Rudra's Storm.

Tbf, while inferior to other options, it's not really that bad.
It's a good option, just not the best anymore.

My post was in response to the comments on the recent few pages that seem to imply Twashter/Taming is a bad combination for some reason. Things like this


While I agree that subbing /black mage and casting sleepga on the mobs is a waste of potential, I disagree that the difference between Twashter/Taming and Twashter/Centovente is as significant as some are trying to make it sound. Centovente's max damage still has an edge over sari's max damage, but how big is that difference really? When you're getting a full stack of buffs and mobs are getting all their debuffs the numbers skew to huge proportions, at which point the offhand dagger choice actually feels like it has a minimal impact. Take my recent dynamis jeuno run for example. I used Twashter/taming sari the whole time on this run



With a small bit of percentile rounding to aggregate the numbers, I was able to put up 12 million damage over the course of the run, 9 million of which was rudra's damage, and 3 million of which was white damage. I got off 210 weaponskills over the course of the run, and we do a scoreboard report at the end of the run which reports total number of weaponskills for everyone in the group (//scoreboard report wsavg gives the total number of weaponskills per person). Everyone in our group has a pretty consistent weaponskill frequency for their job from run to run. I've tried out Twashter/Centovente too, and my total damage output and number of weaponskills is extremely consistent regardless of which combo I use. And that bears repeating. Even when I go Twashter/centovente and try to weaponskill at 1000, my total number of weaponskills per run is extremely similar to when I try to weaponskill closer to 2000 with Twashter/Taming sari

Here's the thing. The amount of damage centovente can put up over taming sari offhand is kneecaped by the fact that as soon as you throw out a weaponskill the mob dies. Period. Trash mobs just die to weaponskills when your get strong buffs, and I build TP so fast that my twashter/taming sari runs have almost the same weaponskill frequency as my twashter/cento despite the extra wait for TP with the sari. You wouldn't think that should be the case but it really is. The downtime of switching between mobs, running our of mobs entirely and having to wait for the next pull, and running from point A to point B has a far bigger impact than my offhand dagger selection. My damage numbers in any given dynamis run are always consistent regardless of which offhand I use, so much so that the number of times Su5 mobs one shot me has a bigger effect on performance than which offhand I select.

I've parsed Twashter/Taming and I've parsed Twashter/centovente in all the dynamis zones and honestly I can't tell the difference between the two combinations because individual variances from run to run have a bigger impact than my dagger selection. So is centovente a stronger dagger? Probably yes. On paper it's absolutely stronger, but that's against mobs with infinite HP weaponskilling ASAP. In a real world scenario with actual in game mechanics that line blurs so heavily that I can't see the difference. Is that difference nearly as pronounced as some are trying to make it seem? I really don't think it is.

If you have lags in dynamis and cant fire off WS asap above 1000 tp its not a scenario with actual game mechanic. If you uave same amount of WSs with /cento trying to ws at 1000+ and /sari trying to WS at 2000 than you

a) you have acc issues with cento and make less tp
b) you fail to ws soon enough because of lags

In normal no lag or small lag scenario you can easily WS just above 1000 and Cento is way better than Sari then. Its not even close, its like 25% higher DPS.

So is the argument that twash/cento or tauret/cento wins?

It's neither. Your specific usage matters more than what goes where.

If you need to ask, it doesn't matter which you go with, you're under performing either way.

This is what you're looking for, a blanket answer with no context for a blanket question with no context. It's on page 188.

Every page of every guide is "I don't know how to play, tell me how to mimic you"

My argument is that spreadsheets and calculations paint a different picture from what actually happens during play. Essentially this.


I base my assessments off in game experiences and data I collect and analyze after my runs. Twashter/centovente is powerful, and I don't doubt that it pulls ahead of twashter/taming, but it's not a 25% dps difference like Simon says it is. The conditions that he bases his numbers on don't exist in the game. There is never a scenario where you can stay engaged to a mob 100% of the time, nor do mobs have unlimited HP. In practice you spend more time in dynamis trying to engage mobs (often to have someone else kill it before you get more than a swing or two off) than you do actually fighting the mobs. Most NMs are zerged pretty hard, and nothing lives for long. When your support gives you a full stack of buffs and you start incorporating actual gameplay stuff into the equation there's a stark difference between the theoretical spreadsheet world and real gameplay.

I do get off more weaponskills when I go Twashter/Cento than I do with Twashter/Taming. But it's not a crazy increase, and because TP flows so fast at capped delay reduction the actual difference in average power level is actually comparable. The difference between twashter/taming versus twashter/cento is like background noise in a sea of instruments. Sure it affects your overall numbers, but it's not as significant as some people think it is.

