Page 2 of 8 FirstFirst
1
2
3
4
... LastLast
  1. #21
    So stacking ress in yellow gems for a priest healer, ís beter than 20 int/20 res?

  2. #22
    The Patient Thoelle's Avatar
    Join Date
    Aug 2009
    Location
    Denmark
    Posts
    233
    How does damage scale compared to this? See that would be cool

  3. #23
    using this data could you "estimate" how much resi you would need to be at where a 60 stamina gem would gain you more EH than a 40 resi gem?
    That is certainly possible, but as I said in the guide its not quite that cut and dry, increasing your EH by increasing your damage mitigation ALSO increases the relative effectiveness of heals on you (any given heal counts for more EH). Where as increasing your EH by increasing your displayed health through stamina only increases the total pool size. I may very well make a graph explaining it this week if I have time.

    So stacking ress in yellow gems for a priest healer, ís beter than 20 int/20 res?
    That is a question to be theory crafted by members of your class/spec, and ultimately a choice you need to make based on your own personal preferences and play style, do you want more survivability, or more healing throughput?

    How does damage scale compared to this? See that would be cool
    I have considered making graphs showing the trade off between dps/resil, as well as healing/resil, the issue is that these graphs would be different for every class, and for some classes like mages the DPS portion would need to be tested with multiple different itemization schemes (stacking mastery vs haste vs crit). I would need to find perfectly geared members of each class to run dps/healing tests on dummies 10-20 times to get decent sample sizes. All in all its just far too time consuming for me to do for every class/spec by myself, and it ultimately something I think is best left to people who actually play the class and know it better. Im not saying I will never do it, but its not in my immediate plans for the future.

  4. #24
    I have already sent many to look at this, thank-you for the work that went into this.

  5. #25
    Dreadlord Nightfury Treann's Avatar
    Join Date
    Sep 2009
    Location
    Colombia Spire
    Posts
    632
    i got a question about the classes:
    if i got 4k ress as a warrior, and a mage 4k, it is the same % reduction?
    "#Got99ProblemsAndTreannAintOne" ~ElyPop

  6. #26
    For resil, yes. But you factor in armor afterwards, so overall it will be different.........not to mention passive talents or buffs that reduce damage.

  7. #27
    i got a question about the classes:
    if i got 4k ress as a warrior, and a mage 4k, it is the same % reduction?
    As Asmodejjj said resilience scales exactly the same for every class, however a character's total damage reduction after resil, armor, and talents will be different based on your class as plate wearers obviously have more physical dmg reduction than cloth etc.

  8. #28
    4132 resil still doesn't stop frost mages and T2 idiots from blowing me up.

  9. #29
    This is seriously one of the best graphs for WoW I have ever seen related to PvP. Thanks for the read!

  10. #30
    I have a hypothetical question for you.
    effective health equals damage reduction times the current health right?
    This would mean that with very high resillience (around 10.000 so I am just asking this question hypothetically) the resillience drops causing the effective health to reduce, however the equation you use to calculate effective health does not have this effect.
    I'll give an example to what i mean: at 22,000 resillience my damage reduction is -18,05%. This would make my effective health 81,95% right? rather than 100% (or 900,52%) which your equation shows.
    Just curious if you agree with me.

  11. #31
    Dreadlord Nightfury Treann's Avatar
    Join Date
    Sep 2009
    Location
    Colombia Spire
    Posts
    632
    you mean if i got more health i receive more damage?
    or less?
    "#Got99ProblemsAndTreannAintOne" ~ElyPop

  12. #32
    I have a hypothetical question for you.
    effective health equals damage reduction times the current health right?
    No, effective health % is calculated like this: 100/(1-[dmg reduc %/100]). For example for my current resilience of 4807 the damage reduction is 45.70% so the equation looks like this: 100/(1-0.4570) = 184.16%

    This would mean that with very high resillience (around 10.000 so I am just asking this question hypothetically) the resillience drops causing the effective health to reduce, however the equation you use to calculate effective health does not have this effect.
    I'll give an example to what i mean: at 22,000 resillience my damage reduction is -18,05%. This would make my effective health 81,95% right? rather than 100% (or 900,52%) which your equation shows.
    Just curious if you agree with me.
    If im understanding correctly here you are pointing out that at extremely large resilience numbers the damage reduction equation starts REDUCING your damage reduction and will eventually go negative. Let me point out again that these equations are just approximations, they are not the actual equations used by the game, they just emulate them very closely. With that said the equation does go negative at extreme values that is true, its possible (even likely imo) that the actual equations used by the game go negative at extreme values as well. The simple explanation is that Blizzard doesn't care because no one will hit those values in Cataclysm, its just not possible, and when the next expansion comes out and resilience values inflate the scaling will change based on character level like every other rating does (crit rating, hit rating, etc).

