Curious Profession Sims

[Jelly 0]

This is my armory: http://eu.battle.net/wow/en/character/hellfire/jellybottom/advanced

As you can see I'm a herbalist/alchemist and was planning to change to tailor for an assumed dps increase. Curiously enough when I ran simulations using simulationcraft herbalism came out ahead. If possible I'd want a second opinion on this, maybe someone can find an error that I made?

To make the sims I just did a normal one on my armory 25000iterations.

To test tailoring I switched out professions at the top, removed the lifeblood action and added the lightweave 3 enchant.

If anyone has current sim information of the best professions or can take a look at this i'd be happy!

Thanks in advance

Jelly

[Jelly 0]

Okay so i was a bit busy during christmas but decided to try and persue this a bit more. This is what my sims say (if we assume that simcraft handles the extra haste from herbalism correctly, which I'm still doubting).

Assume simmed dps for static 320int/320int profession = 0

Using my armory as the base this gives us the follow values for each profession combination:

Tailoring/Herbalism = +1106dps

320int/Herbalism = +706dps

320int/Tailoring = +289dps

Engineering/Tailoring = +204dps

320int/320int = 0

320int/Engineering = -83dps

This sounds really wierd and the sim gives an insane value for herbalism at +706dps compared to a normal 320int profession.

Same comparison but with the t14heroic gear (simulationcrafts BiS):

Tailoring/Herbalism = +1111dps

Engineering/Tailoring = +971dps

Engineering/Herbalism = +968dps

320int/Herbalism = +546dps

320int/Engineering = +476dps

320int/Tailoring = +469dps

320int/320int = 0

Atleast the value of herbalism seems to drop with better gear, but it's still higher than any other single profession.

Edited December 26, 2012 by Jelly

[Damien 1,514]

Sorry for not getting back to you earlier. We have also been very busy with Christmas :P

I'm still busy actually, but I will double check these results later today or tomorrow and let you know what I find.

[NightShade 14]

Not sure here but as affliction haste is a high priority so the sim craft could be picking up the Lifeblood that you receive from herbalism and that is adding about 2880 haste right now at 600 herbalism. It has a 2 min cooldown but lasts for 20 seconds and if you keep it up a maximum amount of time while in a raid it is very possible that it will be better for you. And it seems that Lightweave Embroidery adds a 15 second chance for 2000 intellect but the proc does not have a guaranteed up time which could be set to a lower priority.

[Damien 1,514]

I've done some more thinking and testing and my conclusion is that in SimC, Life Blood will be cast at the same time as Dark Soul: Misery. So, instead of getting +2880 Haste Rating for 20 seconds (from Life Blood), you are getting 3640.

Also, the default action list for Affliction Warlock will always use Soulburn + Soul Swap to refresh your DoTs, as soon as Dark Soul: Misery is up. Therefore, your DoTs will benefit from the increased Haste from Life Blood for more than just 20 seconds, because you will most likely refresh your DoTs before Dark Soul and Life Blood run out.

[Jelly 0]

Not sure here but as affliction haste is a high priority so the sim craft could be picking up the Lifeblood that you receive from herbalism and that is adding about 2880 haste right now at 600 herbalism. It has a 2 min cooldown but lasts for 20 seconds and if you keep it up a maximum amount of time while in a raid it is very possible that it will be better for you. And it seems that Lightweave Embroidery adds a 15 second chance for 2000 intellect but the proc does not have a guaranteed up time which could be set to a lower priority.

Checking the uptime from the Sims gives this

Herbalism: 18.48%

Lightweave: 26.86%

Synapse Spring: 17.26%

So Lightweave have by far the highest uptime of all the three professions.

[Jelly 0]

I've done some more thinking and testing and my conclusion is that in SimC, Life Blood will be cast at the same time as Dark Soul: Misery. So, instead of getting +2880 Haste Rating for 20 seconds (from Life Blood), you are getting 3640.

Also, the default action list for Affliction Warlock will always use Soulburn + Soul Swap to refresh your DoTs, as soon as Dark Soul: Misery is up. Therefore, your DoTs will benefit from the increased Haste from Life Blood for more than just 20 seconds, because you will most likely refresh your DoTs before Dark Soul and Life Blood run out.

Isn't this less true for the gear-stats that I was using for the simulation? Both my gear and the t14heroic setups are using 4part of the tier set which makes DS have a 40s less cd than lifeblood.

Even so if you check the uptimes the value of haste would have to be insane considering it's up for a lot shorter time, and doesn't line up with DS every time either because of the 4set bonus.

