View RSS Feed

Blogenstein

Affliction Warlock: Spell Hit Evaluation

Rating: 4 votes, 5.00 average.
Continuing the Bink Learns About Other Classes Project, as a followup to the previous Affliction Rotation post, Iím looking at Affliction Stats. First off, I should make a note that switching the priority of Haunt in the rotation turned out to be a DPS loss due to the reduced uptime of the spell, which happened because Soulburn was used more often & using up soulshards that would otherwise go to Haunt.

Anyway, the two primary topics for any Affliction Stat post are Haste & Hit. In both cases itís a question of what values to aim for & whether you should add more or go for Mastery instead, with some contention about how much of either you should take. Since this post has become slightly larger than I first planned and I still have more haste plots to generate to nail down a few theories, Iíll just stick with the Hit section for today.

Hit
Hit is perhaps the easier one to talk about, so Iíll start there. Typically Hit is the most important stat for any spell caster, as individual spells do quite a lot of damage in Mists. For Affliction, it has a lower value which drops it down around or even under Haste & Mastery. This is because every Affliction ability is either a DoT or a channeled cast, rather than the cast/single hit spells that are the mainstays of other classes. The result is that the damage lost by having to recast an individual spell that missed is only one tick, which can be anywhere from 25% of the total spell damage to less than 10%.

Even though Agony, Corruption and Unstable Affliction have the highest damage return from time spent casting, the actual DPS contribution of these spells is a lot lower, in the order of 11-14% each, because of the long duration of the dots. This means that adding an extra GCD of time to recast a dot is wasting that 11-14% of your dps for just over 1 second. In the case of the channeled spells like Malefic Grasp, they simply fail to start, which only wastes about 25% of the possible damage while you recast. Itís for these reasons that hit isnít worth a lot more than haste or mastery.

Itís for these reasons that some people argue in favour of running Affliction without reaching the hit cap, and going on purely quantitative reasoning theyíd be right. However, the default profiles are already at the hit cap so the value generated will be the absolute minimum it can be. Itís worth noting that the value of hit diminishes as you add more, so going from 98% to 99% total hit is more valuable as going from 99% to 100%. Does this mean that you want to have ~12% total hit instead? (SimCraft uses +/- 1000 to calculate weights, so 1000/340 = 2.94% hit)

So do I want to be hit capped?
The answer is yes, youíll want to be hit capped. The reasons for this are all qualitative and boil down to ďmissing important spells really sucksĒ. Haunt and its 25% damage bonus are certainly important, especially since it requires soul shards to cast, which are scarce enough to begin with. Likewise, Soulburn & Soulswap are fairly important, especially since each dot re-applied has its own chance of missing. Lastly, Agony itself is fairly important to keep rolling at 10 stacks, so if you push your recast too close and it misses youíll lose DPS.

Certainly, on average hit will be worth 90-95% of Haste or Mastery, but the impact of events like these on individual fights will be much higher as the value of Hit is an average of encounters where missing had an impact and those where it had absolutely no impact at all. Coming back to stats, if you have a 1% chance to miss, youíd expect to miss 1 in every 100 spells. However, thereís only a 36.6% chance you wonít miss at all (0.99 * 0.99 * 0.99 * Ö. or 0.99 ^ 100), which means 63.4% of the time youíll miss at least once, and a 26.5% chance youíll miss two or more times (100 * 0.01 * 0.99 ^ 99 which is the chance for 100 different combinations of 99 spells hitting and 1 spell missing in a set of 100).

Submit "Affliction Warlock: Spell Hit Evaluation" to Google

Comments

  1. Unregistered's Avatar
    AskMrRobot blog had an article with regards to hit-cap that you might find interesting to read.


    http://blog.askmrrobot.com/2012/11/h...bout-hit-caps/
  2. Binkenstein's Avatar
    Quote Originally Posted by Unregistered
    AskMrRobot blog had an article with regards to hit-cap that you might find interesting to read.


    http://blog.askmrrobot.com/2012/11/h...bout-hit-caps/
    I'm aware of their position, but they're still taking their purely quantitative approach to the whole matter, which is a side effect of people not understanding how stats scale, etc.

    I'm not sure how exactly they optimise stats, and they have a habit of massaging stats to make their system work.
  3. Kaedis's Avatar
    Note that going from 98% to 99% hit is NOT twice as valuable as going from 99% to 100%. Even ignoring the effects of Pandemic (which greatly diminishes the value of hit), missing 1% of your spells is a 1% damage loss. Thus going from 98% to 99% hit is a 0.99/0.98 = 1.020408% damage increase, while going from 99% to 100% is a 1.0/0.99 = 1.010101% damage increase. Thus 98->99 is at most 0.010204% more valuable than 99->100.

    On the qualitative argument side, SimC accounts for misses. The potential to miss a Haunt (or a Soulswap) is outweighed by the increase the rest of your damage. Incidentally, that argument that SimC's profiles are already hitcapped (and therefore improperly valuing hit) is flat false. SimC uses a negative delta for capped stats (like Hit), so it's actually simming *removing* 1000 hit. Beyond that, it's very easy to adjust the profiles to use lower hit values for comparison. Doing so nets a small but very present DPS increase (on the order of about 0.8-1% for a reduction of hit rating from 15% to 10%).

