combat changes 4.0 and sin tanks

[Evolixe]

You don't rely on mitigation as much as you rely on your ability to evade and avoid alltogether.

[KeyboardNinja]

Thanks. Hmm, seems I was under the false impression (since I have hardly played the other two tanks) that shadows rely more on shield/absorb for both spikiness and similar damage taken profiles with jug and pt tanks, while the other two have higher damage reduction even after shadows kinetic bulwark and shadow protection but their shield/absorb is less effective.

 

Well, it's a little more complicated than that. Remember that we have to be very careful about how we define "spikiness", since shadows actually have the lowest damage variance in the statistical sense. Dipstik and I had a really great discussion about this a few months ago, part of which is here: http://www.swtor.com/community/showpost.php?p=8138867&postcount=43 Basically, this graph gives a general intuition for what healers "feel" about the various tanks, whereas this is the statistical variance on mitigation. Note that, in the statistical sense, shadows have the highest peak and also the narrowest density, which means better averages and less variance within those averages (in other words, the exact opposite of the stereotype). This is different from the mitigation cases graph (the first one), which shows how the really great mitigation cases which shadows bring (much better than vanguards) lull healers into a false sense of security, broken by infrequent awful cases that are not much worse than vanguards and far less frequent, but more startling.

 

Defense is certainly a spikier stat, and its spikiness would show up quite starkly in the mitigation cases graph if I replotted in terms of what we know about 4.0 (probably not worth doing yet since we don't know full stat distributions and mod ratios). However, shadow spikiness is so minimal right now that I don't think it'll be an issue, even with the marginal increase in spikiness coming from lower shield/absorb totals. In 4.0, healing a shadow tank without losing your mind is going to come down to the same things it always has: trust in their mitigation and confidence that, as bad as things look now and again, the probabilities are low and the averages are spectacular.

 

Relative to the other tanks…* Shadows have DR which just barely falls behind guardians for practical cases, though is well behind Vanguards. It's their total mitigation package which is so much better though. Shadow defense, even with high shield/absorb builds, is generally within spitting distance of a guardian, while their shield/absorb is generally better than even a vanguard (whereas vanguards have awful defense, and guardians have awful shield/absorb). So, it's not so much that shadows rely on shield/absorb as they rely on their full mitigation package, including defense. If you stripped their defense bonuses away, shadows would be like a variant of vanguard that has 10% less DR, which is not a good situation to be in.

 

Having "excess" defense (which is what 4.0 is going to give us) isn't going to hurt us that badly. Vanguards, yes. Shadows, no.

[MusicRider]

Well, it's a little more complicated than that. Remember that we have to be very careful about how we define "spikiness", since shadows actually have the lowest damage variance in the statistical sense. Dipstik and I had a really great discussion about this a few months ago, part of which is here: http://www.swtor.com/community/showpost.php?p=8138867&postcount=43 Basically, this graph gives a general intuition for what healers "feel" about the various tanks, whereas this is the statistical variance on mitigation. Note that, in the statistical sense, shadows have the highest peak and also the narrowest density, which means better averages and less variance within those averages (in other words, the exact opposite of the stereotype). This is different from the mitigation cases graph (the first one), which shows how the really great mitigation cases which shadows bring (much better than vanguards) lull healers into a false sense of security, broken by infrequent awful cases that are not much worse than vanguards and far less frequent, but more startling.

 

Defense is certainly a spikier stat, and its spikiness would show up quite starkly in the mitigation cases graph if I replotted in terms of what we know about 4.0 (probably not worth doing yet since we don't know full stat distributions and mod ratios). However, shadow spikiness is so minimal right now that I don't think it'll be an issue, even with the marginal increase in spikiness coming from lower shield/absorb totals. In 4.0, healing a shadow tank without losing your mind is going to come down to the same things it always has: trust in their mitigation and confidence that, as bad as things look now and again, the probabilities are low and the averages are spectacular.

