I really like the armory import feature of the new spreadsheet! I was wondering if the internet feature it could be (if it is possible) expanded to look up the stats of items off of the wowarmory/item-info, or any other site like wowhead. Then the item database of the spreadsheet wouldn't have to be updated for new items (or old items not currently in the database). It would help people like me that just hit 80, and need to manually add all the green/blue quest items I'm using until I get into raiding.

Going a step farther, even an option of including PTR items for quickly assessing how the new round of items affect the power of warrior.

Just trying to think of ways to make this DPS spreadsheet even more amazing.

You can either leave it exactly how it is (i.e boss armor value) and as sunder/FF reduce it to approximately the same value that they will reduce it to post patch you only need to change the armor pen rating. As its 25% more effective you just multiply the rating required for 1% cell by 0.8, and then, thats all if the effects are multiplicative. If you want to simulate them being additive then you change the boss armor to 10643 then remove the sunder armor and faerie fire debuffs. Then in the phoney stats column add 25*ArP rating required for 1% cell, along with multiplying this cell by 0.8.

I think I just put in Rallik's formula with the ArP changes. Here's what I did:

AL61 (25% more ArP from rating): =4.69512177*AK54*0.8
C25 (new boss armor): 10643
C57 (new debuffs): =1-(1-4*MAX(U30*V30/100,U32*V32/100,U34*V34/100)/100)*(1-5*MAX(P35,P37,P39)/100)
C58 (ArP from non-procs, multiplied by .81): =(((Q136+C51+IF(A30="Food: ArP",1,0)*C30)/AL61+3*IF(AI24=1,1,IF(AI25=1,1,0))*AL14)/100)*0.81
C59 (ArP when Incisor procs, multiplied by .81): =(((Q136+291+0+C51+IF(A30="Food: ArP",1,0)*C30)/AL61+3*IF(AI24=1,1,IF(AI25=1,1,0))*AL14)/100)*0.81
C60 (ArP when Grim Toll procs, multiplied by .81): =(((Q136+612+0+C51+IF(A30="Food: ArP",1,0)*C30)/AL61+3*IF(AI24=1,1,IF(AI25=1,1,0))*AL14)/100)*0.81
C61 (ArP when both trinkets proc, multiplied by .81): =(((Q136+612+291+0)/AL61+3*IF(AI24=1,1,IF(AI25=1,1,0))*AL14)/100)*0.81
C63 (Base armor multiplier from Rallik's formula*): =(1-(C62+C57)+((C25/(C56+C25))*C62*C57))
C64 (final armor mitigation): =IF(C56/(C56+C25*C63)>1,1,C56/(C56+C25*C63))

Gives an SEP for ArP of ~0.89

*

My question is why would you do that using the found Bugged ArP formula when you know that it will be fixed in live and have to be re-tested and reverified before the final patch goes in? Why is everyone so crazy about it right now? If you want to know to pick up ArP gear or not, just estimate ArP near 1.1 of strength, don't go over 50% passive ArP if you have Grim Toll, and I assure you will be fine for when the real patch comes. The only items worth possibly persuing out of normal current BIS to make a pre-Ulduar BIS set would be Grim Toll, Maybe Valorous Helmet maybe Double Strikes if you have expertise cap (you'd need valorous legs and Obsidian for expertise), Collision Ring, and that's about it.

Players will have access to the updated sheet and enough time to play with it before you are able to collect any Ulduar gear I assure you. Just be patient because I don't want to give you anything when the full correct mechanics are unknown.

Gives us something to do and because spreadsheets are fun. The only reason not to fuck around with it is if you'd be better off doing something else.

Obsidian Greathelm, Chestguard of the Recluse, Darkmoon Card: Greatness, Melancholy Sabatons, Favor of the Dragon Queen and Grim Tool look like it's going to last a very long time.
Also used in some optimal hit/exp setups with known Ulduar drops, some of which landsoul mentioned: Frosted Adroit, Ruthlessness, Legplates of Double Strikes, Shoulderguards of Opportunity.

Though, it can all be summed up pretty neatly with "Don't let anything rot". There's going to be lots of mixing and matching as hit and expertise swap slots.

I'm curious as to your early assessment of Grim Toll over, say, Mirror of Truth (assuming the other slot is used by Greatness or FotFF). Let's make a fair assumption that we are hovering around 200+ hit rating without gemming for it, does data exist to show that Grim Toll's proc puts the combined 619 ArP and 73 hit rating over 84 crit rating and 1000 ap?

I'd test this myself but I had long gotten rid of my own Grim Toll ages ago.

