Tank Points - By Undertanker

[Undertanker 88]

Added Blessed Qiraji War Axe

[Omgdontdie 1]

SQL file has QS at 5.833% proc rate. 75% up time seems a bit optimistic. 

[Undertanker 88]

I did it with a 6.66% proc rate derived from 2 PPM.

30 autos, 40 globals (assuming Sunder Armor and shield slam will get fixed), gives 70 possible situations of proc potential each minute.

70 * .066 = 4.62 procs a minute on average.

4.62 * 10 (duration of proc) = 46.2 seconds, assuming no overlap.

46.2 seconds of 60 seconds is 77% of a minute.

I rounded down to 75%.

 

[Omgdontdie 1]

The SQL files have it at 1.75 ppm.

Shield slam does not proc main hand here.

That gives approx 3 procs per minute (if you are hit capped).

Assuming no overlap, that gives us 50% uptime. Realistically, its not uncommon that the procs overlap. You'd have to use a poisson distribution to estimate a realistic uptime if you account for overlaping.

An expected value of 50% uptime is your best case scenario here. 77% is beyond inaccurate. 

[Undertanker 88]

1.75 PPM gives it a 40 second per min average uptime, 66.66% not 50%.

I am not doing calculations based off server bugs.  Once said bugs are fixed they will display the accurate potential of said weapons.  So I am still pulling from 70 possible attacks, not 61 attacks (skipping shield slam).

Overlapping occurs less with a lower proc rate, I am not going full Sir Newton on probabilities, as it works on both sides, thinks can proc often and you get overlaps, and things could not proc, so we take average and move on with life.

[Undertanker 88]

Updated Quel from 75% uptime to 66% uptime.  Moved rank of weapon as appropriate. 

[Omgdontdie 1]

Shield slam does not trigger main hands here and procs can overlap.

Lets assume you are using an ability for every global, since you should be in a raid setting. Every minute, you will do about 10 shield slams, the rest will be revenge or sunder. Every minute you have 30 special attacks and 30 auto attacks that can proc. At 1.75 ppm, its clearly follows that the expected number of procs per minute is 3.5. If we assume that the proc NEVER overlap, that is still 35 seconds of uptime. If we consider that there is a 5%(I think?) base chance to be dodged and parried, this number becomes increasingly closer to 30. If you are not hit capped (many tanks are not) the value is diminished a bit more. Sure there are other factors, like parry haste, but the fine details aside, lets just go with 35secs.

Keep in mind that it is pretty common for procs to overlap (again, poisson distribution), so your expected uptime will be closer to 50% at best. You would have to be incredibly lucky to achieve 66% uptime.

 

 

[Undertanker 88]

2 hours ago, Omgdontdie said:

 

The SQL files have it at 1.75 ppm.

Shield slam does not proc main hand here.

 

 

1 hour ago, Undertanker said:

I am not doing calculations based off server bugs.  Once said bugs are fixed they will display the accurate potential of said weapons.

 

7 minutes ago, Omgdontdie said:

 

Shield slam does not trigger main hands here and procs can overlap.

 

Thanks, because 2 hours ago is hard to remember.

1 hour ago, Undertanker said:

Overlapping occurs less with a lower proc rate, I am not going full Sir Newton on probabilities, as it works on both sides, thinks can proc often and you get overlaps, and things could not proc, so we take average and move on with life.

Average: a number expressing the central or typical value in a set of data, in particular the mode, median, or (most commonly) the mean, which is calculated by dividing the sum of the values in the set by their number.

Can you stop following me every post I make with the same 2 statements?

Ironically with your hard-on for Ironfoe you don't take into account dodge/miss/parry but want me to do it here?

[Omgdontdie 1]

I'm just trying to help you to get more accurate numbers and correct false information. Calculating buff uptime isn't as simple as taking a mean. 

In regards to dodge/miss/parry, let us revisit my comment...

Quote

Sure there are other factors, like parry haste, but the fine details aside, lets just go with 35secs

Here is some entry level statistics relevant to proc uptime: https://en.wikipedia.org/wiki/Poisson_distribution

[Undertanker 88]

24 minutes ago, Undertanker said:

I am not going full Sir Newton on probabilities

 

[Vilius 1]

3 hours ago, Omgdontdie said:

Assuming no overlap, that gives us 50% uptime. Realistically, its not uncommon that the procs overlap. You'd have to use a poisson distribution to estimate a realistic uptime if you account for overlaping.

