How does armor currently work here?

Discussion in 'General Discussion' started by Cerebro, Oct 28, 2015.

  1. Mokli

    Mokli People Like Me

    Messages:
    1,755
    Better than Monks, that's for sure!
     
  2. rainessa

    rainessa People Like Me

    Messages:
    392
    I don't have a big strong tank to defend me :( A poor fragile bard must do what she can.
     
  3. Kovah

    Kovah Member

    Messages:
    33
    Sorry to beat a dead horse, but is there any chance we can change base to parry to /47.6 instead of /50? I have the following reasons:

    1. it would not change the base parry % for any classes (ie the base parry % would remain at 6% for War/Pal/SK, and 5% for bards)

    2. Illicits logs implied that a 10% parry modifier would help a knights chance to parry. Currently at /50, there is no increased chance for a knight to parry with a 10% mod. With a /47.6 base however, a knight would go from a 6% chance to parry to a 7% chance to parry (consistent with Illicits logs)

    3. currently, every riposte modifier, and every dodge modifier does something for at least one of the classes listed on the item. For the 13 items that have a parry modifier however, only 2 have a impact for the listed classes (Ring of reflex for war/bard and aegis of earth for war). That doesn't seem right to me. The other 11 items do absolutely nothing for the classes listed. If we change the base on parry to /47.6, I'm pretty sure every single one of these parry items would do something for at least one of the classes listed (which aligns with the results from riposte and dodge modifiers), except for shield of spiked energy which was removed from the game anyway. see Search EQ Items (allakabor.com) for all items with these mods. changing the base to /47.6 would make it so warriors and rogues only need a 1% increase in parry to get to the next tier, which allows even the lowly 2% modifiers to work.

    4. We originally had the parry base at /45. It was noted that this seemed to low, and it was bumped to /50. The logic behind the bump wasn't that /50 was absolutely correct, just that /45 was too low. 47.6 is a nice sweet spot between the two numbers, which seems entirely reasonable, and still maintains the base parry percentages (noted in point 1). It was mentioned previously that our formulas are likely not perfect. I'm not saying 47.6 is perfect, but I believe its the more correct choice given Illicits logs and the general logic of point 3.
     
    Cadsuane likes this.
  4. sowislifesowislove

    sowislifesowislove People Like Me

    Messages:
    762
    I think Torven did mention looking into changing it because basically even with POP out only like 4 items in the game woth those modifiers would ever effect those stats
     
  5. Placer

    Placer Member

    Messages:
    524
    With everyone at 65, what would be the various mod cutoffs for the classes? The orb of evasion I want to say it was called from VT did seem like it made a difference on a beast. But that might have been wishful thinking or the agility somehow.
     
  6. sowislifesowislove

    sowislifesowislove People Like Me

    Messages:
    762
    well it was realized recently that you achieve no more skill points into defense, riposte, etc from 61-65. So whatever works/doesn’t work will be the same in POP
     
  7. Placer

    Placer Member

    Messages:
    524
    I don't what the numbers were to/from but wasn't there a 10 difference in some things that got doled out 2 pts per level?
     
  8. Devour_Souls

    Devour_Souls People Like Me

    Messages:
    395
    "TAKP Resources" ---> "Skill Caps by Level"
     
  9. solar

    solar I Feel Loved Staff Member

    Messages:
    568
    Our skills aren't 100% right now, I've got some research I've done but not implemented yet. I believe the only one that should go up at 65 is Alchemy from 180 to 200.
     
    Neealana likes this.
  10. Torven

    Torven I Feel Loved Staff Member

    Messages:
    2,462
    The Illicit data also suggests that our current riposte mods are too good and his dodge and rip mods did nothing on AK. I'm not going to cherry pick a single data point to conform the logic to and ignore the others.
     
  11. Kovah

    Kovah Member

    Messages:
    33
    I don't believe there is any inconsistency in how Illicits logs show that dodge worked, and how its actually working. The Illicit logs implied the 8% dodge modifier did nothing for knights, and that is consistent with our current math that shows a knight would need a 17% increase in dodge to get any results.

    Riposte might not be working perfectly, and we can continue tweaking that, but should we not try and improve parry with the information we have in the meantime?
     
  12. Torven

    Torven I Feel Loved Staff Member

    Messages:
    2,462
    If I ever modify Parry then I'm nerfing Riposte with it. I'm not going to change the divisor to something obviously incorrect just to make mods work because that's what we want them to be. If I ever revisit this problem and figure out a way to narrow it down further to something that is more likely to be correct and it ends up nerfing a bunch of mods that shouldn't have worked to begin with then players will throw a fit.

    Without leaked source code or a running AK server, this is a problem that is not going to likely ever have a clear-cut answer. I'd have to try and play with the numbers and think of all the ways the formulae might have been modified and parse logs for more data. This involves spreadsheets and a lot of time. Time is something I'm running out of for this project and frankly this issue isn't worth much.