    Agony is of the least concern for hit rating. Assuming the 4717 hit cap, you'll have precisely 12.0 seconds (raid-buffed) to refresh Agony from the first time it becomes eligible without DPS loss through Pandemic until it actually falls off. If you're waiting until the last tick, you're not playing Affliction properly (Pandemic was added for a *reason*). With lower hit levels, you simply need to ensure you're starting your refresh attempts earlier.

    Now, your argument on fluctuation is valid. Dropping below the hitcap introduces RNG where there existed not before, which means an increase in the standard deviation of your DPS. However, since SimC is the average of many iterations (including ones with bad strings of misses), this is both a positive and negative standard deviation. Some fights, your DPS will be even higher than the DPS increase SimC predicts, others it will be lower. However, this noise is *completely* overwashed by the inherent RNG associated both with spells with variable damage ranges and crit itself, and thus is an extremely poor argument for ignoring SimC's results.

    Note, however, that the net DPS gain of being below hitcap is still rather small, so if you don't *want* to have to deal with misses and recasts, hitcapping really isn't that much of a DPS loss. It still *is* a DPS loss, though (however small), so attempting to play it off as simply an eccentricity of SimC is simply dishonest, particularly with this blog's seeming dedication to using theorycrafting to back it's statements.
    Updated 16-12-2012 at 08:16 AM by Kaedis
  4. Binkenstein's Avatar
    Quote Originally Posted by Kaedis
    Note that going from 98% to 99% hit is NOT twice as valuable as going from 99% to 100%. Even ignoring the effects of Pandemic (which greatly diminishes the value of hit), missing 1% of your spells is a 1% damage loss. Thus going from 98% to 99% hit is a 0.99/0.98 = 1.020408% damage increase, while going from 99% to 100% is a 1.0/0.99 = 1.010101% damage increase. Thus 99->100 is at most 0.010204% more valuable than 98->99.

    On the qualitative argument side, SimC accounts for misses. The potential to miss a Haunt (or a Soulswap) is outweighed by the increase the rest of your damage. Incidentally, that argument that SimC's profiles are already hitcapped (and therefore improperly valuing hit) is flat false. SimC uses a negative delta for capped stats (like Hit), so it's actually simming *removing* 1000 hit. Beyond that, it's very easy to adjust the profiles to use lower hit values for comparison. Doing so nets a small but very present DPS increase (on the order of about 0.8-1% for a reduction of hit rating from 15% to 10%).

    Agony is of the least concern for hit rating. Assuming the 4717 hit cap, you'll have precisely 12.0 seconds (raid-buffed) to refresh Agony from the first time it becomes eligible without DPS loss through Pandemic until it actually falls off. If you're waiting until the last tick, you're not playing Affliction properly (Pandemic was added for a *reason*). With lower hit levels, you simply need to ensure you're starting your refresh attempts earlier.

    Now, your argument on fluctuation is value. Dropping below the hitcap introduces RNG where there existed not before, which means an increase in the standard deviation of your DPS. However, since SimC is the average (including ones with bad strings of misses), this is both a positive and negative standard deviation. Some fights, your DPS will be even higher than the DPS increase SimC predicts, others it will be lower. However, this noise is *completely* overwashed by the inherent RNG associated both with spells with variable damage ranges and crit itself, and thus is an extremely poor argument for ignoring SimC's results.

    Note, however, that the net DPS gain of being below hitcap is still rather small, so if you don't *want* to have to deal with misses and recasts, hitcapping really isn't that much of a DPS loss. It still *is* a DPS loss, though (however small), so attempting to play it off as simply an eccentricity of SimC is simply dishonest, particularly with this blog's seeming dedication to using theorycrafting to back it's statements.
    Good point on the percentages.

    I'm not dismissing it as an eccentricity of SimCraft, but rather how people interpret those values, combined with the fact that players generally can't react as fast as the simulation when spells miss.
  5. Kaedis's Avatar
    Quote Originally Posted by Binkenstein
    Good point on the percentages.

    I'm not dismissing it as an eccentricity of SimCraft, but rather how people interpret those values, combined with the fact that players generally can't react as fast as the simulation when spells miss.
    Depends on how they are simming them. SimC has varying levels of "player skill" selectable, which provide varying chances to "oops" on a priority item and fail to recognize that that item should be executed. Elite is the default and has a 0% chance of this. I personally run it on "Good" instead, gives more realistic results (and those results still show Mastery > Hit).
    Updated 16-12-2012 at 08:22 AM by Kaedis
  6. Binkenstein's Avatar
    Quote Originally Posted by Kaedis
    Depends on how they are simming them. SimC has varying levels of "player skill" selectable, which provide varying chances to "oops" on a priority item and fail to recognize that that item should be executed. Elite is the default and has a 0% chance of this. I personally run it on "Good" instead, gives more realistic results (and those results still show Mastery > Hit).
    Ah, but by how much? Most sims I've run put both spells within the error margin.
  7. Kaedis's Avatar
    Quote Originally Posted by Binkenstein
    Ah, but by how much? Most sims I've run put both spells within the error margin.
    Assuming you meant "stats" instead of "spells":

