Warrior DPS Calculation Spreadsheet - Page 63

Claertes · 05/22/09, 12:42 PM

Note: Executioner is not better than Berserking for both Arms and Fury, at least not yet. For Executioner to be better than Berserking, SEP for ArP needs to be greater than 1.2627 for Fury, and 1.4568 for Arms.

Not a big thing (but now I'm a little bit confused).

Doesnt Executioners uptime depends on hits per second, which depends (partic.) on hasterating (and choosen buffs a.s.o.), which can be very very different?
I'm just confused about so much SEvalue-accuracy in such a basic statement

.

RPZip · 05/22/09, 1:29 PM

I found a few minor bugs in the newest spreadsheet (and probably earlier ones, but I first encountered them with this one).

When I imported from Armory, I get two errors; the first one is about my boot enchant, because the spreadsheet has it in a different format than it is on Armory (Minor Speed and +6 Agility versus Minor Speed +6 Agility) and that Hoperender is not a supported mainhand. It's spelled with a space on the spreadsheet, but it actually doesn't have one in the item name. Should be easily replicated if you try to import Nomepunter - Illidan.

I ran a quick thread search to see if this had been mentioned earlier and didn't see it come up.

Goatsey · 05/22/09, 2:04 PM

Originally Posted by Phoenix

Got something funny going on with the spreadsheet. All my unbuffed stats are bang on aside from my crit, AP and health. All my gems, race etc are correct (I checked bazillion times because I thought I was going mad).

My AP is 72 too low, my crit is 0.13% too high and my healths way out - im at 22951 in-game but the sheet says 27485.

Any ideas ?

The unbuffed stats don't include strength of arms for the ap calculation, they are included in calculations but not for the UB ap. No idea if you changed your race to your current one which could explain the crit. Not really sure about the massive hp compared to actually in game though.

Paradosi · 05/22/09, 2:22 PM

Not sure how much this impacts the spreadsheet but:

World of Warcraft - English (NA) Forums -> Upcoming Jewelcrafting Change

Ferboten · 05/22/09, 2:25 PM

Basically the 27 str/arp/exp gems become red gems, with a net effect of us losing out on 3 socket bonuses

Blackburn · 05/22/09, 2:30 PM

Originally Posted by Ferboten

Basically the 27 str/arp/exp gems become red gems, with a net effect of us losing out on 3 socket bonuses

Not quite, we're also losing the metagem requirements, so we're now forced to socket at least 2 blue hybrid gems. It's a pretty terrible change when considering the extra perks that other professions get besides just straight up stat bonuses, especially once epic gems become available and the JC gems don't provide the same stat bonus as they do now.

Zasm · 05/22/09, 3:20 PM

Hi I was messing around with the sheet for about an hour last night comparing my current gear and upgrades. I was getting some weird results with the SEP values of ArP mostly around .85 range, last run of the spreadsheet about two weeks ago clearly showed that ArP>STR with a SEP of around 1.2 for me.

I have grim toll, betrayer, 4/5 8.5 so far and mostly nice gear with it all currently gemmed for ArP and sitting at 46.69% ArP but with the recent ArP cap changes its showing that just strait up gemming for STR is now the best idea again... With my current ArP values I'm pretty sure I'm not going over the cap with only 1 ArP oriented trinket. I don't exactly understand why the SEP value dropped so much considering i wasn't even over the ArP cap to my understanding.

So back to gemming all strength?

sigh

Stink · 05/22/09, 3:45 PM

If you try swapping out the Grim Toll for a non ArP trinket you will see your ArP SEP value return to what you expect it to be. The reason for the devalued ArP values with an ArP trinket equipped is because you are effectively wasting any ArP you have over the cap while the trinket is active. The ideal scenario would be to be at 100% passive ArP and not use a proc-based trinket, but until that situation is realistic we will have to find some middle ground between the two. Making educated gear decisions is going to rely even more heavily on testing different sets and looking at the expected DPS returns rather than just assigning SEP values to gear.

Aieeja · 05/22/09, 5:19 PM

mh, maybe i'm wrong or understanding it wrong, but when ArP cap now is at 100% tooltip-display, then 1232 ArP rating would be enough for 100%, but (with Mjolnir equiped) in the spreadsheet the SEP for ArP is showing >1 only when i have less then 45,88% ArP (35,88% equip + 10% battlestance).
Which means that i would be capped with Mjolnir proc when i have 1109 ArP rating (90% ArP + 10% battlestance).
Since the armor penetration from battlestance isn't shown in the tooltip, i'm wondering if battlestance counts into the cap or acts like sunder and FF ? :>

Machinator · 05/22/09, 6:28 PM

Brainstorming on avoidance streaks, feel free to point out errors. Apologies ahead of time for clarity or formatting.