An expected value of 50% uptime is your best case scenario here. 77% is beyond inaccurate. 

I agree with you in principle about Poisson being a better method than what Undertanker is doing, although I prefer to just do Monte Carlo sims since it's easy to add in non-uniform events (like no shield slam procs) and other irregularities.

So I did just that for this one case of Quel'serrar proc uptime.  For now I didn't model dodge/parry/miss.  I assumed every 4th gcd was Shield Slam, and did runs for both SS can and cannot proc.  Other gcd's were assumed to be capable of proc.  In each case I did 1 million runs of 3 minute fights.  These are the results:

2 PPM w/ SS procs: 57% uptime

2 PPM w/o SS procs: 50% uptime

1.75 PPM w/ SS procs: 49% uptime

1.75 PPM w/o SS procs: 45% uptime

Undertanker I understand your desire to simplify, but when you oversimplify you get very inaccurate results.

[Omgdontdie 1]

Vilius, you did the math that I was too lazy to do. Those numbers look very realistic. Currently, QS is a bit under tuned because sunder isn;t proccing weapons, but it should be getting a sizable buff after the next dev update (assuming the fix works).

[Undertanker 88]

I'll update to 50% uptime in a bit.  @ work atm.

[Undertanker 88]

Updated Quel with 50% uptime and adjusted ranking for it.

This puts it barely behind Sickle of Unyielding Strength for an Orc.   Seems legit.

[Cetus 0]

Got a question about skill TP vs. +hit TP.

You've put skill TP on 2, and +hit TP on 18.

As I understand it, getting 305 weapon skill results in a different formula to calculate base miss chance like this:

At 300 weapon skill (difference between def skill of boss and weap skill of player is >10:

Base chance to miss (%) = 7 + (defense skill – weapon skill – 10) * 0.4
= 7 + (315 - 300 - 10) * 0.4
= 7 + 5 * 0.4
= 9%

At 305 weapon skill (difference is equal (or less) than 10):

Base chance to miss (%) = 5 + (defense skill – weapon skill) * 0.1
= 5 + (315 - 305) * 0.1
= 5 + 10 * 0.1
= 6%

In other words: the first 5 weap skill gives a 3% reduction in base chance to miss. Should the racials therefore not be worth at least 3*18 = 54 TP?

Thanks for your reply in advance.

[Undertanker 88]

The benefit for the based reduction in base chance to miss is more so for duel wield, you'll still need the +Hit for your abilities such as shield slam, revenge, sunder which are critical in your threat holding.

The importance of the +5 Skill is the 8% average rage-gain from white hits and the small increased damage for damage to threat conversions.

[Cetus 0]

I was under the impression that the 3% reduction in base miss chance also affected yellow hits. The only other thread where I saw this discussion was in AQ bis + weap skill by Omakaroni. There too people did not seem to agree on the way weap skill affected yellows, if at all.

Of course if weap skill is only on whites, that diminishes the use for me considerably as a tank (I was stoked a/f about gettin 3 hit effectively).

I would love the person that could conclude that debate once and for all.

[Ragward 2]

as far as I am aware +weapon skill was only to affect your glancing damage and not your "miss/hit chance" i think there must have been some confusion with some threads where they tried to convert + weapon skill into hit rating

[Kolmar 3]

Here's the conclusion to the debate, there's a TL;DR at the end.

The hit cap is partly based on weapon skill difference. The weapon skill of an NPC is their level * 5, so a level 60 mob has a weapon skill of 300. 61 = 305, 62 = 310, 63/skull mobs = 315, vanilla mobs do not go higher than 63 so this is as high as NPC weapon skill goes. 

The miss chance is 5% by default, and the formulas used to determine the hit cap after weapon skill calculations are as follows:

If the difference between your weapon skill and a mobs weapon skill is 10 or less, then this formula is used to determine the hit cap:

Skill difference * 0.1 + Base Miss Chance (5%) = Hit Cap

So lets say for example that you are a level 60 character fighting against a level 62 mob. You have 300 weapon skill, he has 310 weapon skill, the difference is 10 weapon skill.

10 weapon skill difference * 0.1 + 5% base miss chance = hit cap

10 * 0.1 + 5 = 6% hit cap

 

If you fight against a mob that has a weapon skill difference of 11 or more, then the formula changes a bit.

