BLM needs some love and RDM needs a nerf

[Galvuu]

snip

Percentiles are median scores. They use all of the data available.
The only exception to this statistical validity are the max and min cases, where you're only taking into account the top and bottom most score, and the 99% and 1% percentile, which are pretty much all the outliers (even though the scores come from all the datapoints available, these still matter very little).
5%~95% is statistically significant.
(answering you, but it's a general statement about how the percentiles work; the make use of all parses available)

[PrismaticDaybreak]

Percentiles are median scores. They use all of the data available.
The only exception to this statistical validity are the max and min cases, where you're only taking into account the top and bottom most score, and the 99% and 1% percentile, which are pretty much all the outliers (even though the scores come from all the datapoints available, these still matter very little).
5%~95% is statistically significant.
(answering you, but it's a general statement about how the percentiles work; the make use of all parses available)

The differences between amount of parses makes the deviation quite high, I don't believe FFlogs to be currently reliable.

[Llugen]

Yes the F it can be argued with. Vercure and Verraise cost you DPS, Verraise probably could use a slightly steeper cost since Lucid Dreaming and Dualcast are powerful when used in conjuction with it. Embolden is only a maximum of .75% increase in rDPS. If Vercure got a slight potency reduction maybe 50 or so and again an increased MP consumption, but not too much since DC'ing it is already nearly 10% of RDMs MP.

costs you TWO gcds vs having the entire party wipe? See also: broken.

Furthermore even if embolden is apparently a 1% increase, it still pushes SMN/BLM out of favor very disproportionately, as you'll always have at least two physical classes (tanks) and running 4 casters would really be an issue for loot.

[Galvuu]

The differences between amount of parses makes the deviation quite high, I don't believe FFlogs to be currently reliable.

Do you mean the amount of parses each job has? They're all on above 15k except for poor MCH.
Otherwise, the "deviation" argument only works for the 99% and 1% percentiles because they cluster the statistical outliers of the data (you can argue, with merit, that this is true for anything above 95% and below 5%).
The rest are median scores, so they're very resistant to deviation by design- and appear from the data contained across all parses available.
(As in, the percentiles are inherent normalised in respect to all available data)

costs you TWO gcds vs having the entire party wipe? See also: broken.

Furthermore even if embolden is apparently a 1% increase, it still pushes SMN/BLM out of favor very disproportionately, as you'll always have at least two physical classes (tanks) and running 4 casters would really be an issue for loot.

I did this twice today xD
It's such a feelsgood moment.

[Llugen]

I personally have cast FIVE verraise in 10 GCDs in v4 and with lucid for the 5th one had MP to continue my rotation mostly unhindered. By Square Enix's own logic (see their argument for SMN with raise in all of 2.x and 3.x), that is insanely broken.

[PrismaticDaybreak]

costs you TWO gcds vs having the entire party wipe? See also: broken.

Furthermore even if embolden is apparently a 1% increase, it still pushes SMN/BLM out of favor very disproportionately, as you'll always have at least two physical classes (tanks) and running 4 casters would really be an issue for loot.

If one person is dead any good RDM is going to let the healer deal with it since its usually automatic for them. 3+ is usually my threshold for raising unless its both tanks or a tank and healer or any combination of tanks and healers being dead. The only time a Verraise is absolutely required is the previous conditions. And one person dying is almost never the the entire party wiping, don't be an alarmist. Embolden isn't even 1% it's a maximum of .75% and that's being generous and counting tank DPS.

[HoLoFoNo]

Here is the ghetto spreadsheet showing the ranking of each of the DPS on Lak and Susano. I've ranked them on each percentile from 1 to 9, 1 being best and 9 being worst. Their average placement is shown and based of that their rank is also shown. Obviously there are a lot of moving parts involved in this including party composition, kill time, deaths and most importantly this is only representative of the classes dps and not the offensive utility they bring. However, we all now what offensive utility these classes bring. This only addresses the performance of these classes in fights thus far.



http://imgur.com/a/pKydy if the picture isn't working

[PrismaticDaybreak]

Here is the ghetto spreadsheet showing the ranking of each of the DPS on Lak and Susano. I've ranked them on each percentile from 1 to 9, 1 being best and 9 being worst. Their average placement is shown and based of that their rank is also shown. Obviously there are a lot of moving parts involved in this including party composition, kill time, deaths and most importantly this is only representative of the classes dps and not the offensive utility they bring. However, we all now what offensive utility these classes bring. This only addresses the performance of these classes in fights thus far.



http://imgur.com/a/pKydy if the picture isn't working

I'm assuming the numbers are extracted from FFlogs? Not being snarky, legitimately asking.

[HoLoFoNo]

I'm assuming the numbers are extracted from FFlogs? Not being snarky, legitimately asking.

Yeah using the statistics page. It pulls in around 15k results from each class (MCH at 7k and RDM/SAM at almost 50k)

[PrismaticDaybreak]

Yeah using the statistics page. It pulls in around 15k results from each class (MCH at 7k and RDM/SAM at almost 50k)

Honestly, the more I think about it RDM probably wouldn't hurt too much from somewhere between 20-40 potency reductions, maximum 50. My issue is that if it gets nerfed player skill can only make up the difference so much based on how the class itself operates, the ceiling to a mild extent is based on RNG and there's not much standing room between it and the floor. Which is why I prefer to push for ceiling/floor increases to under performing classes.