The Mitigation Parse thread (discussion)

[Swansong]

Originally Posted by Kreynium

Well to see so many dif parses has anyone got the formula for high end ac raiding yet?

I see mostly old material here.

Is there a cap..hard? soft?

The Hard cap is 3226ac.

[Bedavir]

There is no easy answer, but there really are no noticable caps on ac at this point in time. It's up to each warrior to decide how much or little they want. What we can say for sure from the data gathered so far is that there are beneficial returns from ac at all levels. Even Sulain and I (2 highest serverwide) benefit from each additional ac. I wish it were as simple as a single formula ... but it isn't. There is no such thing as 'enough ac' just like there's no such thing as 'enough hp'. Increases in either will continue to affect tank survivability.

What I can safely say, though, is this: I could be in the 19k-range unbuffed if I had chosen a different route and haven't yet regretted my choice to pursue ac. It really does trivialize tanking in EQ.

[Lluianae]

Old material works too, since the highest ATK in EQ atm was found in Omens of War and the highest has not increased since (we only don't know against what). ATK tends to vary it seems but in some cases not as much as DB/DI are.

For now however, there is no such thing as a hardcap, as to perhaps more 'softcaps' as diminishing returns progresses I can't say.
The mitigaiton formula of a player v mob however is being worked on! As things come more into focus, I'm pretty sure knowing the Knight return is going to be really helpful (supposedly approx 1:3 for anyone interested).

If possible, can these be shifted to the discussion thread please Yoda?

[Raaj]

Moved posts from the deposit thread to here, so if you're confused by some of the replies or something seems out of place - that's why.

[Lluianae]

Thanks Raaj

[Tatum]

Originally Posted by Bedavir

Doh nm ... I see that now. Defensive parses (while not completely useless) aren't terribly helpful for comparison sake. Unless you go in by hand and restore everything to the proper DI it just muddies the water.

Tatum for future fights where you know you'll be parsing are you comfortable using evasive instead?

Depending on what the target is I can, not sure i would live through razorclaw evasive tbh.

[Daze]

Thanks for posting this stuff, i used to be an AC whore., and i'm gonna be one again

[Bedavir]

Originally Posted by Pershon

From what I understand you are merely trying to eliminate the chances of receiving a killing round, correct? It means that you are lowering the chances of taking a large flurry and quint/quad, but that that combo will still kill you. Doesn't that also increase the chance that the said quint/quad will kill you b/c you don't have the hps to survive them?

If a mob hits for 5k and lands a full quad on me I'm still alive. If a mob hits for 6k and lands a full quad on me I'm dead. If a mob hits for 6k and lands a full quad on a hp tank from my same progression level, he's dead too. Can hp tanks at my level buff to 24k+? 6k quad hitters are far from the worst mobs. BBlade was telling us how he survives a 6400 quadder + flurry, yet he doesn't have more than 36k hps despite her max theoretical round being well over 37k. My chances of receiving a quad assuming a 0.5% chance average DI20 hit (which is right about what I get on most mobs give or take a tenth of a percentage point) is 0.0000000625%. That means that over 1 in 1,000,000,000 hits statistically will be a DI80 round max quadded round. Say you had a 2% chance of average DI20 hit ... your chances of a DI 80 round are 0.000016%, which means that your chances of a DI 80 round are one in 1,000,000. One in a few billion vs a million.

So what's the take away? Nobody is at any real risk of DI80 full quad rounds ... so even using them in your argument negates it's practical value.

I still never saw any numbers on what the odds are that those rounds happen on say me versus someone with 500 more ac.

I provided a pretty good example just a few days ago, and you can find it seated at the tail end of the stickied thread up top. The difference in unbuffed duel wield ac was 750, but one tank used a shield the entire time while the other used it for enrage. (bringing the level close to but not quite at 500). If you want to see the stats go look at it for yourself. Now he didn't have a max hit like a mac truck ... it was less than 4k with capped shielding ... but his dps was fierce. Check out the delay between swings and then factor in the fact that he flurried.

Tank A
DI 20 hit: 0.00%
DI 15+ hit: 5.61%
DI 10+ hit: 15.72%
DI 80 round (all DI20): Statistically impossible to calculate
DI 60+ round (all DI 15+): 0.00098% (~1 in 100,000 swings)
DI 40+ round (all DI 10+): 0.061% (~1 in 2000 swings)

Tank b
DI 20 hit: 2.19%
DI 15+ hit: 14.91%
DI 10+ hit: 24.97%
DI 80 round (all DI20): 0.000023% (~1 in 1,000,000 swings)
DI 60+ round (all DI15+): 0.0494% (~1 in 2000 swings)
DI 40+ round (all DI10+): 0.389% (~ 1 in 200)

Anyways long story short, rounds of nothing but max hits are about as common as a lightning strike, so even using them in the argument "I can soak a max or closer to the max hit than you" is a poor argument. What tanks usually die from is some bad combination of DI30-DI60 rounds that are not well spaced. This becomes even more real when you consider not the mobs that hit for 8k+ slowly (you can count on heals btw most rounds), but rather consider the mobs that hit for less, but faster, with more accuracy, and flurrying.

On meaningful content, an extra 2000hp lets you absorb maybe an extra 8 total DIs. I'd rather be 50x less likely to take that DI60+ round and 10x less likely to take that DI 40+ round than have the ability to soak an extra 8 total DIs. Yes I still do get a DI60 round on occasion, but statistically speaking a DI60+ round on me would be a DI70+ round with less ac.

[Bedavir]

This needed to be saved.

Yakk:

Originally Posted by Yakk

Alice has lower AC:
100 95% of the time
200 5% of the time

Avg: 105 damage

Bob has higher AC:
100 99% of the time
200 1% of the time

Avg: 101 damage

Alice takes 4% more DPS than Bob. That is a pretty small margin, so naively you'd think it shouldn't have much impact on incoming damage.