    As for the equation for effective health continuing to go up despite the damage reduction going down that is simply because the equation used to approximate the curve is no longer accurate at those extreme values due to the way I created it. The equation in the graphs for effective health was created from the data points I got from the game in the actual playable resilience range. The equation is designed to approximate the curve of the data points in that range. If I were to run calculations to determine approximate data points for damage reduction and effective health outside the playable range (such as those extreme values), I could extrapolate and equation that would more accurately fit the larger curve, and I imagine it would be somewhat bell shaped (going up faster and faster then leveling off and going down quickly). I didn't do that simply because it would make the equation waaaay more complicated than it needs to be and really add no information of value.

    I hope this answers your questions, if I misunderstood you or you have further questions please feel free to ask and I will do my best to answer them.

  13. #33
    Thanks for answering my questions, you've made clear to me what I wanted to know, thank you.

  14. #34
    Thanks sooo much for this, never really understood resil scaling but now it seems simple (:

  15. #35
    Legendary! Blueobelisk's Avatar
    Join Date
    May 2011
    Location
    New York
    Posts
    6,671
    On your TI what kind of regression did you use?

    Why do you think its not just a linear regression?

    Your effective health is based off stamina ( hp ) and damage reduction alone right? Which means me as a frost Mage would have a worse relationship between resilience and effective health compared to a tank right?

    Don't take these questions as me being aggressive, I'm just asking out of curiosity because I considered finding an equation for resilience scaling before too.

  16. #36
    On your TI what kind of regression did you use?
    Im not really sure what you mean here, I didnt manually enter any type of regression. I used excel's trendline function to curve-fit an equation that closely matched the data points I pulled from the game, I then took that equation and entered it into the TI-83. Once in the TI-83 I used it's table function to enter X values and get the associated Y values for the equation.
    Why do you think its not just a linear regression?
    Well you can see for yourself the original data points do not have a linear regression, they look very much like quadratic polynomials in shape and that is what the equations that best fit were. That last graph in the guide is data pulled directly from the game, no calculation or extrapolation, just raw facts. My conclusions were simply drawn from observations of the data.

    Your effective health is based off stamina ( hp ) and damage reduction alone right?
    Not quite true, those are the primary factors however dodge, parry, armor, damage reduction talents, and the attacker's natural chance to miss all affect your total effective health as well. How all of these different factors stack together would take an entire guide itself to explain so its more than I can do here. This guide looks exclusively at how much your effective health is increased by resilience in percentages. When you calculate your effective health using these percentages and your base health you are only getting part of the picture, your total effective health is actually higher but it is different depending on the type of attack (melee vs spells) and character positioning (if they are in front of you for melee attacks). The best way to look at the number you generate from this guide is it is the minimum effective health you have, other factors may raise it higher, but baring damage increase debuffs (zerker and FC debuff) it should never be any lower. It is also actually fairly close to your total EH vs spells since there isnt much else affecting that damage (most casters are spell-pen capped in PVP).

    Which means me as a frost Mage would have a worse relationship between resilience and effective health compared to a tank right?
    The numbers in the above graphs are true regardless of your class or spec. You will have a lower EH than a tank simply because you will normally have less stamina and resilience than them. More resilience will also be more valuable to them because they already have more and resil has increasing returns as shown in the graphs.

    Don't take these questions as me being aggressive, I'm just asking out of curiosity because I considered finding an equation for resilience scaling before too.
    I don't take questions as attacks, peer-scrutiny is the best way to ensure that this kind of work is accurate. I am only one person, having other people look at the data exposes me to new perspectives and new ways of looking at it, which is always good. I hope I have answered your questions, if you have any more please feel free to ask.

  17. #37
    Legendary! Blueobelisk's Avatar
    Join Date
    May 2011
    Location
    New York
    Posts
    6,671
    Maybe you didn't know but your calculator can make trendlines for points like excel can. I guess I read too fast and assumed you backed up the trendline equation by using your calculator too.

    In case you wanted to know, Stat->Edit and list your points, 2nd quit to mainscreeen, stat -> scroll right to "Calc" -> and choose your regression, on main screen pick "L1 comma L2" and it'll show your equation.

    ----
    Yes I'm sure your numbers were all values that came from your own armor test by putting diff pieces of resil armor on/off to get your 33 values.

    I was just wondering why you chose quadratic. I had read somewhere...That the resil scaling was more complex than that, that there were multiple equations depending on what resilience you were talking about.

    But you're right, that line is VERY best fit. I'm sure that r-squared is close close close to 1. (If you know what I mean).

    ---
    Ahh I didn't mean effective health in general, I meant YOUR effective health as in the one you used in the graph. That graph EH is only based on Hp and damage reduction right?

