Connecting the DoTs: A Summoner Guide

[Havenchild]

Thinking for myself and waiting on a proper analysis to be done are in no way related. If I want to know what is actually better, the only way to accomplish that is to wait until the analysis is done, or do it myself. Doing some parses and changing some gear is probably the least accurate way to come to an actual understanding of what is going on behind the scenes.

Been watching the thread as always and this is where I had to draw the line. This is not a good way of thinking when aiming for high parses. Do not feel because the formulas dictate a number, that this is how it works in game, especially on a patch by patch basis. You should always value you're parses higher in the end no matter what. Sure, you can have a BIS or coefficients thrown at you all day, but if you can maintain higher numbers using another build, you should exploit that or come to understand that what works in game is far superior to any simulation or math done in a calculator. Going on behind the scenes and what gives the most damage don't always line up.

Now just to figure out how a few numbers of this stat may impact overall damage in theory, is different from how it will impact overall damage in execution. Regardless we should WANT to understand how each stat CAN affect the damage done by SMN / Pet anyway.

[T0rin]

Been watching the thread as always and this is where I had to draw the line. This is not a good way of thinking when aiming for high parses. Do not feel because the formulas dictate a number, that this is how it works in game, especially on a patch by patch basis. You should always value you're parses higher in the end no matter what. Sure, you can have a BIS or coefficients thrown at you all day, but if you can maintain higher numbers using another build, you should exploit that or come to understand that what works in game is far superior to any simulation.

Now just to figure out how a few numbers of this stat, may impact overall damage in theory, is different from how it will impact overall damage in execution. Regardless we should stand to understand how each stat CAN affect the damage done by SMN / Pet anyway.

This discounts the reality that there may be minor differences in execution between parses. I can do 10 different parses with the exact same opener and routine and come up with 10 different results, without changing anything. Unless you execute everything perfect, every single time, no, this is a really bad way to come up with an analysis of which secondary stats are better. Any given kill, take A1S for example, can vary, simply because of RNG. Resins go out on different people. Spasers go out on different people. Number of jumps before kills vary. Etc. etc. Changing gear between one attempt and the next doesn't do anything to quantify the real differences between your potential. And this is what is important, your potential, not your execution.

If we were robots, and the fights had no RNG, and we executed everything exactly the same way every time, then comparing two parses would be a legitimate way to determine which gear (stats) were better. But in reality, we aren't. But you know what is? A simulation.

If you can maintain higher numbers with inferior gear, it is not an issue that the inferior gear is actually better, it is an issue with how you play the game.

And all of this analysis should be able to be proven out in the game. If the math says that SS is worth X and Crit is worth Y and Det is worth Z, that should actually be provable in the game. This isn't just some exercise for funsies, the whole point is to determine the actual outcome of something happening in the game, to determine a real value to something more than an anecdotal guess.

If the reality is that SS > Crit > Det, and someone can pull higher numbers with a Crit heavy set than an SS heavy set, that just means they are screwing something up when using the SS heavy set, or that the method for determining the value of SS vs Det was flawed. You're proposing a really bad hypothetical, simply to get behind the idea that a flawed human execution of something is more analytically accurate than a mathematical analysis.

[Dervy]

Exactly. I can use the exact same set and have a variation of 50DPS simply because of human error, latency, good vs bad damage ranges, critting all my high potency skills, the list goes on.

The only, true way, to test stats and different "builds" is in a vacuum where RNG and Crit are normalized, with human error removed from it.

Any other method isn't scientific. You need to eliminate as many variables as possible when conducting tests.

[Havenchild]

snip

There's two things kinda wrong with your statement. The first is, that there would be a better rotation than the one you found before that would better utilize the let's say for purpose SS, you have gained. In theory this is true, but in actual in game use it's not. Maybe on Striking Dummy? But in a fight? Highly Unlikely for the exact reason you mention, disconnects of a fight, RNG that can mess around with you and such. That is what slightly reduces the use of Spellspeed as well. How many casts it gave you over the course of the fight actually almost never changes or changes very little because of how fights play out. Now if we are looking at SS, like a modifier similar to DET, that's completely different.

The second way you are wrong, I would take you back to the 2.5+ series a sec. Many people stood by on calculators and sims alone, but several people could not pull the numbers I was able to do, when I would sacrifice yes, main stat for secondaries.
Many people claimed I was wrong regardless of reasoning and logical testing in game showed otherwise. It was only after I showed parse data relative to what I was saying, that people began to question how things actually worked again with SMN.

Turns out, that the old damage calculator was slightly off and the NA scene largely as a whole, didn't actually understand that DET scaled higher with more main stat. However the JP scene was more privy on this. Turns out I was proving the JP calculators in game without even knowing of either calculators existence. Have I tried builds from multiple people before? Absolutely, and executed my rotation the exact same way because even then, when Spellspeed had stronger influence on our cast times and such, it still never changed my rotation or gave an extra cast. Many people went full INT builds by maximizing ilvl, but that stuff was bloated with Spellspeed.

