Dragoons: A Rotation Reborn

[nanaoei]

Thank you so much!
I love having these on a notepad for reference and i really appreciate it.

haha, fortunately i do main a str melee so i have the fortune of having the crafted accessories... and the pentamelded head for shits and giggles.
The gains are absolutely terrible. I do have to admit that it feels good to look [back] at how hard you worked for the pieces as they serve as a small reminder.
Or it could be complete nightmare where a much-too-high number sticks out like a sore thumb in your head.


Just wanted to add a comment although it already occurs to most of you guys:
The raids have different accuracy requirements, so feel free to downsize on your accuracy gearsets (or make more) for the different turns/primals!
I only say this because most people stick to their t9/t13 sets and don't really look back anymore, especially when things are being farmed now,
but I don't believe it's too much extra work in order to get a slight extra out of the gear sitting in inventory. (( it's like neglected old barbies or transformers toys man.

Thanks again.

[SunnyHirose]

I've checked EMX's spreadsheet, and I think I figured out a reason the value it gives for skill speed is significantly higher than I expected: it models DoT clipping as "you clipped it? NO LAST TICK FOR YOU".

Because it assumes the last tick of Chaos Thrust is already lost as soon as you hit 342, it doesn't model the increasing--albeit minuscule--likelihood of losing a tick with more skill speed.

(Yes, messing with this does mean different BiS... unless you were going for the Black Truffle, in which case nothing changed.)

[Dervy]

I guess you pointed your fingers on what I was trying to figure out? :P

[SunnyHirose]

I guess you pointed your fingers on what I was trying to figure out? :P

medic_ihavenoidea.gif

My personal spreadsheet models (which I'm about ready to scrap due to their relative inflexibility...) don't model DoTs that way, and I was just curious what he did.

[Dervy]

I'm not going for any of the sets I posted. I will not be BiS optimised, unless I blow another 10m on earrings.

[Dervy]

Aah fair enough then. Yeh, the model is constructed in a way that assumes DoTs are on their own internal timer and the final DoT tick happens as the debuff falls off. So applying Chaos Thrust at 29.99s for example, would clip the DoT by 0.01s (which is dumb). I think I can fix this issue pretty easily to be honest. DoT's have an internal global timer of 3s. Therefore as an average, a DoT tick happened +-1.5s of the Chaos Thrust reapplication. So, couldn't we just make the assumption that if we reapply the DoT outside of that +-1.5 bracket, as an average, the DoT will clip?

It's really tricky to accurately model anything in this game because of the internal global timer for DoTs. If you start your rotation 2s into the next global DoT tick, that could be the difference between clipping and not clipping DoTs at any given SkillSpeed (theoretically), which really sucks. Unless you or anyone else has a better suggestion?

@Tetsuya, I'd go for http://ffxiv.ariyala.com/PAF8.

[MrYaah]

Aah fair enough then. Yeh, the model is constructed in a way that assumes DoTs are on their own internal timer and the final DoT tick happens as the debuff falls off. So applying Chaos Thrust at 29.99s for example, would clip the DoT by 0.01s (which is dumb). I think I can fix this issue pretty easily to be honest. DoT's have an internal global timer of 3s. Therefore as an average, a DoT tick happened +-1.5s of the Chaos Thrust reapplication. So, couldn't we just make the assumption that if we reapply the DoT outside of that +-1.5 bracket, as an average, the DoT will clip?

It's really tricky to accurately model anything in this game because of the internal global timer for DoTs. If you start your rotation 2s into the next global DoT tick, that could be the difference between clipping and not clipping DoTs at any given SkillSpeed (theoretically), which really sucks. Unless you or anyone else has a better suggestion?

@Tetsuya, I'd go for http://ffxiv.ariyala.com/PAF8.

I don't think the way to model it is to think about dots as being clipped. You have to look at it like a skill with upfront damage and a dot ticking in the background, if you're constantly reapplying you're never missing a tick, but if you reapply very frequently obviously you're not making the spell tick any faster so you're wasting potential dot ticks and the only benefit is that you've extended the duration by newduration-oldduration +the upfront damage of the spel. if you have a rotation that never drops the dot tick you can just assume the number of dot ticks is rotationduration/3, where the remainder gives you the probability of getting the +1 dot tick because you applyed the buff initially right before a dot tick. You can probably normalize out the last dot tick and just assume that if say, you're getting 21.5 dot ticks the value of the last tick is .5*dotpotency.

That being said when you have a rotation where the dot always falls off the math becomes much easier because you know you got an exact number of ticks and you can bundle the tick damage with the upfront damage on an application by application basis.

[MrYaah]

Is the Crafted Set superior compared to the Non-Crafted BiS Set with the 410 Determination?

Yes, any set that includes crafted gear doesn't exclude non crafted gear in its calculations so if the full i130 set was superior you'd end up still end up using those.

[SunnyHirose]

I model DoT effects as if they are continuous rather than coming in discrete ticks:

Modeled DoT potency = [Full DoT potency] × MIN([active length of DoT] ÷ [Full DoT duration], 1).

The core assumption here is that the ticks have an equal chance to be synced to any time (and since they start pulsing outside reasonable player control, this is sane enough, I think). Thus, the portion clipped is exactly how much DoT potency you should expect to lose over an infinite number of runs. So, losing the last tick because you clipped by 0.01s is treated like the freak accident it is.

[MrYaah]

I model DoT effects as if they are continuous rather than coming in discrete ticks:

Modeled DoT potency = [Full DoT potency] × MIN([active length of DoT] ÷ [Full DoT duration], 1).

The core assumption here is that the ticks have an equal chance to be synced to any time (and since they start pulsing outside reasonable player control, this is sane enough, I think). Thus, the portion clipped is exactly how much DoT potency you should expect to lose over an infinite number of runs. So, losing the last tick because you clipped by 0.01s is treated like the freak accident it is.

You can simplify the dot potency equation down a lot to


Modeled Dot potency = [Dot Tick potency] * [active length of DoT] / 3