[Damien 1,514]

I think that the default SimC duration for a fight is between 450 and 500 seconds. In this lapse of time, you will be able to cast Life Blood 5 times, at 0s, 120s, 240s, 360s, and sometimes 480s. You will be able to cast Dark Soul at 0s, 80s, 160s, 240s, 320s, 400s, and sometimes 480s. This means that you can get 3 casts of Life Blood (out of 5) that are synced with Dark Soul, which is already a huge boost.

[Jelly 0]

I'm not convinced that actually is enough to make such a huge difference. Even if all lifebloods were lined up with DS it would give you 3640 haste with a 18% uptime compared to 2000int with a 26-27% uptime. I mean if the uptime was the same I could see an argument for lifeblood since haste might be valued at 2.2 or something, but with such a huge difference in active time It's a bit hard to accept.

[Damien 1,514]

Tailoring is not 2000 Intellect with a 26-27% uptime. You lose the 180 Intellect back enchant for it, while Lifeblood does not replace anything, it is just something extra.

I got the following stat weights for the Tier 14H profile in Simc, without professions and with Relic of Yu'lon instead of Essence of Terror (so that the Haste proc does not get in the way): 4.14 for Intellect and 2.95 for Haste (with a scale delta of 3000).

With Tailoring, you get 4.14*2000*0.27-4.14*180 = 1490. With Herbalism, you get 2.95*0.19*2880 = 1614.

So Lifeblood is a bit better. Now, there are probably things that make it even better. This is really difficult to model or to properly understand, and this is why we use Simc, because interactions between spells become too complex, especially when it comes to procs.

[Jelly 0]

Tailoring is not 2000 Intellect with a 26-27% uptime. You lose the 180 Intellect back enchant for it, while Lifeblood does not replace anything, it is just something extra.

I got the following stat weights for the Tier 14H profile in Simc, without professions and with Relic of Yu'lon instead of Essence of Terror (so that the Haste proc does not get in the way): 4.14 for Intellect and 2.95 for Haste (with a scale delta of 3000).

With Tailoring, you get 4.14*2000*0.27-4.14*180 = 1490. With Herbalism, you get 2.95*0.19*2880 = 1614.

So Lifeblood is a bit better. Now, there are probably things that make it even better. This is really difficult to model or to properly understand, and this is why we use Simc, because interactions between spells become too complex, especially when it comes to procs.

Hmm that does sound even wierder since the value of herbalism dropped of when I switched to the 14H profile compared to my current gear. Since my statweights the last time I checked was around 4.35 for int and 2.48 for herbalism. By that logic the value of lifeblood should increase while simming the T14H profile?

[Damien 1,514]

Hmm that does sound even wierder since the value of herbalism dropped of when I switched to the 14H profile compared to my current gear. Since my statweights the last time I checked was around 4.35 for int and 2.48 for herbalism. By that logic the value of lifeblood should increase while simming the T14H profile?

By default, simc is using a scale delta of 1,000. You need to manually set it to 3,000 or 2,880 if you want to properly compute the value of the haste gain that you are getting from Lifeblood. Similarly, you need to set the scale delta for Intellect to be 2,000. Also, the default T14H profile is using Essence of Terror (a HUGE haste proc), so this may be interfering with the results.

I already explained all that in my previous message. You need to be very careful when doing simulations and it is important, if you are testing for a particular thing to remove all the potential noise.

[Zagam 1,982]

Math, statistics, simulations, magnets...how do they work?

[Jelly 0]

By default, simc is using a scale delta of 1,000. You need to manually set it to 3,000 or 2,880 if you want to properly compute the value of the haste gain that you are getting from Lifeblood. Similarly, you need to set the scale delta for Intellect to be 2,000. Also, the default T14H profile is using Essence of Terror (a HUGE haste proc), so this may be interfering with the results.

I already explained all that in my previous message. You need to be very careful when doing simulations and it is important, if you are testing for a particular thing to remove all the potential noise.

There is no need to change the deltas. Incase you meant separating the test to just be lifeblood vs lightweave it's kind of a useless comparison since you yourself said there is complex interaction between spells. There is also no use increasing the deltas if you only run a sufficient enough of simulations (not iterations here) to make sure there is no rng - "A good way to check your results are stable enough is to run Simulationcraft twice (with deterministic_rng and seed being disabled, their default values) and ensure you get similar scale factors.". You can read about this http://code.google.com/p/simulationcraft/wiki/StatsScaling#Default_deltas

Really the only thing we can do differently here to make sure things are working correctly is run a bigger amount of sims (even with the Essence of Terror equipped), as it stands now we have a pretty small sample size. We probably need to scale the stats for both my gear and t14h setup a couple of times (using the Essence of Terror for the t14h one, or just redo the profession comparison without that trinket). As it stands the answers we've gotten are not consistent, but considering that the only descrepancy is scale factors it's easy to write it off as a case of having too small of a sample size.