    At last sim (Daerius - US Lothar), I had a value of 2.00 for Mastery and 1.62 for Hit, over 10000 iterations. Error bars are 0.13 for hit and 0.09 for Mastery, which is barely half of the delta between the two values.
  8. Binkenstein's Avatar
    Quote Originally Posted by Kaedis
    Assuming you meant "stats" instead of "spells":

    At last sim (Daerius - US Lothar), I had a value of 2.00 for Mastery and 1.62 for Hit, over 10000 iterations. Error bars are 0.13 for hit and 0.09 for Mastery, which is barely half of the delta between the two values.
    Just re-ran T14H, which shows 2.71 Mastery, 2.59 Haste & 2.32 Hit for a hit capped profile.

    Even with the weights showing like this, I still stand by my assertion that hit is worth capping for Affliction for qualitative reasons, ie: hit is the easiest stat to observe the impact of when you don't cap it.

    I can't view your affliction gear as you're currently logged out as Demonology.
  9. Kaedis's Avatar
    Quote Originally Posted by Binkenstein
    Just re-ran T14H, which shows 2.71 Mastery, 2.59 Haste & 2.32 Hit for a hit capped profile.

    Even with the weights showing like this, I still stand by my assertion that hit is worth capping for Affliction for qualitative reasons, ie: hit is the easiest stat to observe the impact of when you don't cap it.

    I can't view your affliction gear as you're currently logged out as Demonology.
    Ya, sorry, it's actually the same gear as my Aff set, I just tend to log out in that Demo spec because I use it for soloing. Makes it a pain to do an import, though.

    On the hit note, I *do* agree that hitcap has qualitative benefits, not the least of which is not having to pay attention for and react to misses, which allows far more attention for boss mechanics and for optimizing proc usage on DoTs. Depending on the user, these more abstract effects *may* overpower the additional DPS benefit of base stats inherent in sub-cap. Personally, *I* hitcap, despite my argumentation against it, because I'm also the raid leader, so I *need* to have as much attention as I can on the fight, since I have to not only optimize my own performance, but watch the overall progress of the fight and keep the raid organized and working together properly.

    I'm just arguing that for *true* ideal-scenario optimization, hitcap is sub-optimal. For those with Dalek-level concentration, it's proper to not be hitcapped. For most people, whether you are hitcapped and if no how far below the cap you are should be entirely based on how well you can pay attention for misses while also managing fight mechanics properly.
  10. Unregistered's Avatar
    Please keep in mind that 99.xxx % of published/quoted scaling values from SimC are from default options settings like 'Patchwerk type fight', 'single target' and 'Next to perfect player'. As soon as you play with these options towards 'high movement', 'multiple adds' and 'just me' scaling values tend to shift from mastery>>haste>hit towards haste>=hit>mastery.

    But as today's habit in nearly every niche tends to break things down to just 1 number, we'll have to live with 'old school' vs. 'new age pros'. Just my 2 cents
  11. Binkenstein's Avatar
    Quote Originally Posted by Unregistered
    Please keep in mind that 99.xxx % of published/quoted scaling values from SimC are from default options settings like 'Patchwerk type fight', 'single target' and 'Next to perfect player'. As soon as you play with these options towards 'high movement', 'multiple adds' and 'just me' scaling values tend to shift from mastery>>haste>hit towards haste>=hit>mastery.

    But as today's habit in nearly every niche tends to break things down to just 1 number, we'll have to live with 'old school' vs. 'new age pros'. Just my 2 cents
    That's why I usually run them with "LightMovement" as the fight type.
  12. Kaedis's Avatar
    Quote Originally Posted by Unregistered
    Please keep in mind that 99.xxx % of published/quoted scaling values from SimC are from default options settings like 'Patchwerk type fight', 'single target' and 'Next to perfect player'. As soon as you play with these options towards 'high movement', 'multiple adds' and 'just me' scaling values tend to shift from mastery>>haste>hit towards haste>=hit>mastery.

    But as today's habit in nearly every niche tends to break things down to just 1 number, we'll have to live with 'old school' vs. 'new age pros'. Just my 2 cents
    Do note that Affliction has perhaps the *least* loss due to movement thanks to Kil'jaedon's Cunning and a high dependency on long channeled spells. Player skill doesn't affect the mastery/haste values much, and multi-target fights actually significantly *amplify* the value of Mastery, as a much larger portion of your damage comes from your DoTs on those fight (though multitarget also amplifies the importance of haste thresholds considerably as well, as it's far more common for your DoTs to fall off). I will agree that for *most* players, hit will be of higher value than the sims suggest, as most players will make mistakes, particularly in high-attention phases of fights. I can't think of any possible way that Mastery would tank that much, though. It simply scales too well.

Trackbacks

Total Trackbacks 0
Trackback URL:
Powered byvBSocial.com and MMORPG