(Weapon skill difference - 10)*0.4 + 2 + Base miss chance (5%) = Hit cap

So lets say you are a level 60 character fighting a level 63 mob. You have 300 weapon skill and he has 315 weapon skill. The weapon skill difference is 15 which is more than 10, so the second formula kicks in. 

(15-10)*0.4 = 2

2 + 2 + 5 (base miss chance) = 9% hit cap

 

So what happens if you fight a level 63 mob with 305 weapon skill? Well, the difference in weapon skill becomes 10 (305 vs 315), that means that the first formula is used instead of the second one. That means that against a raid boss, if you have 305 weapon skill then you will only need 6% hit to reach the cap. This applies to your special abilities that require a meele weapon too, Sunder/Revenge/Heroic strike/etc all cap at 6% with 305 weapon skill against raid bosses, the only exception is Shield Slam because the ability does not require a weapon to use, only a shield, thus it's treated as always being at 300 weapon skill, and so you will always need 9% hit to cap Shield Slam. Whether it's worth it to get 3 additional hit solely to cap Shield Slam is up to you guys, I'm just explaining how the stuff works. Also all of this is assuming you are using a 1-hander & shield or 2-hander, we're not taking the dual-wield white miss penalty into account. If you don't trust this then you can test it yourself by going 6% hit with 305 weapon skill on ZG/MC bosses and record your misses with an addon, the only attacks I have ever missed are Shield Slams (due to it needing 9%), all the other attacks have never missed. 

 

 

TL;DR:

300 weapon skill = 9% hit cap for white/yellow attacks against bosses

305 weapon skill = 6% hit cap for white and some yellow attacks against bosses, yellow attacks that do not require a meele weapon to use like Shield Slam still need 9% hit. Whether it's worth it to get 3 additional hit solely to cap Shield Slam is up to you guys, I'm just explaining how the stuff works. 

[Omgdontdie 1]

Ive noticed that weapon skill doesnt affect some yellows in the same way. For example, ive had over 9% hit after weapon skill and missed an execute. For the same reason they dont proc main hand weapon hits, id expect that shield slam, sunder, bloodthirst, and execute are not affected by weapon skill. Only evidence I have is for execute tho, so can't really prove the other ones w/o more testing.

Also, in response to Ragward - according to the Ely source code, you actually gain a lot of +hit from 300->305 skill against bosses here. 305 weapon skill on this server is a big deal, intentional or not. It also slightly reduces chances of your target block/dodge/parrying your attacks, which is increasingly relevant for tanks since you always attack from the front.

[CononC 0]

Heya!

Crap, I wrote a post to explain the weapon skill stuff, but then managed to forget my password and get my account temp disabled. And of course someone else beat me to the punch. But I'm still gonna post, with references! Even made a new account to post as fast as possible, how desperate is that?! Time to try to get some love then, get ready Cetus!  I posted earlier about the weapon skill calculations before the source code was known. The calc's were in the right direction, but a bit off - looking at the source code available now, the calculations can be simplified into the following:

References to the source code - if you don't believe me, browse the code until I (or hopefully you :P) stand corrected.

#_________________________________________________________________________________________________________
From the function at https://github.com/elysium-project/server/blob/master/src/game/Objects/Unit.cpp#L2724
we can see, that the weapon skill applies to melee spells too.

the relevant lines are (the first if-element deals with PvP)

    else if (skillDiff < -10)
        hitChance = 93.0f + (skillDiff + 10) * 0.4f;        // 7% base chance to miss for big skill diff (%6 in 3.x)
    else
    hitChance = 95.0f + skillDiff * 0.1f;

Here it is, unambiguously put. PLEASE NOTE: This means that the weapon skill also affects yellow hits!

#_________________________________________________________________________________________________________
From the function at https://github.com/elysium-project/server/blob/master/src/game/Objects/Unit.cpp#L3056
we can see the "normal" white hit calculation for melee hits.

the relevant lines here (again, first if-element deals with PvP)

    else if (skillDiff < -10)
        missChance -= (skillDiff + 10) * 0.4f - 2.0f;       // 7% base chance to miss for big skill diff (%6 in 3.x)
    else
    missChance -= skillDiff * 0.1f;
#_________________________________________________________________________________________________________

Surprisingly, the both functions give out the same number when equal weapon skills and mob defense numbers are applied.

Thus, the yellow swings and white swings have the same miss % on this server!

