Some views on DPS balance that everyone cares about

[Myon88]

Hmm, all this time I've assumed that the workings behind RDPS calculations were common knowledge. Not the actual math mind you, just the principle behind it. "Damage gained from buffs is credited to the buffer" is simple enough, right? Now that makes me wonder about all the other times people bring it up, especially in the other threads about the ranged. Did the people arguing about statements like "ranged should do the same rdps as the melees" actually know what they were agreeing or disagreeing with?

And yes, the shift to looking at rdps as a metric has some interesting ramifications:

• It means that 100% selfish jobs like mch/sam/blm perform the same regardless of what kind of party they're in. Conversely, the support oriented jobs are dependent on their team performing well in order to feed rdps back to their own parse. I'm not so petty as to suggest that this is a privilege that selfish jobs should be grateful for, but I won't deny it is kinda nice.

• It also means that the true ceiling of a particular job may only be visible in a party that caters perfectly to that job's buffs. Take ninja for example, a sam/smn/mch + ninja comp where everyone has strong burst every minute would likely produce a more impressive ninja rdps number than a mnk/blm/brd + ninja one.

• The other interesting thing is it also means that jobs have the capability to hurt the rdps numbers of their teammates. Here's an example. You're a black mage with xenoglossy ready and trick attack is active. If you xeno instead of fire4 to make sure it lands in trick attack, your ninja will be credited more rdps from this trick attack window, but you yourself may lose dps later if you need to move but don't have that proc. If you played more conservatively and saved that xeno for later, you've safeguarded your own parse but your ninja loses out because you did not pad their trick attack as hard as you could.
  
  Of course, ideally you would fulfill both conditions by planning ahead better/being a better black mage. But this is a curious inversion of the adps/personal dps status quo, where the responsibilities were reversed. If a buff was active the onus was on you to burst as hard as you could under it, because if you didn't, the only one to lose out was you yourself.

[NaiXizi]

It's possible and has been a thought some have had with how the current rDPS system on fflogs is causing a widespread undervaluing of synergy comps. I had a long discussion about it with Zeppe (of many world firsts fame, big fan) the other day because he was asking for my insight into what I believed was the best comp for world/blind prog and week1 clears.

The problem with fflogs is ensuring the model is correct enough to give as fair a representation of things as possible. Something I've worked on with more pressing sources of attribution (notably DoTs as mentioned, DH and all sorts), but it's still one of those things where we have to ask ourselves "is rDPS being calculated as sensibly as it could be?". I don't fully know, something still feels off to me, it's gotten a lot better for sure, but it's not quite there which can lead to incorrect conclusions of matters when we're talking about differences of a few hundred potentially reshuffling the whole list (because honestly, things are much closer than they have ever been for most jobs).

I've initiated another project that should help address being able to solve that query and more.

Yes, there is a lot of doubt about the calculation method of rdps. Underestimating the gain of a critical hit buff will cause this data to be biased. I have an idea. The current rdps is to average the critical hit buff to the players, so it is better to add a range value. For example, all the Critical Hit Buffs are given to the initiator of the Critical Hit Buff as a piece of data, and all players who absorbed the Critical Hit Buff are given to them as a piece of data. Taking a maximum value and a minimum value in this way can intuitively reflect the balance of the job?

[Vendalwind]

Myon is on about exactly what i was thinking. I get the simplicity of measuring buff contributions as they currently are, but if selfish dps dont have incentive to perform within those buff windows then maximum party synergy isnt incentivized, and people with that knowledge will play even more selfishly. Also it doesnt reflect accurately then the skill of players that know how to utilize buff windows optimally.

That is a very unfortunate side effect in my opinion.

How weird would it be to, for the purpose of parsing single players only, award the full 100% of whats gained and given on both ends? Itd have its own host of problems, but as an additional metric it would be interesting. And problematic eh? But man im just not sure what the right way too look at it would be? Even a 50 50 split would have problems because selfish dps would then be gaining from party comp significantly without any actual effort.
Is there a right answer?

Maybe just 5% should be given to selfish generally just so its a noticable and measureable performance increase even if minorly.

[Lastelli]

Myon is on about exactly what i was thinking. I get the simplicity of measuring buff contributions as they currently are, but if selfish dps dont have incentive to perform within those buff windows then maximum party synergy isnt incentivized, and people with that knowledge will play even more selfishly. Also it doesnt reflect accurately then the skill of players that know how to utilize buff windows optimally.

That's a non-issue. Let's consider the 3 selfish dps:
1) BLM: in general, BLM doesn't give a damn about party buffs because its damage is very stable and its resources are mainly used for movement purposes;
2) SAM: it has a 60 seconds cycle (tsubame gaeshi + higanbana), so in order to maximize its own performance it also needs to perform well in trick windows;
3) MCH: wildfire every 120 seconds + 1 hypercharge during every trick. The only possible issue here is queen summoning, but the general consensus here is to use it together with reassemble every 60 seconds, so it naturally lines up with trick windows.

So there's really no need to incentivize selfish jobs to perform well during trick attack windows. They're already doing the best the can for the NIN by being selfish.

[Karshan]

That's a non-issue. Let's consider the 3 selfish dps:
1) BLM: in general, BLM doesn't give a damn about party buffs because its damage is very stable and its resources are mainly used for movement purposes;
2) SAM: it has a 60 seconds cycle (tsubame gaeshi + higanbana), so in order to maximize its own performance it also needs to perform well in trick windows;
3) MCH: wildfire every 120 seconds + 1 hypercharge during every trick. The only possible issue here is queen summoning, but the general consensus here is to use it together with reassemble every 60 seconds, so it naturally lines up with trick windows.