 

Relative to the other tanks…* Shadows have DR which just barely falls behind guardians for practical cases, though is well behind Vanguards. It's their total mitigation package which is so much better though. Shadow defense, even with high shield/absorb builds, is generally within spitting distance of a guardian, while their shield/absorb is generally better than even a vanguard (whereas vanguards have awful defense, and guardians have awful shield/absorb). So, it's not so much that shadows rely on shield/absorb as they rely on their full mitigation package, including defense. If you stripped their defense bonuses away, shadows would be like a variant of vanguard that has 10% less DR, which is not a good situation to be in.

 

Having "excess" defense (which is what 4.0 is going to give us) isn't going to hurt us that badly. Vanguards, yes. Shadows, no.

 

Hmm, I have underestimated the value of defence for shadows. Need to reconsider and think about it. Btw need titles for the axes. Not sure what I am looking at particularly in the second. First, P is in the common sense of probability? X-axis mit% is mitigation in percentage (yes I assume that x-values are 100*)? And what is each graph title (low, high, etc.).

[KeyboardNinja]

Hmm, I have underestimated the value of defence for shadows. Need to reconsider and think about it. Btw need titles for the axes. Not sure what I am looking at particularly in the second. First, P is in the common sense of probability? X-axis mit% is mitigation in percentage (yes I assume that x-values are 100*)? And what is each graph title (low, high, etc.).

 

Probability is the common sense of probability. As in, a higher value of P indicates that the case in question is more probable. All data points should sum to 1 (for this reason, the interpolation is slightly deceptive). The mitigation percentages are as you said, mitigation in that expected case. For example, you'll notice that there are 100% mitigation cases, and they have a non-zero probability, since defense and resist are a thing. You'll also notice that even the lowest mitigation cases are better than 0%, since we have armor and internal resistance. Finally, the low/mid/high/average labels indicate the damage profile used to weight the probabilities. Obviously, a dodge (100% mitigation) is a more probable case when a large percentage of the incoming damage is M/R, etc.

[MusicRider]

Probability is the common sense of probability. As in, a higher value of P indicates that the case in question is more probable. All data points should sum to 1 (for this reason, the interpolation is slightly deceptive). The mitigation percentages are as you said, mitigation in that expected case. For example, you'll notice that there are 100% mitigation cases, and they have a non-zero probability, since defense and resist are a thing. You'll also notice that even the lowest mitigation cases are better than 0%, since we have armor and internal resistance. Finally, the low/mid/high/average labels indicate the damage profile used to weight the probabilities. Obviously, a dodge (100% mitigation) is a more probable case when a large percentage of the incoming damage is M/R, etc.

 

OK, that's fine. The first plot is the probability mass function, by discretising mitigation (strictly speaking it should not have been a continuous line though although I can see why you want to fit a polynomial order 3 or 4 for shadows? in order to approximate the PDF, although there is no unique solution by doing so). However, not sure why it needs discriminated in to high-low damage as the amount of damage mitigation expressed in percentages is irrelevant to the initial values, and in any case what would be the damage interval boundaries. Also unclear to me why the PMFs in the four plots are different in this case.

 

Neverthless, if focus on the average, then this graph shows that shadows have smoother damage mitigation with guardian being the worst out of the three. Hmmm, interesting.

 

EDIT: Correction on inaccuracy.

Edited October 5, 2015 by MusicRider

[KeyboardNinja]

OK, that's fine. The first plot is the probability mass function, by discretising mitigation (strictly speaking it should not have been a continuous line though although I can see why you want to fit a polynomial order 3 or 4 for shadows? in order to approximate the PDF, although there is no unique solution by doing so). However, not sure why it needs discriminated in to high-low damage as the amount of damage mitigation expressed in percentages is irrelevant to the initial values, and in any case what would be the damage interval boundaries. Also unclear to me why the PMFs in the four plots are different in this case.

 

Neverthless, if focus on the average, then this graph shows that shadows have smoother damage mitigation with guardian being the worst out of the three. Hmmm, interesting.

