Thank you SE

[Dizzy_Derp]

So...Yes this is 70th percentile.
If you go to 100% percentile... thereby adding in the 30% of LEAST SKILLED players... Paladin rises to the top, but only by a thin hair line.
The lower the percentile you go into, the more skilled the player base, the more Paladin falls in the ranking.

Top 5 highest parsing pld dps
1. 3590.1
2. 3580.5
3. 3329.0
4. 3303.0
5. 3292.9
Top 5 wars
1. 3216.3
2. 3215.9
3. 3198.8
4. 3194.4
5. 3153.6
You are wrong.

[Tegernako]

Top 5 highest parsing pld dps

1. 3590.1

Top 5 wars

1. 3216.3

You are wrong.

How the heck is a slight increase in casting time going to cost them 300-500 in dps to make them so much worse?

Do you even read lol?

Not the first time they have ignored what was actually being said.


http://forum.square-enix.com/ffxiv/t...=1#post4287561

[Fyce]

Top 5 highest parsing pld dps
1. 3590.1
2. 3580.5
3. 3329.0
4. 3303.0
5. 3292.9
Top 5 wars
1. 3216.3
2. 3215.9
3. 3198.8
4. 3194.4
5. 3153.6
You are wrong.

Even if Makeda got it wrong regarding how percentiles work, the point still stands: PLD is only sitting at the 1st place at the 90th and above. Any other percentile will see PLD being 2nd or 3rd place, with WAR being 1st.

But if you only want to only look at the top 5, which represents something like 0.02% of the current total parses, then sure. However you should understand that some people actually want relevant data, and not cherry picked parses.

[FallenWings]

Imho opinion. The only relevant data for balance should be the ones taken from the top percentiles because that is where things are played properly and towards the peak. It's been seen time and time again that the general playerbase is incapable of bringing out any class' true potential when they, to this day and probrably forever more, shove PLDs in the MT slot and have an unnrcessarily high ShO uptime as well as do the majority of the tanking. Thus the data is unreliable because the majority of PLDs are not given equal enviroments to bring out their potential and/or are bad.

Excuse the grandma, im on phone and cbs fixing it atm

[Dizzy_Derp]

if you only want to only look at the top 5, which represents something like 0.02% of the current total parses, then sure. However you should understand that some people actually want relevant data, and not cherry picked parses.

While I have no actual data to back this up I do have a pretty good guess why this is, there was a HUGE skill gap in 3.xx when it came to warrior, with most not being able to come even close to the top players. Now most of those less skilled players hopped to pld and drk due to the bad spot war has been in and how difficult and punishing the rotation is, i believe the warriors remaining have a much higher skill lvl across the board than the pld and drk base especially when it comes to the percenrile range you are pointing out. Also there are not nearly as many warrior parses as the other two so the data when looking at percentiles is skewed in the direction of war. The best metric is to look at what a job is capable of. Because pld is capable of higher dps that cannot be debated it is fact.

[Fyce]

While I have. O actual data to ba k this up I do have a pretty good guess why this is, there was a HUGE skill gap in 3.xx when it came to warrior, with most not being able to come even close to the top players. Now most of those players hopped to pld and drk, the warriors remaining have a much higher skill lol across the board than the pld and drk base. Also there are not nearly as many warrior parses as the other two so the data when looking at percentiles is skewed in the direction of war. The best metric is to look at what a job is capable of.

Susano and Lakshmi are encounters that people can very easily speedrun and pull padded parses out of. The top 5 parses are not "what these jobs are capable of", it's more about the group as a whole being exceptionally good.
The question is: can you reproduce that in savage? And in super savage? As well as: how much people can reproduce that?

If things aren't balanced for a very small subset of players/groups, then it simply becomes an exception. Perfect balance is impossible to achieve. So trying to reach it while screwing everything for the vast majority of players is probably not the best thing to do in a MMORPG.
Balancing things around the very top can be a good thing in a competitive game, but not really here. There's nothing to gain from that. Especially if you base your balance on a very small and innacurate data.

[BluexBird]

Excuse the grandma, im on phone and cbs fixing it atm

LMAO this just made my day! xD

And Dizzy_Derp has the right of it. When we are looking for the job that has the highest damage potential, the accurate place to look at is what the top players are capable of doing. He even did a great job explaining why the percentiles are skewed toward Warriors.

[Dizzy_Derp]

Susano and Lakshmi are encounters that people can very easily speedrun and pull padded parses out of. The top 5 parses are not "what these jobs are capable of", it's more about the group as a whole being exceptionally good.
The question is: can you reproduce that in savage? And in super savage? As well as: how much people can reproduce that?

The group does not have to be good to get good damage, that's a misconception, I was still able to get a 99th percentile parse in Susano in a full pug with no cheese at a long af time. People wanna act like it has to be set up but it doesn't. You just gotta be good. And yes reproducing that at savage will be possible, it will actually be easier because more tanks will play conservatively due to the higher skill required so getting that high will be a cake walk.

[aleph_null]

Even if Makeda got it wrong regarding how percentiles work, the point still stands: PLD is only sitting at the 1st place at the 90th and above. Any other percentile will see PLD being 2nd or 3rd place, with WAR being 1st.

But if you only want to only look at the top 5, which represents something like 0.02% of the current total parses, then sure. However you should understand that some people actually want relevant data, and not cherry picked parses.

At lower percentiles you still see a lot of groups pushing pld into MT role and war into OT role, that's why for anything below optimized runs the edge goes to war, not because of the job inherently doing more dps, but because they're pushed into the lazy role. If you put two equally skilled players into a group as a pld and a war, and tell the pld to pull the boss without shadewalker, take all cleaves/busters without doing tank swaps unless necessary, obviously the pld would do less dps since they can't stay in sword oath all the time. For a fair comparison you need both tanks in dps stance nearly all the time, which is only possible in an optimized run.

Btw I also disagree with picking top parses from the dps leaderboard like dizzy posted above, since some of those parses are obviously padded (some had really slow kill time at 8 min+, some even had 80%+ balance uptime only on the tank). I'd rather look at static speed ranking and choose the pld/war/drk in the fastest groups for comparison.

[Fyce]

The group does not have to be good to get good damage, that's a misconception, I was still able to get a 99th percentile parse in Susano in a full pug with no cheese at a long af time. People wanna act like it has to be set up but it doesn't. You just gotta be good. And yes reproducing that at savage will be possible, it will actually be easier because more tanks will play conservatively due to the higher skill required so getting that high will be a cake walk.

Okay, then, let's look at the 99th percentile if that's what you want. Average DPS are: PLD-2884, WAR-2808, DRK-2783 (at the time of writting. The numbers already changed a little bit). As you can see, there's a HUGE difference between the 99th percentile and the 100th. This clearly shows how bad it is to assume that the top 5 is representative of anything.
In short, the difference between tanks is no way near what some people (you included) say it is, simply because using top parses only is meaningless for 99.98% of players.