Thread: Best in Slot for SMN after 2.1?

Originally Posted by T0rin

Well, this is discouraging, but helpful. Obviously my methodology for determining BiS is flawed, I guess the question is... what should be done to fix it? It seems based on preliminary results, that the old and new sets are very close without SS factored in and will be even closer with SS factored in, but obviously this disconnect between stat weights and damage output needs to be corrected.

Because there are several factors that theoretical values such as "int weight" alone cannot account for independent of class, rotation, and scenario.

One of the main things overlooked when comparing stats is that a number if pieces of gear with crit (the real BiS) also come with DET, such as the majority of hero accessories, the allagan weapon***, and the SMN doublet. So there is already a "moderate amount of DET" there to begin with.

I understand what you were trying to do was promote a superior result in overall damage, but our dots only take a small amount of time to cast, the rest of the time there IS a per-cast importance of damage on things like ruin II (especially when you try to filter in as many auto attacks as possible), where crit holds significant weight compared to det. DET doesn't even guarantee more damage, it just raises the two ends of the "damage range" things can hit for (very slightly)-- which, theoretically** (not as often in reality), will add to more damage-- but the majority of the time those two windows, whether one has a higher floor than the other, will hit about the same without regard to that floor. At least this is so when considering the base amount of DET that comes with the crit BiS.

Also your formulas do not account for burst, where crit plays a much more important role than DET.

While it's only a chance, and people seem to have a bias when comparing it to other classes like bard, a wise man once said "every little bit counts," and maximizing that crit chance will help more when you try to push a boss or a mechanic in a certain phase than DET's slight raise to the "consistent damage window."

Originally Posted by T0rin

Det affects auto-attacks, why would you think it doesn't? I mean, even something as simple as mousing over Determination on the Character screen

What I meant by DET having 0 effect on auto attacks is that compared to crit, it basically has no significance. I hit for maybe 5-7 less damage (as a window range) when I'm completely naked with the Relic +1. That's at 202 DET. At 282 DET, I hit for ~5-7 more. That means it can take around 12-15 DET to *maybe* raise the two edges of that damage window by 1. And like before, it doesn't guarantee that one window hits higher than the other. I find it impossible that raising the edge of that window by even +2 more, with sacrificing that much crit, can ever be worth it. One extra crit would be worth more than a lot of +2 auto attacks, which aren't even guaranteed to be +2 compared to the other floor.

I know you spent hours doing the theoretical math, but that doesn't make you right. It's also not nice to say something is "as simple as" when you are promoting a flawed system.

Originally Posted by Seobit

Not this shit again. Can we just ignore him this time please?

Also all hail Torin, savior of pantless summoners all over the world.

I understand that inexperienced players will sometimes jump to the newest bandwagon. Even when that wagon is propelled by completely arbitrary and flawed wheels

PS: Kobe

It should be pretty straightforward to put together a model for DPS output for a SMN, since it's not as "internally dependent on its own cycle annoying infinite mana regen bullshit" that characterizes BLM sequences.

If you want you can use my workbook I put together for the physical classes: http://www.fileswap.com/dl/qBoh3iIFB/

It's not particularly user-friendly but most of the stuff in there should be usable for all the basics (SMN direct damage, DoTs). The only significant changes that would need to be made is to add columns to simulate the pet with its independent cast rate (otherwise the pet would be trivial to add -- actually you could simulate it as the same cast rate as the player, but model the SS proc enhancement). Also something in the timings formulas to add time for Contagion.

Edit:

Alternatively you can try to get a hold of pandabearcat / panda's simulator which seemed reasonably solid and add in SMN definitions. IIRC that's programming-intensive though.

Originally Posted by Kicoe

Is there an average damage per cast for our damage spells mathed out somewhere? Crit vs Det BiS, crit hit rate & crit damage included?

There is, a page back. The CRT set has a 0.01% 'lead' in damage per cast over the DTR set, without factoring in SS, which is what I'm working on now. The DTR set has more SS, so in the end, it will likely make up the different and then some, but will not be very far ahead at all. I think they are going to be pretty damn close to equal, in the end.

Originally Posted by xxalucard

One of the main things overlooked when comparing stats is that a number if pieces of gear with crit (the real BiS) also come with DET, such as the majority of hero accessories, the allagan weapon***, and the SMN doublet. So there is already a "moderate amount of DET" there to begin with.