Let's do the math for 3 hits in a row:

3 hits:
Alice gets hit for 600 damage 1 in 8,000 times
Alice gets hit for 500 damage 1 in 140 times
Alice gets hit for 400 damage 1 in 7.4 times
Alice gets hit for 300 damage 6 in 7 times

Bob gets hit for 600 damage 1 in 1 million times
Bob gets hit for 500 damage 1 in 3367 times
Bob gets hit for 400 damage 1 in 34 times
Bob gets hit for 300 damage 33.7 in 34.7 times

Notice the difference. This difference is magnified the more hits-in-a-row we are talking about.[1]



If a 500 damage spike kills you, the Bob will die 24 times less often than Alice. From a 4% drop in average DPS.

Now, in the real game, this isn't about DI 80. It is about the chances of a high DI round. And a tank with decent AC already has a near-zero chance of a DI 20 single-round.

The additional AC starves the rest of the DI 11 to 20 rounds. They are already a small percentage of the damage you are taking, so if they are reduced further they don't show up in the average damage taken that strongly.



Shield AC, given a warrior 45% return-after-softcap and shield-block AA, is pretty huge.

For a quick rule-of-thumb, for a warrior add twice your shield's raw AC to your displayed AC to figure out the "real displayed AC effect" of that shield[2]. So a 3357 AC tank with a 130 AC shield has the mitigation of a 3617 AC tank. Plus you get shield-block, which is another chunk of damage avoidence.

Footnotes:
[1] [A^m](A+B)^n = (n choose m)A^m*B^(n-m)

where [A^m] means "select the A^m term from the following polynomial" (stolen from combinatorics notation).

Noting that both Alice and Bob's chance of a 100 point hit is pretty close to 1, we can generate a relative chance of being hit for various values on a 4 hit round without actually doing the "choose" calculations.

Here is the full calculation for a 4 hit round with Alice and Bob:

800: Bob is about 625 times less likely to be hit for 800 (1 in 100 million vs 1 in 160,000).
700: Bob is about 120 times less likely to be hit for 700 (1 in 250,000 vs 1 in 2000).
600: Bob is about 23 times less likely to be hit for 600 (0.06% vs 1.4%).
500: Bob is about 4.4 times less likely to be hit for 500 (4% vs 17%).
400: Bob is 18% more likely to be hit for 400 (96% vs 81%).

Note that Bob is has a 0.07% or less chance of being hit for 600 or more.
Alice, meanwhile, has a 1.4%+ chance of being hit for 600 or more.

One can generalize this effect to a more smooth situation: the variance-damping effects of high-AC can have an impact far beyond what the average-reducing effects of high-AC would indicate even in a more realistic case.

[2]
Disp AC =~ 1.6 * Raw Worn AC
Real Shield AC =~ 1/.45 * Real Non Shield AC

So shield AC is "really" worth about 2.22... times as much as a non-shield. Of that, 1.0 is already factored into your displayed AC, leaving 1.22... times your shield AC unaccounted for on your displayed AC.

1.22... times your shield AC, times the display factor of about 1.6, is about 2.0.

So take your displayed AC with your shield. Add in twice your shield's AC. That is how well you mitigate damage, with a shield, not counting shield block, as a Warrior.

Edit: named the warriors, extended the "how to work out the chance of being hit for X" footnote

[Kudedor]

In reference to this:

Thanks for posting in the parse thread, Aanelar! That is some very interesting stuff. I recall Lluiane or someone positing that NPC attack was very low. That kind of swing in the DI distribution would seem to support the notion. I can't wait to see further tests along this line.

[Bedavir]

Great work indeed Aanelar. What would be an interesting counter-experiment would be to then find out how much AC you had to pile back on to that warrior to restore you to the same base level of mitigation that you started. Ie how much ac does it take to reverse the ~100attack and ~120 strength. Maybe then we can begin to understand the practical value of attack/str debuffs and even begin to some up with closer estimations of mob attack.

[Lluianae]

Thank you for posting that

It's nice to see the reverse of ATK reduction in action, if you're ok with it could I forward the pictures to some druids? Also could you possibly say what the warrior's Raw AC (total gear AC) is?

Whilst we aren't entirely sure what +STR does in terms of relative numerical ATK gained for NPCs (could it be the same as for players?), we do know that it's substantial in tandem with +ATK. This pretty much reinforces my view that pets may just be the key, after all, they are the controllable mobs.


** edit by Yoda :
shame on you Lluianae for posting this in the wrong thread and forcing me to move it to its correct place.
btw, I agree with what you said to Aanelar.
**

[Lluianae]

Sorry Yoda

[Yakk]

That is a good point.

We can control pets, and they seem to act allot like mobs, combat-wise.

We can boost their ATK, we can boost their STR, we can boost their STR without boosting their ATK. We can figure out if STR has a different effect than just raw ATK (it probably does, given that SOE mobs store STR and ATK seperately (some random developer mentioned it)).

[Lluianae]

We can also debuff them too, although granted, harder to do since debuffs dont last anywhere near as long as buffs do. But if that was do be done, Arenas would be the ideal parsing place (as long as no griefers come).

I was thinking...

Don't charmed mobs retain the STR and ATK of their former selves, they just hit for 'significantly' less?

Charming mobs could also be an option too, perhaps. Although alot more risky too :/ Dire Charm wru

Another thing - I was talking to a guildy mage about pets, asking about when mage pets get summoned which foci = summoned with Phantom Plate. Pets use Real AC just like mobs apparently, so maybe some Pet vs Pet parses with varying ATK/AC could be done. That said I have no idea what pet mitigation defensives do formulaeicly :/