From this crappy table you can see what happens when the weapon skill goes up. Key points:

The base miss chance is 9% when skill = 300 and enemy defense = 315.
The base miss chance is 7.4% when skill = 304 and enemy defense = 315.
The base miss chance goes down by a whopping 1.4% between 304 and 305.
The base miss chance is 6% when skill = 305 and enemy defense = 315.
The base miss chance is 5% when skill = 315 and enemy defense = 315.

wskill    def        diff        miss %
290    315        -25        13
291    315        -24        12.6
292    315        -23        12.2
293    315        -22        11.8
294    315        -21        11.4
295    315        -20        11
296    315        -19        10.6
297    315        -18        10.2
298    315        -17        9.8
299    315        -16        9.4
300    315        -15        9
301    315        -14        8.6
302    315        -13        8.2
303    315        -12        7.8
304    315        -11        7.4
305    315        -10        6
306    315        -9        5.9
307    315        -8        5.8
308    315        -7        5.7
309    315        -6        5.6
310    315        -5        5.5
311    315        -4        5.4
312    315        -3        5.3
313    315        -2        5.2
314    315        -1        5.1
315    315        0        5

And getting back to my earlier post https://forum.elysium-project.org/topic/32234-weapon-skill-from-300-to-323-and-the-implications-therein-for-warriors-and-perhaps-others-too/ - the Edgemaster's don't give a direct +hit (reduction to miss) of 2.48%. They actually give +hit (reduction to miss) of 3.2%.. I stand also corrected... but it feels good, the Edgemasters are even more overpowered than I originally thought :D

I'll make a nice spreadsheet later! Just had to post this while shitting my pants, so I get to say the last word :P!

best regards
Cononchet, tauren warrior of Anathema

[Undertanker 88]

Yet many people are reporting (rogues and warriors) that with > 9% hit on gear yet with +weapon skill they are recording missing on yellow hits.   So I am hesitant to give +skill a further boost in performance for a tank perspective, as currently I only have it's value derived from the glancing blow pen reduction and the reduced chance to be dodged/parried/blocked/missed.

 

[Knife 2]

3 hours ago, Undertanker said:

Yet many people are reporting (rogues and warriors) that with > 9% hit on gear yet with +weapon skill they are recording missing on yellow hits.   So I am hesitant to give +skill a further boost in performance for a tank perspective, as currently I only have it's value derived from the glancing blow pen reduction and the reduced chance to be dodged/parried/blocked/missed.

 

"many people" = ~0 guys with a name and a screenshot of gear / combat log?

50%+ of the population believes hit on gear counters evasion(dodge), and "many people say" should be comparable to a copypaste of source code?

go and TEST it it's really really not hard to autohit anachronos for 5 min with varying weapon skill, some people actually did this...

[Undertanker 88]

31 minutes ago, Knife said:

"many people" = ~0 guys with a name and a screenshot of gear / combat log?

[CononC 0]

EDIT: was proven wrong on this, phew! Leaving this here for archival purposes :d

Heya!

Heh hee, you know what? I had typo'd my spreadsheet yesterday and it produced the correct numbers, so I overlooked this in my last post.

But.

If we look at the source code
https://github.com/elysium-project/server/blob/master/src/game/Objects/Unit.cpp#L3056

the equation there for normal white hits is

missChance -= (skillDiff + 10) * 0.4f - 2.0f;

when skillDiff < -10.

And if we calculate with this..

skill    def        skilldiff    normal miss
290    315        -25        9
291    315        -24        8,6
292    315        -23        8,2
293    315        -22        7,8
294    315        -21        7,4
295    315        -20        7
296    315        -19        6,6
297    315        -18        6,2
298    315        -17        5,8
299    315        -16        5,4
300    315        -15        5
301    315        -14        4,6
302    315        -13        4,2
303    315        -12        3,8
304    315        -11        3,4
305    315        -10        6
306    315        -9        5,9
307    315        -8        5,8
308    315        -7        5,7
309    315        -6        5,6
310    315        -5        5,5
311    315        -4        5,4
312    315        -3        5,3
313    315        -2        5,2
314    315        -1        5,1
315    315        0        5

so the equation is majorly bugged :DD and it should read

missChance -= (skillDiff + 10) * 0.4f + 2.0f;

instead. And this + was the typo I made, so the numbers looked all good to me.. Sorry to rain on everyone's parade :D please doublecheck this for me, I couldn't sleep because something was bothering me... maybe it was this 8)

best regards
Cononchet, tauren warrior of Anathema