Deathwing

I thought I'd create a thread for this since I didn't get any hugely negative feedback on it. Plus, I'm sure the rogues don't want me hijacking their threads anymore.

I fixed the DPS column in the 2H tab. This spreadsheet is designed for sustained dps versus a level 63 mob, so that column really isn't that pertainent. Look to the 80%/20% for the dps numbers you want. I have a few questions of my own:

I looked at Chalon's spreadsheet and I liked the pulldown menu/lookup table he implemented. Anyone got a quick guide I could read so I can implement this?

Additionally, Chalon modeled Crusader at 37 str MH and 24.5 str offhand. How did he come to those numbers?

Any and all feedback is welcomed.

EDIT: new file posted, see post #705 for explanation:

http://www.rcsolid.com/deathwing/DPSWarriorv0.485.xls

 

Thorb

Hmm, checking it atm, I'm kinda trying to do the same thing atm but it seem I'm complicating my life alot. Anyway.

I'm not sure I get the yellow hit mod/white hit mod thing.

You seem to be using 2.4 weapon speed normalization for stuff like ww/op, etc, even though your main hand is a dagger.

Also I think the whole point of stuff like this is to compare specs versus one another, 2-h is easier to do then dw in theory too.

The thing with crusader and flurry is that weapon speed change the up time obviously. The slower your weapon speed, the higher the uptime you will have on flurry and crusader due to instants being a constant. Stuff like windfury will increase the uptime on crusader/flurry alot. This is very hard to model exactly but it's possible to have some kind of estimate. What I did for crusader is assume 1ppm on autoattack. That give you a % per swing then you calculate the number of swing you have per second and get an uptime from that. Obviously it's an estimate and the higher the uptime the more likely you are to overlap, same with flurry.

Lastly I think using some kind of "hard coded" 55/45 formula to spread the value of specials versus white dmg is very flawed. The whole point of doing a spreadsheet like this is to not do that. You know how much rage per second you generate, you know how much rage per second stuff like bloodthirst and ww cost, you should be able to calculate how much special you can do in that time. You could use a 80%/20% rule for "other" specials and executes for the last 20% of the fight though and assume you spam execute as soon as it's up (got enough rage for it and your weapon delay is reset, possibly giving 1 bonus for using bloodrage). Also not sure about heroic strike but do you calculate that it lower your white dmg, thus your rage generation and that it's using the special hit table?

But overall good work, nice presentation too.

 

malthrin

The Warrior dps and Flurry/Crusader uptime models are quite a bit more complicated than Rogues due to the feedback of white damage into rage. Lacking the statistics background to even try to address that, I'm tempted to just model it and run it on the Beowulf a million times or so for some ballpark figures. Can anyone offer an optimal instant attack pattern for DW fury? Specify with/without windfury for bonus points.

 

Deathwing

Yellow/white hit mod is basically a multiplier for all yellow and white damage. It takes the hit table for those types of hits and averages it down into a damage multiplier. It's very hard to look at excel formulas and understand what the author was trying to do. For example, let's say your yellow hit table consisted of just misses, hits, and crits, this is how I would get the yellow hit mod:

M*0 + C*2 + (1 - M - C)

M and C being miss and crit chance, respecitively. So, averaged out, every yellow hit would be multiplied by that equation above(I simplify it down in my spreadsheet.

I thought dagger normalization only applied to rogues. Is it 1.7 for warriors too? It would make a small difference in dps.

I thought crusader was a flat 2 ppm(1 ppm in offhand?). Changing weapon speed shouldn't affect that at all. The difficulty I saw was from it overlapping with a previous proc.

Yes, I agree the 55/45 isn't that great. I put that in there for comparison's sake. And for the time being, my gear seems to adhere to that rule after a long night of raiding. You were spot on with my 80%/20% rule, that's exactly what I do.

For heroic strike and cleave, the costs of rage and white damage are associated into the DPR column. And the white damage cost is associated into the DPS column. Same for RPS column and rage cost. Hence why if you notice, HS gets slowly worse as gear improves.


Separate question for anyone. Does Kings multiply every stat, regardless of source? As in, does it work on other buffs, like fort?

 

Deathwing

Not sure what you mean by instant attack pattern. Even with my relatively low level of gear(see profile), I can sustain BT and WW on cooldowns, even though that's not the most optimal thing to do yet. But even at this point, I'm debating whether it's better to stay in berserker stance for WW, or battle stance for overpower, and just dump extra rage into hamstring(better for flurry).