 

EDIT: Correction on inaccuracy.

 

Ah, sorry! The low/mid/high/average distinction seems unclear. The magnitudes that I'm referencing are the relative amount of M/R damage. See the Ideal Tank Stats post for a more precise definition (basically, I'm splitting bosses into three buckets based on percentage of M/R damage vs F/T). That's why the values are different, since you're actually looking at very different mitigation profiles and probabilities.

 

In any case, you're right about the "shadows being the smoothest" conclusion from the PMF, which is part of why dipstik and I were digging deeper, ultimately settling on the discrete approach. The problem with the PMF is that it doesn't match the intuition that healers get when healing a shadow tank vs a vanguard tank. I don't think that particular intuition is invalid in any way, so the goal was to quantify it and identify exactly what in the class design and itemization was making shadows feel spiky, even when the math says that, statistically, they aren't. I mean, we already had PK (probability-of-kill), but that particular regression is a little too ad hoc to be useful in these sorts of measures, and it also ignores important attributes like mean mitigation. Hence, the discrete plots.

 

Incidentally, the polynomial fitting just makes the plot a little easier to consume, visually. It does imply some falsehoods, notably stemming from the fact that there is no "in between" data in a truly discrete plot. I still like it though because it reflects quite directly what healers do during a boss: namely, incrementally interpolate false states and probabilities from discrete events. As your eyes follow each individual curve, think about the way in which you might be interpolating between discrete mitigation events during a boss fight, landing at each point along the curve. The fact that the vanguard curve has the least inflection makes it feel more consistent, despite the fact that it is strictly worse in almost any objective metric. The shadow and guardian graphs have significant inflection (it might actually be useful to formalize the derivative of the fit function, to reduce this inflection to a single value) – especially shadows – and that inflection mirrors the intuition when healing a shadow that they are "spiky". They're not really spiky, they're just inconsistent in their magnitude of goodness.

[allexj]

Ah, sorry! The low/mid/high/average distinction seems unclear. The magnitudes that I'm referencing are the relative amount of M/R damage. See the Ideal Tank Stats post for a more precise definition (basically, I'm splitting bosses into three buckets based on percentage of M/R damage vs F/T). That's why the values are different, since you're actually looking at very different mitigation profiles and probabilities.

 

In any case, you're right about the "shadows being the smoothest" conclusion from the PMF, which is part of why dipstik and I were digging deeper, ultimately settling on the discrete approach. The problem with the PMF is that it doesn't match the intuition that healers get when healing a shadow tank vs a vanguard tank. I don't think that particular intuition is invalid in any way, so the goal was to quantify it and identify exactly what in the class design and itemization was making shadows feel spiky, even when the math says that, statistically, they aren't. I mean, we already had PK (probability-of-kill), but that particular regression is a little too ad hoc to be useful in these sorts of measures, and it also ignores important attributes like mean mitigation. Hence, the discrete plots.

 

Incidentally, the polynomial fitting just makes the plot a little easier to consume, visually. It does imply some falsehoods, notably stemming from the fact that there is no "in between" data in a truly discrete plot. I still like it though because it reflects quite directly what healers do during a boss: namely, incrementally interpolate false states and probabilities from discrete events. As your eyes follow each individual curve, think about the way in which you might be interpolating between discrete mitigation events during a boss fight, landing at each point along the curve. The fact that the vanguard curve has the least inflection makes it feel more consistent, despite the fact that it is strictly worse in almost any objective metric. The shadow and guardian graphs have significant inflection (it might actually be useful to formalize the derivative of the fit function, to reduce this inflection to a single value) – especially shadows – and that inflection mirrors the intuition when healing a shadow that they are "spiky". They're not really spiky, they're just inconsistent in their magnitude of goodness.

 

wait so defense chance is worth stacking? could have sworn it was almost worthless in pve and in pvp is was even taboo to stack such a thing?

[Aelanis]

wait so defense chance is worth stacking? could have sworn it was almost worthless in pve and in pvp is was even taboo to stack such a thing?