You also used the wrong uptime for lifeblood in your previous calculation, so really it should be 2.95*0.18*2880 = 1529. But really this should realistically be even higher since lifeblood gets lined up with DS during some activations, so a more accurate comparison would be (assuming your 3640haste during DS is correct). 2.95*0.18*3336 = 1771 for a fight with 5 applications of DS.

I'll see if I can get some extra time to run the required scale factor simulations, otherwise we can probably agree that herbalism seems to currently be the best profession for warlocks.

[Zagam 1,982]

The one thing people need to remember is that Affliction Warlocks NEVER EVER EVER stick to the duration of cooldowns and procs used when played optimally. Dark Soul lasts for 20 seconds, but a skilled Affliction Warlock will extend his DoTs to cover anywhere from 40 seconds with UA to 50+ seconds with Agony. Same can be said with Lifeblood and Lightweave. HOWEVER, my interjection here is that Tailoring beats Lifeblood because you can do much more damage with a 2000 Int proc that you can extend with clever use of DoT management far better than you can use the extra Haste.

Haste for Affliction is far less important anywhere between a haste breakpoint. The logical breakpoints are:

1) 3043

2) 4717

3) 6637

4) 9778

5) 13737

Affliction benefits from Haste to speed up the ticks of DoTs to gain an additional tick at certain thresholds, namely the ones listed above that cover multitudes of extra ticks. These breakpoints are the most significant due to the huge gap between the next threshold. Lifeblood gives you 2880 Haste, and this would NOT push you JUST ENOUGH to the next threshold to be effective. Intellect has no threshold...more is better. Intellect also has a much higher stat weight contributing more to your DPS. To do optimum DPS, you have to be considering the 4pc bonus which reduces the cooldown of Dark Soul to 80 seconds, so you're only going to line up Lifeblood the first and 4th time you use Dark Soul if you're using both on cooldown, which you should be. Simulations are nice, but you can't let them do all the thinking for you. You CAN outperform simulation values by thinking outside of the box and truly mastering your class.

[Jelly 0]

While that is true the interaction between DS and lifeblood increase the haste by quite a bit (2% for my character). You also need to consider that while int has a higher value than haste, it's not close to being a huge gap as demonstrated by our scale factor sims (if we can assume they are somewhat correct).

As you can see in my other post I already calculated the DS+lifeblood interactions, although I took the values from Damien and he made a mistake thinking it lined up 3 times. Even so over a long fight the value of lifeblood will be higher than 2880haste, comparing with scale factors shows it as a bigger increase than tailoring

As shown before with either my scale-factors or the t14h:

4.14*2000*0.27-4.14*180 = 1490 for lightweave

As for herbalism it's a bit more tricky to actually get the correct value of haste given by lifeblood during DS. But even assuming you'd gain no extra haste for lining them up (which you do), it would look like this:

2880*2.95*0.18 = 1529

Now basically the problem is to know if the extra "uptime" you can get on lightweave will outdo lifebloods value. Not really an easy solution since the same thing applies to haste, even though in a lesser degree. To be fair the "uptime" increase on tailoring would only have to increase with 1% for it to catch up to lifebloods values, but if you started adding the extra haste from lining up with DS it becomes a bit further away.

3226*0.18*2.95 = 1712 is a bit of a better value for lifeblood, probably not correct either since it's more complicated than this.

Now we are talking about a 4% increase in "uptime" for lightweave, which isn't really hard to accomplish since the duration is 15s and all you'd need to do is cast a dot right at the end to get a big increase. Same thing still applies to lifeblood though, but that is way harder to calculated, and have less of an impact.

Hard to say which is best, without the 4pc lifeblood would probably be miles ahead, but as it stands now they are pretty close, with the tailoring player having to make sure to get as much use out of the proc as possible.

[Zagam 1,982]

While that is true the interaction between DS and lifeblood increase the haste by quite a bit (2% for my character). You also need to consider that while int has a higher value than haste, it's not close to being a huge gap as demonstrated by our scale factor sims (if we can assume they are somewhat correct).