    If people want to come up with better fitting solutions, then great I'll be glad to review it-- but don't cherry pick data and be neutral about which way it ends up instead of trying to make the answer be what you wish it to be. The number of items with working mods would be something I would check for and I agree that should be considered. However I think it's likely that the divisors are multiples of 5 since the 4 known examples are and the formulae were made long before those items existed.
     
    Sketchy, Mechaike, Cillipis and 4 others like this.
  13. Dominar

    Dominar Member

    Messages:
    53
    How will the HP vs AC calculation change for different classes when the level cap goes to 65? Anyone willing to explain their formula?
     
  14. Pithy

    Pithy I Feel Loved

    Messages:
    2,113
    AFAIK, two factors will affect the HP vs. AC question in PoP.
    1. New defensive AAs. Innate Defense 1-5 increases mitigation AC (how hard mobs hit you when they connect) above and beyond Combat Stability 3. I believe it's a straight multiplier to mitigation AC: 10% for CS3, 25% for CS3+ID5. How that additional 15% multiplier on mitigation AC translates into the actual damage you take is a harder question and varies from mob to mob.
    2. New returns on AC over the soft-cap. The new returns are listed here. Warriors, for example, will get 33% over-cap returns in PoP, up from 20% in Luclin. Knights go from 16.7% to 25%. Monks go from 5% to 8.3%, etc.
    ID5 and higher over-cap returns make AC significantly more valuable in PoP than in Luclin. How much more valuable? Harder question. At a guess, though, I'd say that AC beyond the soft-cap is worth about 1.65 times more (33% is 165% of 20%) in PoP than in Luclin for warriors and 1.5 times for knights. So if a knight values 1 AC at X HP now, they might value 1 AC at 1.5*X HP in PoP. Or thereabouts.

    Edit: corrected an error that @solar pointed out below.
     
    Last edited: May 20, 2021
    Cadsuane likes this.
  15. Devour_Souls

    Devour_Souls People Like Me

    Messages:
    395
    Anecdotal, but as best I can tell, from what time I've spent on test, there really is no noticeable difference on the average 500 hitting PoP trash mob between max Luclin level, gear and AAs and PoP max level, gear and AAs. I'm sure that long parses would indicate that the new gear and AAs are doing something, but IMO it's not tangible without those parses.
     
    Last edited: May 20, 2021
    Mokli likes this.
  16. solar

    solar I Feel Loved Staff Member

    Messages:
    568
    What Pithy said is mostly correct. The Combat Stability line of AAs raise the soft cap number by a percentage allowing you to gain more benefit from equipment. They still require you to wear equipment to get the additional benefit. I fixed the monk AC (on the linked wiki page) and verified the rest of the numbers.
     
    Pithy likes this.
  17. Pithy

    Pithy I Feel Loved

    Messages:
    2,113
    Oh, interesting, thanks @solar. I'd forgotten how CS/ID entered into the mitigation calculations. Sounds like Innate Defense in PoP won't really affect the "1 AC = X HP" question, since AC soft-caps are generally easy to hit in bazaar gear. Only the new over-cap returns will affect the incremental value of AC on gear upgrades.
     
    Last edited: May 20, 2021
  18. showstring

    showstring I Feel Loved

    Messages:
    2,373
    FINALLY YES! WE DID IT GUYS
     
    Walex, Break, Crushzilla and 2 others like this.
  19. Neealana

    Neealana People Like Me

    Messages:
    1,141
    And we welcome our new monk overlords...
     
  20. Dominar

    Dominar Member

    Messages:
    53
    After rereading this thread and a few other linked threads I did my own analysis of AC vs. HP. I attempted to the relationship between the change in AC vs the change in damage. I assume you start at full HP and go to 1 (or 0) HP after many hits without a heal. This does not consider the efficiency benefit of higher AC when using partial heals, or heal over time. This does not look at the value of Avoidance AC. This is just attempting to value raw AC vs raw HP.

    In order to compare the HP value of two different AC values, I copied the RollDI20 function and did many simulations of 1 million rolls at different AC vs ATK ratios. I then looked at every sequence of 10 hits and summed the 10 rolls. Next I extracted the 90 percentile sum, 99 percentile sum, and 99.9 percentile for the sequences of 10 hits. Finally for many pairs of runs, I computed the change in mitigation AC vs the change in the sum of 10 swings from the DI20 function.

    I plotted the result and saw that when mitigation AC was approximately equal to NPC attack (min hits and max hits equally likely) a 1% increase in mitigation AC reduced the variable damage by 0.4% - 0.6%. When NPC attack was much higher (up to double player AC), the return on AC caused a reduction in damage of 0.15% - 0.3%. However when player AC was up to double the NPC attack, the returns of an additional 1% increase in mitigation AC reduced variable damage by 0.5% - 0.8%. Note, this only considers variable damage (DI) and not the base damage (DB) which is never mitigated by AC. Ignoring this will overvalue AC slightly, as HP are the only thing that matters to be able to absorbing many minimum hits.