    ---

    It's going to be interesting in 5.0 when they add base resilience to characters and also when they scale down items. It'll be interesting to calculate resilience then.

    Edit: LOL did you see that guy's post in the Warrior FC RBG thread? Linear till 32.5% then it has diminishing returns. 'K guess it is more complex than a single line, which I thought it was.

    I'll do some testing myself later.

    Edit 2:

    Here is the post he said:

    Quote Originally Posted by Blizz I guess?
    Resilience scaling has been modified for linear returns, as opposed to increasing returns. Under the new formula, going from 30 resilience to 40 resilience gives players the same increase to survivability as going from 0 to 10. Resilience now scales in the same way armor and magic resistances do. A player with 32.5% damage reduction from resilience in 4.0.6 should see their damage reduction unchanged in 4.1. Those with less than 32.5% will gain slightly. Those with more will lose some damage reduction, increasingly so as their resilience climbs.
    Yeah that shit isn't true. 2 PVP Ruthless rings, both 160 resil each test.

    0 resil to 160 resil -> 0.00% to 2.01% damage reduction = +2.01%
    160 resil to 320 resil -> 2.01% to 3.98% damage reduction = +1.97%

    Threw on some more gear and did the ring test again.

    2669 resil to 2829 resil -> 28.75% to 30.19% = 1.44% damage reduction
    2829 resil to 2989 resil -> 30.19% to 31.59% = 1.40% damage reduction

    Meh. Will have to do something of my own and compare it to yours at some other time. Is it linear? I'd have to do more numbers because it seems to be going down by .04% for each 160 resil you add. Is that "0 to 10 gives you the same as 30 to 40 resil" thing right? Meh I guess that's what diminishing return means.

    I'm just talking a lot, don't mind me.
    Last edited by Blueobelisk; 2011-10-27 at 01:12 AM.

  18. #38
    Scarab Lord Stanton Biston's Avatar
    Join Date
    Aug 2010
    Location
    Corvallis, Oregon
    Posts
    4,864
    Quote Originally Posted by Blueobelisk View Post
    Meh. Will have to do something of my own and compare it to yours at some other time. Is it linear? I'd have to do more numbers because it seems to be going down by .04% for each 160 resil you add. Is that "0 to 10 gives you the same as 30 to 40 resil" thing right? Meh I guess that's what diminishing return means.

    I'm just talking a lot, don't mind me.
    Read the first post again.

    Resilience scales slightly more than linearly as far as effective health is concerned.

    IE: 50% resilience = 200% health; 75% resilience = 400% health. So if it takes 5000 rating to get to 50% damage reduction, in a linear scaling EH model, it would take 10000 rating to get to 75%, even though at that point the rating is worth half as much raw % pts.

    EDIT: And they said the Defense stat was too mathy.
    Quote Originally Posted by Callace View Post
    Considering you just linked a graph with no data plotted on it as factual evidence, I think Stanton can infer whatever the hell he wants.
    Extraordinary Claims Require Extraordinary Evidence - Sometimes I abbreviate this ECREE

  19. #39
    Stanton just wait until you see the Stam vs Resil guide. I thought resil was complicated until I started working on this, which is ironic because I thought stam vs resil was going to be simple, man was i wrong. It turns out stamina does not have linear scaling on its own. I thought that 1 stam = 10hp 100% of the time and it's just not true at all. It turns out that 1 pt of stam gives roughly 13.9 hp, but its not linear, its actually a 5th degree logarithmic curve which was ridiculous to calculate.

    Here is a taste of the madness, the equation for stam scaling: Y=-58.3767792706829+40.341313886715*ln(x)-9.18011607426721*ln(x)^2+1.06148293401015*ln(x)^3-0.062146290953867*ln(x)^4+0.00146967913451671*ln(x)^5

    Like my other equations it is an approximation of the real deal based on curve-fitting but it is pretty darn accurate anywhere above roughly 500 stam, below that it's off.

    To make matters worse the final graph for resil vs stam is based on 3 axis of data (resil, stam, EH) because resil and stam both make each other more valuable making them interdependent for total scaling, so I either need to represent it with a 3D graph or a contour map, right now im leaning towards the latter. The end graph and the data itself is ultimately difficult to interpret, and drawing a meaningful conclusion from it varies from character to character as everyone has different amounts of both stam and resil. Anyways that's why it's taking me so long, figuring out how to present this data in a way that people can understand is a daunting challenge.

  20. #40
    Mechagnome Rec's Avatar
    Join Date
    Apr 2011
    Location
    Toronto
    Posts
    691
    Thorough and clearly presented. Posts like this make me wish we could +rep.

Posting Permissions

  • You may not post new threads
  • You may not post replies
  • You may not post attachments
  • You may not edit your posts
  •