As you can see in my other post I already calculated the DS+lifeblood interactions, although I took the values from Damien and he made a mistake thinking it lined up 3 times. Even so over a long fight the value of lifeblood will be higher than 2880haste, comparing with scale factors shows it as a bigger increase than tailoring

As shown before with either my scale-factors or the t14h:

4.14*2000*0.27-4.14*180 = 1490 for lightweave

You are assuming a 27% uptime here. This is my argument, as a skilled Affliction Warlock won't be confined to this uptime. Increased Intellect is a pure bonus with no thresholds to measure. Maintaining this bonus over a longer period of time will be a very high and significant DPS gain as Intellect gives Spell Power AND Crit.

As for herbalism it's a bit more tricky to actually get the correct value of haste given by lifeblood during DS. But even assuming you'd gain no extra haste for lining them up (which you do), it would look like this:

2880*2.95*0.18 = 1529

Now basically the problem is to know if the extra "uptime" you can get on lightweave will outdo lifebloods value. Not really an easy solution since the same thing applies to haste, even though in a lesser degree. To be fair the "uptime" increase on tailoring would only have to increase with 1% for it to catch up to lifebloods values, but if you started adding the extra haste from lining up with DS it becomes a bit further away.

3226*0.18*2.95 = 1712 is a bit of a better value for lifeblood, probably not correct either since it's more complicated than this.

You are saying if you line it up with Dark Soul...which I stated you cannot do if you are to use both cooldowns optimally. You'll have Lifeblood at time intervals of 0, 120, 240, 360, 480. You'll have Dark Soul at time intervals 0, 80, 160, 240, 320, 400, 480. If the fight lasted 480 seconds (not many do) you could line them up 3 times, so you can't give the value of 1712 DPS to Lifeblood realistically. Since you're theoretically hitting at 1529 as a standalone value, I would suggest you look at it as 1529 x .66 + 1712 x .33 = 1574.

Now we are talking about a 4% increase in "uptime" for lightweave, which isn't really hard to accomplish since the duration is 15s and all you'd need to do is cast a dot right at the end to get a big increase. Same thing still applies to lifeblood though, but that is way harder to calculated, and have less of an impact.

The extra Haste would do nothing to increase the power of the DoTs unless they pushed them into the next threshold. This was my original argument as to why an Intellect proc is better than a Haste proc. If your Haste value doesn't lock a DoT into an additional tick, that extra haste is very close to being wasted. Yes, I realize MG/DS will tick faster, but they won't tick HARDER like they would under Lightweave. Managing the extra time with a Haste proc would be of much smaller value if you can't achieve the next threshold.

Hard to say which is best, without the 4pc lifeblood would probably be miles ahead, but as it stands now they are pretty close, with the tailoring player having to make sure to get as much use out of the proc as possible.

The player would have to monitor their buffs regardless of whether it was Haste or Intellect, so this is a moot point.

[Jelly 0]

I suggest you take a closer look at the math, although its partly my fault for not showing everything. For the lifeblood value of 1712 I assumed it lined up twice over 5 activations. No delay or anything as you seem to think, so your calculations is wrong. If you want to do it your way you can use another value instead of 1712 (since this is already corrected for only two of them lining up), which would be 1988.

My last point isn't moot, since without monitoring the lightweave enchant is significantly worse than lifeblood. This is when player skill comes in which I admit is a bit less interesting and not of much relevance here.

For my gear I do reach the next haste breakpoint during lifeblood, which could explain the increase in its value for me compared to the t14h profile, even though the statsweights says differently.

In the end simcraft probably shows too low of a value for tailoring while lifeblood is a lot closer to its real value in simulations. The value of tailoring would depend a lot on the fight I'd say. There is also the factor of getting faster MG's, but I'd say this is a minor point compared to the rest.

MG Speeds:

Normal: 3.60

Lifeblood: 3.39

DS: 2.77

DS+Lifeblood: 2.61

This is just taken from my gear though

*edit

DS speeds:

Normal: 10.80s 6ticks

Lifeblood: 10.18s 6ticks

DS: 8.31s 7ticks

DS+Lifeblood: 7.96s 8ticks

Since you never use all ticks (except during bloodlust + DS) not sure how much of a value this is. Probably not enough to skew the results in the direction of herbalism.