 

In PvE, it's not very good in 3.0 because of all the F/T damage rolling around, but that will change in 4.0, when M/R damage becomes much more prevalent.

[Jinre_the_Jedi]

wait so defense chance is worth stacking? could have sworn it was almost worthless in pve and in pvp is was even taboo to stack such a thing?

 

Still not worth taking in PvP.

[allexj]

Still not worth taking in PvP.

 

i was about to say >.>

[truevalon]

wait so defense chance is worth stacking? could have sworn it was almost worthless in pve and in pvp is was even taboo to stack such a thing?

 

In my experience; no. It is not worth stacking defense over shield/absorb. My philosophy for tanking has always been that I will take damage and my job is to reduce the damage I take as much as possible to make the damage taken predictable and smooth; easy to manage for healers. Stacking defense means you gear to not get hit, but when you get hit you will take more damage than those that gear for reducing the damage from hits.

 

KBN talked about lulling healers into a safe sense of security. Nothing does that better than defending 3-4 hits in a row, then taking another 3-4 in the face without proccing shield. Instances like this is why I chose to shelf my juggernaut and level a shadow instead.

[Zoom_VI]

Still not worth taking in PvP.

 

tbh if sorcs ever break their hold on the meta, I think defense will be something to take. Even in the current meta just over 1/3rd of your damage taken is M/R, and the effects of defending a attack is huge.

[Evolixe]

tbh if sorcs ever break their hold on the meta, I think defense will be something to take. Even in the current meta just over 1/3rd of your damage taken is M/R, and the effects of defending a attack is huge.

 

It is also stupid and game breaking. CC should not miss period.

[tXHereticXt]

I'd be okay with reliance on defense if we got some semblance of healing back - which, quite frankly, wont happen any time soon. If at all, so I'm not holding my breath. Arguably the only thing I enjoyed from Champions Online was how it handled defensive powers (Lightning Reflexes), it became a stat with its own DR but then there'd be no point for Absorption so it doesn't work with TOR.

 

Guess I'll be interested in seeing how this'll jumble my stats once 4.0 hits. Hell, SoR's new gear already did that already so my stats are all over the place at this point. From switching set bonuses to rearranging mods and enhancement distributions on the fly... I guess, looking at the bright side, it'll be nice to see something more streamlined instead of this wonky switcheroo nonsense.

Edited October 7, 2015 by tXHereticXt

[Zoom_VI]

It is also stupid and game breaking. CC should not miss period.

 

Is it more game breaking than a 30m CC on a 15 second cooldown?

[Jinre_the_Jedi]

Is it more game breaking than a 30m CC on a 15 second cooldown?

 

[Evolixe]

Is it more game breaking than a 30m CC on a 15 second cooldown?

 

Yes

 

In our defense, I don't think anyone ever asked low slash to be ninjathrow. But we won't complain naturally D:

[sithBracer]

Yes

 

In our defense, I don't think anyone ever asked low slash to be ninjathrow. But we won't complain naturally D:

 

Ignore him, he doesn't even play assassin, he plays VG and just comes here to troll for some reason.

[Iona_Var]

Technically....Lol slash is a 30m mezz, not a cc:p

 

I just wish they'd give hatred more range, like put deathfield, demolish and slow at 15m.

Edited October 10, 2015 by Iona_Var

[Jinre_the_Jedi]

Technically....Lol slash is a 30m mezz, not a cc:p

 

I just wish they'd give hatred more range, like put deathfield, demolish and slow at 15m.

 

Mezz is crowd control.

[Evolixe]

I honestly feel like DF/FiB never needed it's range nerfed after dotspread got taken off of it.

It's kinda doubled up now. Don't need to nerf and/and just either/or.

 

Aside from that, the range nerf to Demo/Vanq was more than justified and the slow is completely fine tbh.

 

 

Ps. I'm not sure how this became the topic at hand in a Combat/Tanking thread.. But I just go with the flow

Edited October 10, 2015 by Evolixe