What do you expect?

You used a scenario that specifically doesn't favor rudra(low pdif) and engage speed matters more than anything(dyna D), of course that's the result you got.....

Spreadsheet has always been ws avg/avg dps calculator, that's it. nobody ever claim it can also calculate your dps when engage speed matters or mobs die in 1 shot. Thats what sims do, not spreadsheet. The "conclusion" you made here isnt anything new.

If you want to calculate dps with mob HP as a factor, use a different tool.

And how does that change anything for our case? Things like that has immpact on builds based on buffs. Like for example it would have impact on MNK veret build because you would waste a lot of Impetus time between engaging.

Difference between Twashtar/cento vs Twashtar/sari is not impacted by this in the slightest. Actually when I think about it, /cento would have one advantage. Its easier to take advantage of SA and TA timers with /cento, because you are not forced to wait to 2000 tp.

The difference between cento and sari in your example is smaller than it should be, because you have acc or lag issues. Number of WS with Cento if you really try to WS asap vs waiting to 2000tp with Sari should be drastic. We are talking about twice longer time spend on TPing to WS.

You know that's complete bull.

At capped delay reduction with 55% or more triple attack rate and samurai roll it takes a couple seconds to build 1000+ TP, and overshooting the anchor tiers by 300-500 tp isn't uncommon. The difference between 1000 and 2000 tp is a second or two worth of swings. You make it sound like sari has a difficult time building tp. Meanwhile when the mob's dead the mobs dead. DPS stops until your find another one. The supply of trash mobs dwindles very quickly after a pull, and moving between them is the limiting factor.

Let me put it this way. I've been keeping track of my dynamis runs now for over half a year. I've done runs with Twashter/cento and runs with Twashter/taming. Neither of them result in a big difference between one setup versus the other by the end of the run. And no, there isn't an accuracy problem. I keep tabs on my accuracy too, and it's consistently around the 95% mark by the run's end. If your hypothesis that offhanding cento should raise my dps by 25% over offhanding sari held true then my numbers in that dynamis run would have in theory been closer to the 14.5 million mark rather than the 12 million mark had I switched my offhand from sari to cento for the run. But in practice that just doesn't happen. The big picture numbers I see are consistent from one run to another, and individual run variances play a bigger part in the final results. You always throw out numbers and math based off your calculations and spreadsheets, but I never see you back any of it up with actual data collected in game.

I base my observations off of what I see from kparser. Actual empyrical evidence collected from real play time. And what I see is that sari offhand and cento offhand are very comparable in actual gameplay environments. Yes I do still think there's a difference. But you're blowing things way out of proportion, and I think personal bias plays a part here. I know how fond you are of your cento, and that's fine. It's a great toy. But don't let that disillusion you. Go out and collect actual data. A LOT of data gathered over months and months. You'll start to see how significant run variances are based off of factors like pulls, deaths, and group composition. And you'll also start to see how difficult it is to distingush the numbers between one setup and the other.

Whats a complete bull? I dont get it. Going from tp return after ws to 2000 instead of 1000+ is twice as long or more.

By virtue of JA delay even existing, you are incorrect.

So we can only discuss if I spend several months on gathering data in dynamis runs on THF? I guess we wont discuss then, because I for sure have better thinga to do lol

Have you ever thought that maybe you are not playing in optimal way instead of math being wrong ;)


This made me laugh :D
Its almost like Cento would be hard to make lol

You're overestimating how important that extra second or two is in the big picture, and because of stuff like triple attack and its accompanying tp overflow that 1000 TP with cento is really more like 1300-1500 most of the time. You know better than that. Even if you try dilligently to never weaponskill over 1500 with cento it'll still happen once in a while because triple attack happens a lot. The difference between going from 1300-1500 with cento versus an extra round of swings with sari to reach 2000 is next to meaningless in the long run, and it's in no way shape or form "twice as long". It's an extra round of swings yes, but you know just as well as I do how trivial it is to get a round of swings in with daggers at capped delay reduction. Also this.



Job abilities and weaponskill animation both play a factor in things. It's not nearly as cut and dry as you make it sound. Actual game environments are a lot messier than spreadsheet worlds. There's a lot going on in tandem in the actual game that isn't represented in those sheets, and when I look at actual parse data I can see the impact of those interactions.

I said twice longer time spend on tping not double the amount of WSs. The difderence in amount of WS will obviously be smaller, but it should easily be 50% more with Cento if played optimaly and really ws asap vs WSing at 2000 with Sari

We need a "same ***, every job, every time" Thread to be it's own thread.

Just call it "no u" and every post is "you're wrong" "no u"

It accomplishes the same goal.

nou