So there's really no need to incentivize selfish jobs to perform well during trick attack windows. They're already doing the best the can for the NIN by being selfish.

Indeed. Moreover, Even as a selfish job I'd tend to target tricks to help with global group DPS because that's what matters in the end.

[Myon88]

You should, yes, as from the standpoint of 'killing the boss as fast as possible', bursting during raid buffs is the correct thing to do. Personally I think ADPS was better at 'teaching' people to play like that, as it was in your own interest to look out for burst windows in that system (of course, it had its own flaws too). Buff alignment wasn't just a suggestion, it was mandatory if you didn't want to be punished.

We are fortunate that SE designed a lot of jobs to have looping rotations, so most of the time you will align naturally as long as you're not making mistakes. There are edge cases though, like the xenoglossy example I mentioned above.

Summoner is another example as it now has the option to execute its rotation in 55s or 60s cycles. If you opt for the former, you may gain more bahamut/phoenix casts which is a clear gain for the summoner, but the tradeoff is that you will drift further away from 1 minute burst windows the longer you do it, hurting the rdps of your party members. There may be cases where this is the right thing to do for everyone too, i.e an extra bahamut is more boss damage than less bahamuts but better alignment, in which case the 'optimal' course of action hurts the rdps of your teammates a little bit.

And probably the biggest difference is poor play on the part of the selfish jobs is not punished as hard as it used to be. Now I say selfish jobs but this goes for everyone in general, 'support' jobs contribute to each others' rdps as well. Say a samurai or machinist drifts their 1 minute cooldowns out of sync with trick attack, who gets punished for that? If this was adps/personal dps, they would get a lower score, but under rdps it might not actually change at all (they're throwing out the same total potency, just shifted around), and the ninja in their party is the one that gets punished.

You could frankly write a whole essay on the pros and cons of rdps/adps metrics. Neither of them I would say is strictly superior to the other, but it's important to know what the failings of each are so you have that context when looking at the stats.

[Nemekh]

Play correctly and to the party's strengths. I will never teach otherwise.

Rankings for the team-based metric that is rDPS has sadly warped how some players approach the game. I never want to see players advocate poor practices just because it might look better on their rDPS.

[Vendalwind]

That's a non-issue. Let's consider the 3 selfish dps:
1) BLM: in general, BLM doesn't give a damn about party buffs because its damage is very stable and its resources are mainly used for movement purposes;
2) SAM: it has a 60 seconds cycle (tsubame gaeshi + higanbana), so in order to maximize its own performance it also needs to perform well in trick windows;
3) MCH: wildfire every 120 seconds + 1 hypercharge during every trick. The only possible issue here is queen summoning, but the general consensus here is to use it together with reassemble every 60 seconds, so it naturally lines up with trick windows.

So there's really no need to incentivize selfish jobs to perform well during trick attack windows. They're already doing the best the can for the NIN by being selfish.

You do also have to keep in mind that currently trick is forced to drift and average of 1 second per minute. So if other jobs are aligning perfectly (which many do and can avoid drift entirely because of their weave windows and non screwy gcd timing), No it will not work. By the second half of the fight their bursts will be missing half the trick. So your above statements are actually incorrect It does take coordination to continually utilize your entire burst in trick windows. And thats not even considering stacked buff moments.

[NaiXizi]

Play correctly and to the party's strengths. I will never teach otherwise.

Rankings for the team-based metric that is rDPS has sadly warped how some players approach the game. I never want to see players advocate poor practices just because it might look better on their rDPS.

Unfortunately, many people use this ranking as a ranking of jobs strength. When people ask which dps is strong, they will judge based on this ranking. As a result, some jobs have been discriminated against, such as RDM and bard.
Create a new type of data to calculate an expected probability based on its own characteristics. For example, the expected probability of 3000 Critical Hit is 20.8%. I used 100 times firepower in the raid, and I expect that less than 21 will cause a critical hit. Then we make a hypothesis that in the raid there is a period of Critical Hit buff, my skill can only be Critical Hit because of the Critical Hit buff, without the Critical Hit buff can not be the Critical Hit. Then based on the overall calculation, if I actually have a critical hit probability lower than the expected probability, then a part of the benefits of enjoying the teammates critical hit buff is my own rdps. However, because different Critical Hit buffs have different probability increases, we should prioritize the Critical Hit buff gains with the lowest probability as our own gains.Such as bard’s Critical Hit buff. Then the benefits of other critical hit buff periods are classified for as critical hit buffs initiate jobs.
If I actually have a critical hit probability higher than the expected probability, then all the gain in the critical hit buff time period will be classified as the job that launched the critical hit buff.
Is this relatively fair? But I think there are still some problems, for example, how will a 100% critical hit jobs be calculated, or how will a high critical hit jobs be calculated? And the fire frequency of some jobs is different from other jobs, for example: some jobs can use 100 times in one minute, but some jobs can only use 50 times in one minute. That's a new problem.

[Lastelli]

You do also have to keep in mind that currently trick is forced to drift and average of 1 second per minute. So if other jobs are aligning perfectly (which many do and can avoid drift entirely because of their weave windows and non screwy gcd timing), No it will not work. By the second half of the fight their bursts will be missing half the trick. So your above statements are actually incorrect It does take coordination to continually utilize your entire burst in trick windows. And thats not even considering stacked buff moments.

That's not the case, because fight mechanics will naturally make other dps also delay their rotations by a couple of seconds, so at the end of the day it makes no difference. The only fight where this doesn't happen is voidwalker, but even ramuh, that's the closest to a full uptime fight for this second savage tier, knockbacks make sure that everything aligns nicely. Not to mention that the rotations of all jobs naturally slightly drift.