Using Grayson's sheet modifications, I show Grim Toll as very slightly better than MoT, within the margin of error.

hit and crit should be roughly equal assuming you can avoid going over hit cap. I traded my MoT in for Grim toll, and based on the proc uptimes I'd say its a good trade. MoT was rarely delivering a 15% uptime, usually lower - and sometimes really low (like 6%). With the grim toll I get 20-23% uptime very consistently. I'm arms though.

A little napkin math about the procs:

MoT: 1000 AP * .45 (2 ap = .9 sep) * .15 (theoretical uptime) = 67.5
Grim Toll: 612 ArP * .77 (rough sep from spreadsheet) * .20 (conservative uptime) = 94.248

and this is with 3.0 style mechanics. Grim Toll should get even better come 3.1, while MoT gets worse.

On my own experience with MoT and Grim Toll, I'd say Grim Toll is better if you have a decent amount of Crit already. +Crit is almost always better than +Hit, but Grim Toll procs a lot more than MoT. ATM I equip both while I wait for my Greatness card and all my Raid parses shows Grim Toll way ahead on proc rate.

Edit: Last Naxxramas raid: MoT 78 procs, Grim Toll 106. My other raid warrior: MoT 96 procs; Greatness 142 procs. Greatness Card is awesome now, and will be better after 3.1.

Using Rallik's data-fitting PTR model of:



where:
A = the half armor value at 80, 15232.5
B = base boss armor, 10643(the value I found from testing)
x = sum of arpen buffs(mace spec/battle stance/gear arpen)
y = product of arpen debuffs(sunder/ff) (1-.95*.8)

Removing the PTR bug (81% effectiveness of additive ArP such as arms stance, ArP, other effects) I get ArP as ~1.44 SEP... I seriously wonder if and how the PTR bug will be removed because it would feel a little awkward gemming/gearing completely ArP even with grim toll giving 100% penetration on proc 20% of the time.

Even if Arp SEP goes to 1+ doesnt this mean that we should still be gemming for it? It does look awkward but we would do the same thing in TBC if we had that option untill we hit the cap.

If the bug is "working as intended" then they should just declare that battlestance provides 8% arp and that arp buff is reverted.
It is just meaningless that they have a bug that "works as not-intended but it's fine".

During my arp tests i noticed the crit deflation difference aswell, but my sample size was always under 5k and I was geared so I didn't focus on that, but paperdoll crit was close than 4.8% every time.
I suppose this can be due to base warrior crit that has been raised by few % in patch 3.08 iirc, but if it is than we should see this even on live.

FileFront seems to be back.

Video Game Downloads on FileFront

Keynote on Arms model:

Based on ability effectiveness:

Mortal Strike every 5 unrestricted (because you dont want to SD first leaving you void of rage)
Overpower every 6 unrestricted
Execute every time it's up, times nochance to overwrite during an MS GCD, which is given by: (1-(1-0.09)*(1-0.09*1.5/{avg attack speed}))*1.5/5, times nochance to overwrite with white attack refreshing rend which is similar, times nochance to overwrite with white attack hitting overpower which is similar, times nochance to overwrite with white attack using slam which is also similar. This deals most of the time 30 rage damage, and other part of the time less than 30 rage damage.
Rend Ticking every 3 unrestricted
Slam every free 1.5s of global cooldown space, multiplied by a scalar A less than one representing amount of estimated collisions. Free global cooldown space is derived from one minus the total space used by other prioritized abilities.
Heroic strike is used every time the player gets over 100 rage.

All of these frequencies are then multiplied again by a scalar B less than one representing the fact that you want to fill every GCD to an extent.

The difficult time I am having is estimating scalar A and scalar B that properly values the human behavior when following the arms rotation. In laman's terms, as humans how often are we not slamming when we can due to prioritizing other skills which is A, and how often are we not hitting prioritized skills on time every time due to trying to fill all global cooldowns which is B.

That's the jist that will go in the spreadsheet and I just have those scalars A and B to work out. Bladestorm is yet undetermined. Since it's instant now, it does more damage than slams as long as you use it and cancel before other stuff comes up.

Is this just so that if you had say 50 rage and you used MS first it would leave you with 20 rage left over for an Execute, which would then leave 10 rage after that, where if you were to Execute you would only have 20 rage left and not be able to MS? What about in lower rage situations where you might have exactly 30 rage? You would have nothing left over after using MS, but would have at least 10 rage for an Overpower if you used Execute or even another Rend or lucky Execute. And if you're going to be stuck at that level of rage you would be "void of rage" until your next auto attack, Bloodrage or outside source anyway.