Either way, I'm guessing what you'll want is BT and WW on cooldown, which is 6s and 10s, or BT on cooldown, and hamstring every ~2 seconds.

 

chalon

Those numbers are just from Kalman's uptime table, which he derived using a Monte Carlo simulation in MATLAB. The main hand value actually will change depending on what weapon you have equipped, whereas the off-hand is constant since you don't have any specials which use the off-hand. For a warrior the numbers are going to be different.

Yeah, anything that directly modifies a stat will get a kings bonus. So for instance a 54 sta fort would be a 59 sta fort if you have kings.

As for VLookup, this explains it fairly well:
http://www.mvps.org/dmcritchie/excel/vlookup.htm

But I guess the short story the way those are setup:
-I made a named range for just the item names themselves. (eg., WeaponsList)
-This named range is used for the validation list in that cell (to create the drop-down)
-Then I created another named range that includes both the item names and the stats (eg., WeaponsStatsList).
-Then the VLookup is applied to WeaponStatsList

 

Thorb

Well the basic formula of 1-(1-C)^3 for flurry uptime work decently for auto attack. It doesn't take into account overlap so the real number is slightly lower then that. To incorporate instant attack into the flurry equation you have to replace "3" by the number of attacks you get into a cycle. A cycle being the time it take to execute 3 attacks. There are 2 cases, either you take the hasted cycle time or the non hasted cycle time. Since this formula doesn't take into account overlap like I said, I personally went with the worse case scenario where you are using the hasted weapon speed.

So say I have sulfuras at 3.7 speed, that's a cycle of 3.7*3/1.3 = 8.5 sec. How many attack do I do during 8.5 sec second using all my instants? Non hasted auto attack are about 0.24 per sec, which give a worse case scenario of about 8.5*-0.24 = 2.04 attack per cycle, with an uptime of 44% or so (25% crit).

Once you start adding windfury, bloodthirst, ww, etc, you get into the 0.5-0.6 attack per sec with sulfuras, giving you an furry uptime of 70%-80%. This seem ok since with full buffs, flrurry seem to always be on once "it get going".

I'm not 100% sure on crusader but the idea behind it should be to have a ppm based on auto attack only. So if crusader is suppose to be 1ppm, this probably mean it's translated in a % per swing on a given weapon. Again, using sulfuras at 3.7 speed, this mean that 3.7/60 = 6.1% chance to proc crusader per swing. If you let autoattack at those rules, you should be getting 1ppm. Once you have the %, you apply it to all your attacks, including instants using something like 0.5 aps * 0.06% * 15 sec duration = 0.45%. This doesn't account overlap. With WF it usually give nearly 60% uptime on crusader, which seem a bit high. To account for overlap is getting into very complex maths though unless you can do it in a simulator. But crusader will overlap much more then flurry will due to it's 15 sec duration.

 

malthrin

Sounds all right. Which one would take priority if you didn't have rage to keep both going? Any Cleave/HS? Any rage lost in order to shift -> overpower can just be added to the cost of overpower, which means you can compare its rage/dmg ratio to other instants and decide when it is and isn't worth OPing. Does that change if you're not Flurried/about to run out of Flurry and are willing to sacrifice extra rage to OP and get Flurry up again? How does Execute figure in?

That kind of thing =)

ed: I suppose it wouldn't be too much more to set this up to test different attack logic sequences anyway.

 

Deathwing

Thorb, 1 - (1 - C)^3 does take into account overlap. By subtracting the chance that your last three hits weren't a crit, it covers all possibilities that will proc flurry, including a string of 3 crits.

Your cycle way of dealing with instants is similar to mine. I prefer to go on instant cooldown because that isn't changable at all, while your cycle time can be.

 

Thorb

Nah it doesn't. If it would statistics would not be that hard wouldn't it?

Say that you have an average of 6 crit per minute and your cycle time is 10 sec. What are the chances that you crit 1 time in each cycle? giving you an uptime of 100%? What are the chances you crit 6 times in a row, giving you an uptime of 2/10 = 20%? What are the chances you crit 3 times in a row followed by once each cycle for 3 crits, etc? Sum all the uptime for each of those models over say an entire minute and you will get the exact uptime including overlap.

We are talking literally hundreds and hundreds of permutations.

 

◊ Shalas

Getting 6 crits in a row doesn't suddenly stop you from getting them for the rest of the minute. The formula is accurate because all you care about is the chance that none of the last three hits were crits, which is exactly what (1 - C)^3 says.

 

Deathwing

RISE FROM YOUR GRAVE!! Man, I loved that game.