True.. but that doesn't affect anything. Itemization is one thing, and contributes to what makes a particular piece BiS, but in the end, it is the set as a whole that confers the most damage, that is BiS. I've never advocated for completely abandoning CRT. The whole point was to find the best mix of gear that would give the most damage. The power of CRT and WD is based on your INT and DTR. You cannot just go balls out on CRT, ignoring DTR (you can't really ignore INT) and expect to automatically have the most damage potential. DTR confers more damage on a point to point basis, but that doesn't mean that because Allagan Tunic of Casting has more DTR than Summoner's Doublet, that it is better. It isn't. The sum of the benefit of the CRT _and_ DTR makes it better.

We long held onto the idea that CRT was better than DTR on a point to point basis, that this other stuff aside, has not changed the reality that we were wrong. DTR comes in smaller quantities, and sometimes in larger sums on otherwise inferior pieces of gear. The point is not to simply go full out DTR and assume that is best. The point is to find the right mix of CRT and DTR that allows you to get the most damage.

Originally Posted by xxalucard

I understand what you were trying to do was promote a superior result in overall damage, but our dots only take a small amount of time to cast, the rest of the time there IS a per-cast importance of damage on things like ruin II (especially when you try to filter in as many auto attacks as possible), where crit holds significant weight compared to det.

But DTR affects all spells the same, DoTs, Ruin, Fester, etc. DTR is better than CRT for every single spell. CRT will never weight more than DTR on a point to point basis. It is mathematically impossible, the damage formula simply does not allow for it right now. Now, at some point in the future, when we have i250 gear with significantly higher INT and DTR totals, CRT can potentially out-weight DTR, since DTR does not scale. But given the gear that exists right now, the closest you can get CRT to DTR's 0.282081229 weight, is ~0.19. You'd need to add 250 iNT and 250 DTR to push the weight of CRT to the level that DTR is at. (and will remain)

Originally Posted by xxalucard

DET doesn't even guarantee more damage, it just raises the two ends of the "damage range" things can hit for (very slightly)-- which, theoretically** (not as often in reality), will add to more damage-- but the majority of the time those two windows, whether one has a higher floor than the other, will hit about the same without regard to that floor. At least this is so when considering the base amount of DET that comes with the crit BiS.

By your very defintion, it raises guaranteed damage. There is a 5% range of variation on damage. Sure, if you added 30 DTR, it is possible that a single cast (low roll) would be less than a single cast (high roll) of the same set without 30 DTR. That is an inane argument to make though, as we're not comparing single good vs single bad damage rolls to determine whether something is better. We're using averages, which when you play for hundreds or thousands of hours, is what matters. When you kill Ex Titan 30 times to get your earring, it doesn't matter that a single cast with a superior set was less than a single cast with an inferior set.. if you took the damage done over 30 titan comes, the superior set always comes out ahead. DTR increases min and max, and by extension, average damage. Just like CRT does.

You could apply the same scenario to CRT. One set with 450 crit casts 10 spells and crits 4 times, and a set with 550 crit casts 10 spells and crits 3 times. Does that make the first set better? No, of course not.

And yes, each point of DTR raises the min and max very slightly. Each point of CRT raises average damage done very slightly as well. It doesn't mean either are not worth it, because we're not talking about adding 1 or 2 points of each, we're talking about adding 50 or 100 or 150, at which point it does make a significant contribution.

Originally Posted by xxalucard

Also your formulas do not account for burst, where crit plays a much more important role than DET.

You cannot account for burst. Burst is front-loading as much damage as possible. CRT is not by definition better than DTR at burst. CRT could be better if you get lucky rolls. It could be worse if you got unlucky rolls. On average, it will be what it its average damage contribution is, and nothing more. You don't determine whether you can kill a conflag reliably by requiring lucky crits. CRT, by its very nature, is worse at burst simply on the merit that you are more susceptible to RNG than with DTR. When you have a DPS check like a 'fast' conflag, you're better off with the set that will guarantee you kill it in time, than the set that could potentially kill it faster if you get lucky, or potentially fail the DPS check if you get unlucky.

Originally Posted by xxalucard

While it's only a chance, and people seem to have a bias when comparing it to other classes like bard, a wise man once said "every little bit counts," and maximizing that crit chance will help more when you try to push a boss or a mechanic in a certain phase than DET's slight raise to the "consistent damage window."

CRT is more important on BRD, because the CRT procs make a significant impact on their ability to deal damage, in addition to the normal contribution it makes. That scenario does not apply to SMN, because the only proc we get on crits is a fairly weak one.