    In computing the final result, there are 3 additional number to explain:
    4/3 --- because reasons. 3 Raw AC gives 4 mitigation AC.
    0.15% - 0.8% --- This is the mitigation AC return for reducing the variable damage. This depends on whether you have relatively high AC (0.5% - 0.8%) or relatively low AC (0.15% - 0.3%) compared to a given NPC.
    5% - 100% --- softcap return. Below the softcap the return is 100% (ignoring low level caps). Above the softcap the return is 5% - 33% depending on class and level.

    Using these numbers compute = total HP / (total mitigation AC) * 4/3 * mitigation return * softcap return

    Here are some results that were relevant to me assuming a mitigation return of 0.6 (low attack NPC and/or high AC player) for level 60 characters.
    Overcap plate tank = 1.5 - 2 HP per AC
    Overcap melee chars = 0.5 - 1.1 HP per AC
    Overcap priests or caster = 0.35 - 0.4 HP per AC
    Undercap priests or caster (or shield AC) = 7 - 10 HP per AC

    When the level cap increases, some classes will see softcap returns increase, making AC up to 50% more valuable compared to AC. Also the new AA (ID5) will move the softcap up another 15% on top of CS3+PE which moved the cap up 12%.

    Tldr; AC is extremely valuable until you hit the softcap, 1AC = 6-12 HP, but when you exceed the softcap, 1AC is worth 1/3 - 2 HP depending on the class return rate with plate tanks seeing the best returns from AC.
     
  21. Pithy

    Pithy I Feel Loved

    Messages:
    2,113
    Cool analysis, @Dominar! The "1 AC = X HP?" question is timeless and, IMO, fascinating :) @Loraen and I made similar attempts to answer it via simulation back in 2017ish. Raev used a spreadsheet, I used Matlab.

    Before digging into the particulars, I think it's worth taking a step back and looking at the context and framing. Do we care about raid mobs or trash mobs? Warriors or knights? If warriors - evasive, defensive, or no disc? What's the key metric to use for this analysis? Is it the average DPS taken, the longest CH rot delay we can get away with while ensuring tank survival with high probability, or something else?

    Those are all just design choices, ways of framing the question. Different answers to those questions necessitate different types of analysis and will lead to different answers to "1 AC = X HP?".

    For my part, I think the most interesting contexts are (1) warriors discing vs. raid mobs, and (2) knights tanking trash mobs. In Case 1, I think the best metric is the longest CH rot delay you can get away with while ensuring tank survival at some high probability, say 99%. In Case 2, I'm not quite sure -- average DPS matters for stuff like mana efficiency in exp grinds, while how long you can survive without a heal matters for tanking trash on raids.
     
    Last edited: Jun 28, 2021
    Ikeren likes this.
  22. Pithy

    Pithy I Feel Loved

    Messages:
    2,113
    Oh, another interesting question is "How much does it matter?" Pick a context and a survivability metric -- say, warriors disc tanking raid mobs and max CH rot delay -- and run the numbers for a toon with max HP raid gear, one with max AC raid gear, and one that "optimally" balances HP vs. AC. How much does survivability vary between the three?

    I'm guessing it's a few percent difference tops, but who knows. It'd be cool to see some numbers.
     
    Last edited: Jun 28, 2021
  23. Pithy

    Pithy I Feel Loved

    Messages:
    2,113
    A couple of random points in favor of HP vs. AC:

    AC only helps with melee damage, while HP helps with spell damage. Some PoP gods have big-ass procs; AC is useless there. Resists help, but only to a point. HP is the god stat for spell damage.

    In the limit of complete risk-aversion -- meaning we don't just want 99% probability of tank survival, but 100% -- AC doesn't factor into the calculations at all. To guarantee tank survival, all that matters is the tank's hitpoints and the mob's max hit and swing rate, because in the worst-case scenario you're going to get hit for max every swing. In that limit, HP is all that matters.
     
    Phaeton and Break like this.
  24. Fadetree

    Fadetree People Like Me

    Messages:
    456
    Echoing Pithy above: to me, EQ's handling of (generalized) AC reduction in damage from armor has always seemed strange, in that it is such a small percentage. Robes and things like that I sort of get, but full plate tank armor really from a % view to me seems kind of minor. I know over really long fights and grinds that it can make a difference in keeping healer mana up, but with the insane DPS of some mobs you're living on the slows and cheal rotations alone anyway pretty much and that ogre might as well be nekkid, especially if it's at max level.

    I think one of the things that made EQ really great was their apparent paranoia about runaway gear effects overpowering everything. I think before they got overwhelmed by mudflation they were really trying to keep gear effects more or less minimally noticeable. They didn't do such a good job on spells and of course emergent gameplay is gonna happen no matter what you do, but the gearing was a subtle aspect that kept the lid on explosive mudflation for a relatively long time.

    Plus I admire the battle mechanics of grouped ranges of hit levels with only a relative small distribution of randomness in each level...it helps scaling and keeps things from getting out of hand. Designing my own battle mechanics lately, I have run into a number of problems that they had solved pretty well.
     
    Last edited: Jun 28, 2021
    Phaeton, solar and Pithy like this.