Given how overpower has huge leeway to use now, Id say - just assume it goes off every 6 seconds on average (which you did), but you can ignore its effect on executes (since you can shift it enough, to realistically, always prioritize execute over it).

Im also thinking if a different approach wouldnt actually yield better results. Since there is a pretty big level of uncertainity with the scalars, and since the user's reaction will tend to cloud the real dps way more then some simplifications in the model, i feel following model would be sufficient:

MS wont be used every 5 seconds, because of GCD issues. In most cases you are left with using MS every 5.5 sec (GCD of 2 skills+OP) + Lag factor, and once per ~minute there is a 6 sec delay on MS (due to OP cooldowns). The lag factor tends to realistically delay the rotation enough to never consider the OP cooldown shift. So overall it ends up being

MS,OP,SLAM,SLAM as a base, and you can replace SLAMs with rend and execute. Duration of the rotation is 5.5sec+lag factor

Rend wont collide with MS in this model, because with 21 sec duration you have 2 "moments" to refresh it always - just after it runs out, or 3 seconds later (due to 6 sec TfB cooldown, its NOT necessary to keep rend always up, and if it would collide with MS its better to delay it 2gcd (where it wont interfere for sure)).

Rend/MS collisions will affect Rend tick damage slightly (but not TFB procs), and collisions should happen ~ once every 4 rends, leaving the uptime of rend in 28/29*100% range.

Executes just replace Slam damage, with the chance calculation like you said, but ignoring the OP for the calculations.

It seems much simpler, and honestly - given how rotations will depend on lag and such, its about as "useful" to gauge dps. Both however will be very useful and accurate models for the main sheet purpose (choosing gear).

Probably its fine on execute, assuming we have 10 overpowers per min + ~33 other gcd before counting lag, it gives around 1.4 average gcd+lag. White swing counting using slam every 6 or so gcd, ends up being about 2.6 sec apart on average. That ends up being around 1 attack/second (i guess we should count rend there too which i think doesnt proc SD, but thats a rough math anyway). 1 attack/second with 9% chance, should result with an SD around 1 per 11 seconds. 1/12 is a bit less, but collisions do happen, and reaction/lag might be higher then 200ms. However given how its our highest damage move, and it happens around once vs 4 swings and 7 other gcd skills, id expect total damage to be around 8.5-9%. Still well see with full ulduar gear

For the same rage, Execute is already more DPS and DPR over Mortal Strike with a Betrayer.

The gap seems to remain around the same even with double the buffs. A new weapon won't be enough to put Mortal Strike ahead. This is also why I still don't understand why you would ever put Mortal Strike ahead of Sudden Death at lower levels of rage.

Well what happens is MS gets bloated with the high ArP, set bonus, sword spec, and 1000 top end on voldrethar. I took the stats down to dummy-unbuffed and got exe up to 8.7% representation. I haven't yet been able to use other gear setups (axe) yet because I haven't yet been able to use the model with full gear selection.

Edit: I found what I did wrong, I was using the wrong ArP multiplier, it's up to about 9.5% now.

Why would I put MS over Execute? I'm not 100% sure at the moment, but there's <19% chance the exe will overwrite while hitting the MS, and you can wait to use it again once you get the rage after bottoming out over MS.

You are absolutely right it does hit harder, but MS's DPS > Execute all the time. It is on a very short cooldown and it really sucks when you push it off

So it's worse to delay an ability that is on a set cooldown by 1.5 seconds instead of favoring the chance of proccing one that is completely RNG based as well as more DPS and DPR? Why does it matter how often you shove Mortal Strike back if you're still going to get more DPS? I've tried both methods of priority and I'm just not seeing Mortal Strike ever being the ability to hit first. DPS has so far always been lower when doing that.

Well it basically comes down to whats better, (1.5+lag+react)/(5+lag+react) of an MS, or [1-(.91)*(1-.09*(1.5+lag+react)/attackspeed)] of an execute, considering crits, deep wounds, and everything.
MS:1.8/5.3*7000=2377.4
exec:.149*7729=1151.6

This is the only thing I really have to go by because no one shows any results besides words that don't mean anything. "my dps is always higher when I use exe of ms" well what the hell am I supposed to do with that? Show me a recount screen or something. Also, from my experience, you have to do a shit load of parses to get non-random results. Doing 3.5k on a dummy over 6 minutes one way and 3.8k on a dummy over 6 minutes another way doesn't really say much.