Anyway, I've been doing some work on my spreadsheet, thought this was a good time to release a new version. Basically, I took what I liked from Chalon's spreadsheet and copied it into mine, I hope he doesn't mind. I used the pull-down menu for selecting gear and enchanted, and I colored it similarly. I've only done this for the 2H sheet. The enchants need some more work as well.

A few notes up front. The skills section is very basic right now. Basically any gear that has a +skill is put in one column. I'm counting on the user to know that the +skill from Death's Sting doesn't apply to the Crul'Shorukh you're offhanding. I got rid of the generic DPS calc and the 55/45 calc, I just use the 80%/20% now.

Please let know of any spelling mistakes, errors, or gear that should be including. I added a quite a bit of gear, so I expect some stuff was entered wrong.

Question for anyone out there: does HoJ only proc off mainhand swings? Still working on modeling that.

 

Thorb

I'll check it out, HOJ should work exactly like sword spec, meaning it can proc both hands and off any weapon attack. I would personally simply count it as a cumulative sword spec. Or simply add +2% since the cumulative effect is very small. 1 * 1.05 * 1.02 = 1.071 or something like that.

Btw can you put the 2-h and DW on the same page? Shouldn't be too bad.

Other edit:
-When I put lvl 60 in the lvl box, I get an effective weapon skill of -5.

If I put lvl 60, change the effective weapon speed to 0 and change the base miss rate of 0.24 in the White Hit Mod to 0.05 (same as specials) and also remove impale, I do not get the same number as the Yellow Hit Mod. These numbers should fit exactly if you remove impale, use the same to hit and remove glancing blows so I'm not sure what's up.

 

Yojimboo

4 things:

1) HoJ is exactly like sword spec in that an offhand proc will produce a mainhand swing. This is going to cause it's combination with a fast offhand/slow mainhand to be better.

2) Is the Net Crit including Berserker stance or should someone using it add a 3 to the crit column somewhere?

3) Are you going to keep updating this with new items? I'm kind of a spreadsheet list/lookup noob...can I just edit the values of some items I won't use to make them new items? I'll mess around and try to figure out how to do it myself.

4) Libram of Constitution isn't adding the HP correctly.

http://ctprofiles.net/482734

 

Deathwing

Yeah, I'm going to update it with new items. I'm going to hold back on that for a little bit until loot tables on most of Naxx is known. You can edit the tables yourself if you want. Just unhide the spreadsheet, and then follow the instructions chalon posted about, if you want them added to the pull-down menu. If you don't care about blue items you know you'll never use, just overwrite them. Please, though, don't edit the spreadsheet and then release it. I'd like to keep control of the releases.

I know Constitution isn't working. I have to add different calc because I can't put it as just +10 sta.

I'll add Berserker stance as an optional buff the user can turn on since you're not always in that stance.

 

Deathwing

I probably can put them on the same page, it would take a bit of work though.

The spreadsheet isn't designed for level 60 because of glancing blows. However, I'll try to work that in my next release.

I'm guessing you don't get the same number because I don't factor misses into the yellow hit mod at all. I assume with the type of people using this, you're going to have +6 to hit. This is probably a bad assumption though, seeing how much +hit is trade for +crit in Naxx. I'll add a miss calc to the yellow hit mod in the next release.

 

Thorb

You should calculate the hit from gear and apply it no? You do already for the 24% penalty right? Btw the thing I did in my similar spreadsheet to have both 2-h and DW on the same page is add a weapon type column, if your main hand is a 2-h, off hand dps is 0. Talents like 2-h spec need to check for if main hand are 2-h, etc. Instead of putting 24% in the formula you can also use a cell for "default miss" that check if main hand is a 2H or no, if it's 2h, default is 5% + skill diff versus lvl if not, it's 24%. Specials stay at 5% + skill diff.

I think it's usefull to have precise to hit calculation since it's one of the weaker (but less costly skill) so in theory if you can stack crit over to hit a sheet like this let you know exactly what you gain and what you lose.

 

Deathwing

Next version is out. Added a bunch of little stuff in this one, biggest one being merging the 2H and DW pages into one. Make sure to read the notes to see how to use it, and read the version control to see what was added this release.

http://www.savefile.com/files/9056225


I'm pretty happy where the spreadsheet is right now. I'll continue adding more things based on feedback and when new items come out. But now, I'm going to shift my attention to writing a java-based combat simulator.

What's the cooldown on Kiss of the Spider?