1. From my post a couple pages ago: "For avoidance streaks, the rage income ratio of each hand is used for determining the chance of a streak, but not its recovery time."
To fix this you could
A. increase the recovery time to wait for the OH for the time you are using HS
1/(1/MH speed + 1/OH Speed), shouldn't it be 1/ ( 1/MH speed * MH ratio + 1/OH speed * OH ratio)
B. add the chance of a HS into the chance of the streak, since you are not gaining rage even if it lands (I think?)

2. Why isnt there a single avoidance streak slip product? One hand cannot sustain the rotation, especially since the avg swing speed is longer than a GCD. For me that would be 15.6% chance and .517s recovery for ~8% slip. Would this be solved by anything in my first point?

3. Completely remodeling avoidance: Replace slip with the chance of getting enough rage for abilities within the last GCD, plus an average of unused rage from other GCDs.

4. Drawing a blank on my statistics, but I think you can find the probability of having enough rage to sustain the rotation(16 seconds probably). Find the chance that: (MH swing/rotation * rage/MH swing + OH swg/rot * rage/OH swg)*(1-avoidance) >= rage needed/rotation. If you can find that you can also find the chance of getting >100 rage. I am unsure if HS frequency could be found the same way as HS/rotation isn't a discrete number.

PS. For the the 3.0 of the sheet, I think it would be worthwhile to expand a lot of formulas out from single cells.

Edit: I can confirm that the rage formula on live does not match the formulas given to us by Kalgan for f = 3.5 or 2.5. Rage is 70-80% of expected.

Origence · 05/22/09, 8:04 PM

I just noticed in the sheet [Girdle of Razuvious] has 88 stamina instead of 85
And [Clockwork Legplates] have 2064 armor instead of 2054.

I checked why health and armor values didnt match and found out why

soullink · 05/23/09, 8:30 AM

I still think the Heroic frequency is too high by around 8% but the sheet is back to being "closer to reality" figures.

What people tend to forget is that the sheet doesn't model execute at all.
If you use execute under 20% boss health instead of heroic strike, hs percentage compared to white damage percentage drops significantly.

The current version of the spreadsheet tells me the following percentages:
17.08% of damage is white hit
28.54% of damage is heroic strike

If you keep in mind execute and that the spreadsheet models perfect play, those numbers seem very accurate to me.

TheGhost305 · 05/23/09, 9:43 AM

I´m really confused about the new Spreadsheet i downloaded.In the Old Spread it was ArP>Str when you got enough passive ArP..not in the new one it is Str>ArP..and there are just Naxx25/Malygos25 Items used inclusive Betrayer of Humanity?!Did i miss something?Or why the change in 2 days..I´m really confused!

Conquistador · 05/23/09, 11:01 AM

I love the fact that you assembled the BiS list Landsoul, I do appreciate it. However I noticed that with all your suggested BiS items, you'd be about 100 expertise short of the cap. In your opinion, is that deficiency meant to be made up by gems? That seems like a lot of sockets to use on expertise gems.

Erfinda · 05/23/09, 1:26 PM

Originally Posted by Machinator

...
3. Completely remodeling avoidance: Replace slip with the chance of getting enough rage for abilities within the last GCD, plus an average of unused rage from other GCDs.

4. Drawing a blank on my statistics, but I think you can find the probability of having enough rage to sustain the rotation(16 seconds probably). Find the chance that: (MH swing/rotation * rage/MH swing + OH swg/rot * rage/OH swg)*(1-avoidance) >= rage needed/rotation. If you can find that you can also find the chance of getting >100 rage. I am unsure if HS frequency could be found the same way as HS/rotation isn't a discrete number.
...

Just gonna comment on 3 and 4 on this as a whole topic.

The problem with finding that probability is that we still need to know the distribution of rage generation in the trailing x seconds. For flurry uptime, flurry procs are a Poisson process that we model as an exponential distribution. Flurry procs are independent, continuous, and occur on an average rate over a period of time. Flurry uptime is just modeled by the CDF function of an exponential distribution. We can't apply the same stochastic process to rage generation because it does not consist of independent events and is definitely not a Poisson process. It's definitely not a normal distribution either unless we sample a lot and apply CLT to it, which is just as problematic as the current system.