Originally Posted by xxalucard

What I meant by DET having 0 effect on auto attacks is that compared to crit, it basically has no significance. I hit for maybe 5-7 less damage (as a window range) when I'm completely naked with the Relic +1. That's at 202 DET. At 282 DET, I hit for ~5-7 more. That means it can take around 12-15 DET to *maybe* raise the two edges of that damage window by 1. And like before, it doesn't guarantee that one window hits higher than the other. I find it impossible that raising the edge of that window by even +2 more, with sacrificing that much crit, can ever be worth it. One extra crit would be worth more than a lot of +2 auto attacks, which aren't even guaranteed to be +2 compared to the other floor.

Why are you thinking about all of this in terms of one or two or ten attacks? Your RNG performance and potential to do better with crits is completely irrelevant to average damage output. Average damage is all that matters, because your performance does not boil down to 5 auto attacks or 7 casts of Ruin, it boils down to thousands upon thousands of them. And after thousands of them, all of this RNG bias is washed away, and all you are left with is averages. These formulas prove that on average, DTR makes a larger contribution than CRT. You simply cannot argue the merit of CRT vs DTR based on such a small set of data. It is inconsequential.

Originally Posted by xxalucard

I know you spent hours doing the theoretical math, but that doesn't make you right. It's also not nice to say something is "as simple as" when you are promoting a flawed system.

No, I spent hours casting the same spells over and over and over and over to collect the real data used to prove the theoretical math. Proving the math is important because only then can the math be relied upon as a basis for analysis of the damage done. If you can't prove the math, then you have no way to prove that anything is better than anything else, and then we're just having a completely pointless discussion. Math, as a means to prove value, is all that matters when it comes to issues of BiS. Nothing else can determine what is really best, and certainly not anecdotal "well I cast Ruin four times and did less average damage with 48 more DTR than I did with 73 more CRT" situations that you're describing.

I admit, my system of judging BiS is flawed. Obviously I didn't know that at the time. The end result and conclusion was flawed because of how the stat weights are applied and used as a measure for comparison. When you get back to average damage output, these 'new' and 'old' BiS are practically identical before you even factor in SS, which means that in the end, the 'new' is still going to better, though only very slightly. The work done wasn't all in vain, it proved a lot of assumptions false, and it brought to light the reality that INT > DTR > CRT > SS. That is an important step to the final conclusion about what is best.

Spent a few hours today programming a sim to parse out some psuedo realistic numbers for different sets... still trying to add more complexity to it, right now it just keeps all 3 dots up, festers during ogcds, aetherflows when it is off of cd and charges = 0(during ogcd), casts ruin 2 when an ogcd is needed and otherwise just spams ruin. Going to work in shadow flare and miasma2/contagion/pet attacks next. But got some preliminary data...

Simulation complete - stats = WD: 71, INT: 499, DTR: 262, CRT: 565, SS: 378, GCD: 2.46, PGCD: 2.96, total = 64955 damage (old BiS)
Simulation complete - stats = WD: 71, INT: 499, DTR: 305, CRT: 494, SS: 389, GCD: 2.45, PGCD: 2.95, total = 64972 damage (new BiS)

Really close.

Edit: Interesting... decided to compare my current gear (with 426 base accuracy) against the same set, but with myth and allagan pants... unexpected results:

Simulation complete - stats = WD: 71, INT: 490, DTR: 280, CRT: 563, SS: 408, GCD: 2.43, PGCD: 2.93, total = 66090 damage (overmelded vanya pants)
Simulation complete - stats = WD: 71, INT: 499, DTR: 279, CRT: 536, SS: 389, GCD: 2.45, PGCD: 2.95, total = 65055 damage (myth pants)
Simulation complete - stats = WD: 71, INT: 499, DTR: 262, CRT: 570, SS: 389, GCD: 2.45, PGCD: 2.95, total = 65201 damage (allagan pants)

I think spell speed is better than anyone (except maybe Kevee) has given it credit for.

To be honest, I use BIS threads to get a foundation or idea of where I should go with gear. So much of the supposed BIS is RNG related so I tend to get what I can and mix and match. We have run BC every week since it launched and have only seen the Allaghan head once (blm got it), the boots twice (yay), the ring twice (again yay) the belt, chest and legs zero times. Of course the chest is no big deal for me. I did get the hands, which I don't use yet due to accuracy cap and have only dropped once. Garuda and titan items are also RNG drops, no luck for me yet.

My tome items are guaranteed so I use what I can. Point being that I still do decent DPS and bosses still die. BIS is a guideline for me, but in the end, I can still only use what I have in my armory chest and if the end result is very close between CRT and DET sets, I am not going to sweat the small stuff. My thanks to those that have the head for numbers to do the math however.