 

Deathwing

I think people are blaming the warrior dps problem on the wrong source. Yes, better gear means more white damage which means more rage. For while in your gear progression, that then means more yellow damage. But after you're hitting BT and WW on cooldown, and full bar execute when the mob gets to 20%, what else is there to spend your rage on? I'm nearing that stage already, as the jump from blues to MC gear is much greater any other gear progression jump. Your options after that is hamstring, cleave, and HS. HS is an obvious no-no, except in the most extreme cases(C'Thun phase 2, from what I hear). Cleave is purported to have a smallish threat bonus too, so until it's proven one way or another, I shy away from that unless there's 2 mobs to hit. Hamstring is your best option, since it has better damage/rage than single target cleave anyway. But just look at the damage/rage numbers on hamstring compared to BT, WW, and Execute. Basically, after a certain point, the extra rage gained from more white damage is being converted into very inefficient yellow damage.

So, where's the DPS coming from? Execute. I'm sure I don't have to point the damage/rage efficiency on execute spamming. I'm thinking of adding a separate field that breaks up the 80/20 model into 80 and 20, just so people can see what a difference execute makes. A fury warrior is ~100 dps behind combat dagger rogue before 20%. Quick and dirty solution I guess to put a somewhat lengthy cooldown on execute, 6 to 10 seconds maybe. Higher rage executes really kill the efficiency. But that said, I'd rather not see anyone nerfed, rather rogues improved. It'd be a shame to put all this effort into my warrior just to see his dps punked.

Lastly, this is a sort of a disclaimer. I'm not going to say my calculations are accurate. I will say they are at least in the right area, so you can use it as comparison tool between classes, and gear within the class. That's all I intend for this spreadsheet, to compare dps. I welcome and encourage anyone to unhide the sheets and look at the calculations(most are on the DPS sheet though), but I understand if most don't. Sometimes I go back and wonder wtf I'm doing in this cell. Excel doesn't help much either, it really isn't design to handle stuff of this complexity. That's why I'm building a simulator.

 

Yojimboo

Is HoJ fixed in this version?

Also, Unyielding Strength items aren't updated with patch (ring wasn't atleast).

And, is it possible to calculate a DPS excluding execute?

http://ctprofiles.net/482734

 

Deathwing

Yes, I modeled HoJ in this one. It's surprisingly competitive with even the trinkets from Naxx. I knew it was great, but not that great. Sorry about missing the Unyielding set, I'll put that in the next release as well as non-execute dps field.

 

Yojimboo

Yeah, HoJ should be purple. It also comes into account for when a warrior has a slow mainhand and fast offhand, the fast offhand will still proc a mainhand swing, which makes it amazing. Your spreadsheet now helps to see which trinket is better given a specific configuration.

You also forgot to add the new WSG legs, but I figured out how to do it myself, just saying for your next release =).

Also, I have no idea how this would work as PvP is burst DPS, but is there a way to see what your DPS would be like on a player? Would you just put the enemy level down to 60?

1 more thing - Untamed Blade's proc doesn't seem to be taken into account anywhere.

Thanks so much for this, it's amazing, and I can play with it at work and it looks inconspicuous.

http://ctprofiles.net/482734

 

Deathwing

Burst dps in pvp is pretty hard to calculate. Most of the calculations use a long time base to smooth out any inconsistencies(string of misses, no crits for 20s, that sort of thing). You can shorten the fight duration on my spreadsheet currently, but anything less than 30 seconds and I will not guaruntee any type of accuracy. Additionally, PvE can be modeled because the fights are at least somewhat predictable. In pvp, there's a lot thing the other player could do that would screw up any kind of predicted dps, and there's just no way to account for an enemy player's actions without running thousands of scenarios.

Level 60 enemies doesn't work in my spreadsheet. The whole purpose of enemy level was to deal with a pve only mechanic, glancing blows. Doing level 60 would require a lot reworking of the formulas. I will keep it in mind since Thorb mentioned it too.

I'd suggest, if you're trying to determine which gear is better for pvp, you evaluate them in the same way this spreadsheet does for pve. Most of the time, gear that's better dps in pve is also in pvp. You'd then have to take into account things like +skill being overvalued in pve, and how much sta the gear has.

EDIT: ah, I hadn't gotten around to weapon procs. Mainly because there's just so god damn many. Anyone know where I can get accurate PPM or %chance/hit info on weapon procs?

 

Yojimboo

Right, and you can kind of see it too. For example, Edgemaster's Handguards are the best gloves in the game for PvE DPS (test it out), but they'd suck in PvP compared to Flameguard/Annihilation.

EDIT: What about sword spec?

http://ctprofiles.net/482734