Edited December 28, 2012 by Jelly

[Zagam 1,982]

I was only using your values as a comparison. I haven't looked into it specifically, so there's always that. I think I am partially impartial to stats other than Haste because of how they work when combined. The more haste you have, the less effect it has. I know that putting them together multiplies your ratings together for a larger effect, but haste has a heavy hitting diminishing return that many overlook. The formula for cast time looks like this:

Cast Time = Base Cast Time / ((Spell Haste Percentage / 100) + 1)

Now let's use something with an arbitrary number. Let's say the Base Cast time of Malefic Grasp is 4.0 seconds. I may or may not be correct, but let's assume I am. Now let's assume you have 0 haste. Your cast time would be:

Cast Time = 4.0 / (0/100 + 1) = 4.0 sec

Now let's look at increments of 5%, which would be 2125 rating.

5% = 4.0 / (5/100 + 1) = 3.8095 sec

10% = 4.0 / (10/100 + 1) = 3.6364 sec

15% = 4.0 / (15/100 + 1) = 3.4783 sec

20% = 4.0 / (20/100 + 1) = 3.3333 sec

...

100% = 4.0 / (100/100 + 1) = 2.0000 sec

I wish this forum supported a chart format so I could show you all of the figures. Basically the first 5% of Haste decreases your cast time by 0.1905 seconds. The next 5% decreases it an additional 0.1732 seconds. The next 5% decreases it an additional 0.1581 seconds. While Haste is good, there is a reason it's good to a point and then stops being as beneficial as other stats.

To give you some sense, my Warlock uses the 6637 Haste threshold, so he sits at 15.62% haste without any buffs. With 5% haste buff in raids, I sit at 21.40% haste. Lifeblood would push me up to 45.68% which would only decrease my casting time by about .04 seconds which is the similar effect haste would give you in going from 5% to 15% haste. So you can see that the more haste you have, the worse Lifeblood gets. Lifeblood would LOSE MORE VALUE the better your gear is. This is because of the diminishing return effect listed above. If Lifeblood is good at all, it would be for Warlocks who do NOT push the Haste thresholds I recommend and decide to go for full Mastery. Lifeblood would have a much larger effect on that playstyle than the playstyle I recommend.

I do value your math and appreciate the opportunity to do some theorycrafting. I just hope in some way I've enlightened you on how Haste REALLY works and what procs like that mean.

[Jelly 0]

I don't disagree with that, and I didn't try to argue against the DR of stats such as haste. You can see how heavy the DR is in my MG comparison for example.

When I first made this thread it was because I thought simcraft valued lifeblood too highly because of some wierd mechanic or bug, but after this back and fourth conversation it's more likely that Lightweave is the one not being handles optimally with refreshes etc, while lifeblood probably works pretty well in the simulator.

I'd agree that tailoring is the better profession in most cases, but that lifeblood is quite a good alternative to normal 320int professions (which seem to preform the worst, not counting skinning/mining..). The same rules that apply to lightweave would also explan why engineering sims so low.

Is this a similar conclusion that you would draw?

[Zagam 1,982]

Engineering is the default profession for BiS profiles loaded into Simcraft...so not sure where you're seeing low simulation values. Synapse Springs should have a 16.6% uptime (10 sec, 60 sec CD). I don't think either of them simulate accurately, but they're probably close.

[Jelly 0]

I know it's the default one. But changing it for herbalism gives a higher sim, which is was started me off on this journey, initially assuming that lifeblood was handled incorrectly somehow.

Maybe there is something wonky with how I sim. If you have time you could test it yourself, just switching engineering for herbalism in the BiS profile. I was actually planning on switching to what I assumed was the best professions (Engineering/Tailoring), but I wanted to check what I'd gain before actually bothering to do so.

I think my initial assumption about those to being the strongest is still correct and that simcraft just isn't using them to their full potential, atleast not with the default code. Could always try and revising some of it to see if we can make it reapply dots just when buffs are about to end.

[Zagam 1,982]

I just removed Engineering and placed Herbalism in and saw a 30 DPS loss...essentially negligible. Like I said, Simcraft is neat, but I've beaten their estimated max dps simulation for me by 20k before. I haven't really bought into it a great deal. I use it as a guide, not a bible if that makes sense.

[Jelly 0]

I agree with that aswell. Most fights aren't even close to being simmable (word?). The only somewhat compareable fight is normal Feng and even then you can just outplay the simulation. I use it as a tool to check reforges/gemming and before this, professions.

Will probably end up changing to tailoring then as I'd agree that it's the better profession, although my pockets will cry of all the expensive Jade Serpent potions.

[Zagam 1,982]

Level alts! I had the 600 of all profession achievement rather early. With some movement of items between my toons and market knowledge, I let the game level my professions and make buckets of gold. Never ever ever buy crafted potions, flasks, etc. It's too easy to level a farmer. Don't buy...unless it's from me ;)