RNG has been existent in raiding for a long time. In T10 with charges/ tethers, T11 tethers, T13 Megaflare/ Earthshakers, T6 Flower debuff, T5 Dread stun....etcetc.

Now understand this does not take away from the fact that, now in Heavensward, Spellspeed is different, because it is. However, regardless, you should still go by what you see in game. A simulator is not server code. Nor are simulators always updated where as patch by patch, things can be tweaked behind the scenes without notice. If you would recall, stat scaling has been adjusted before between patches.

You don't have to trust what you see in game, but that's on you. However, I would strongly recommend against saying any Simulator or Calculator as the top result over what actually executes in-game better. That's not bad advice, that's kinda just common sense, I think.

Any other method isn't scientific. You need to eliminate as many variables as possible when conducting tests.

Do you guys think, that it is impossible, for someone to have little to no human error involved, when they understand the rotation they execute that much? I don't think so, and I feel a lot of people that post ridiculous numbers on DPS boards make it there with very little human error in between runs. You are definitely correct in, it's not based on science but what works in-game.

[Havenchild]

The only, true way, to test stats and different "builds" is in a vacuum where RNG and Crit are normalized, with human error removed from it.

Dont think I'm entirely disagreeing with you. When testing for coefficients and even potency per second,its instrumental to make a vacuum with static circumstances. However, what I'm saying is that gives you the general direction, that does not mean other lanes don't exist that can possibly be better outside of the vacuum which torin eluded to.

[T0rin]

snip

The simple reality is that if the game doesn't back up the numbers, then you need to look at the numbers. The odds of there being human error involved are just as likely in a mathematical analysis as there are in executing a rotation in the game. That doesn't take away from the merit of either, it just means that you need to be very diligent about both. If the analysis is done right, then nothing in the game short of player error should counteract the data. Going back to your pre-3.0 example.. if the analysis is saying one thing and the reality is showing another, that points to a potential flaw in the analysis, but does not discount the idea that the analysis should definitively be able to prove out what is analytically better than something else.

And this is comparing dummy parse sims vs real fights, of course there will be differences in the impact of different stats. What if you wrote a sim that actually simulated real fights with normalized RNG on mechanics?

[Sidra]

From a conceptual level, I am not understanding how Spell Speed could have been weighted so highly in the initial testing. It has no impact on Fester, Painflare, Deathflare. And we know that it has less impact on dots than determination does. It allows for more ruins, and somehow impacts the pets attack speed. But as crit and determination both also buff the pet in some way, it would seem like Spell Speed would be well behind the other 2 stats.

With that being said, SE is notoriously stingy with letting on how stats impact the game. I will let science do its thing here and see where it all ends up. I've definitely been wrong before. Thanks to everyone doing the hard work to figure this all out.

[Charybdis]

I'm not disagreeing about the concern/surprise over the Spell Speed weights. I am just as surprised myself. But there's a way to talk about it and a way not to. Everyone is curious/surprised/polarized. It's what drives us to DO the testing, find the formula, find the numbers, collect the data, etc. There's nothing wrong with Dervy, Judge, Sleigh, T0rin, and myself collecting this data. Speculation is also warranted and appreciated. Hell, speculation is the reason Dervy finds some of the things he does. But there's no need to bring negativity into the thread or say that the efforts we do are a waste of time. We don't deviate the thread so hard that people AREN'T getting help because they certainly still are. Discussion is good. We want discussion, we are not omniscient and may miss things. But don't discredit the hard work we are doing in the name of science.

Minorinz frustrated me not because of the doubt over Spell Speed, which is 100% justifiable. I have it. It was over the way they presented it, like doing the work we are doing is a waste of time which I don't agree with. Don't discredit the hard work being done to gain an understanding of "just how much" a stat is really affecting us. Or not.

[Charybdis]

Thanks again for all the work you are doing.

snip

I'll likely circle around to this after I finish the SS/Damage Pet model and then SS/DoT model. It seems that I misunderstood how OBVIOUS the recast timer reduction on SS works with pets. Hopefully it's just the same reduction of our own Spells/GCD which is already figured out. It's last on my list for the time being though.

[Bhuni]

DWT > Fester at 1 GCD in. Fester + Deathflare on the last Ruin II.

Well that's only really doable with the 2P1F opener as opposed to the 1P2F as Fester will be cooldown when you Trance, unless you Painflare last but by then your DOT's would start falling off before you can Tri-Diaster in Trance.

The opener aside I'm not sure about using Painflare in one Aetherflow cycle just so that you can Fester twice during Trance with 10% Magic increase.
Do the two 10% Festers even make up the 100 potency loss by using Painflare over Fester?