Sign in to follow this  
Followers 0
Ghuerilla

Stats Help... Assassination

17 posts in this topic

So I have spent a couple days spamming LFRs and BGs to get some gear for my alt assassination rogue, and I am now sitting at 516 item level. And I have become thoroughly confused now on the stat priority. AMR says Haste>Crit>Mastery. The icyveins guide says Mastery>Haste>Crit. And the post in green here:

http://www.icy-veins.com/forums/topic/4293-reforginggemming/

says Crit>Haste=Mastery. I know that mastery will boost damage when actively pooling energy before envenom; haste will give more rupture ticks and energy regen; crit will generate more combo points, and give loads more 2-piece procs. But after looking at more than one credible source and seeing many different guides on stat priority it seems impossible now to determine which is the best. Does anyone have some insight on this?

Here is a link to my armory for now:

http://us.battle.net/wow/en/character/arthas/Gorillaa/simple

Right now I just copied what AMR told me to do although I skipped a few expensive enchants and the belt buckle to avoid wasting money on quickly-replaced gear. (Haste>Crit>Mastery.)

Edited by Ghuerilla

Share this post


Link to post
Share on other sites

It's a little bit up for debate still, just based on weapon damage, and gear levels.

The way I was taught was sub-520, you go Mastery>Haste>Crit, and over 520 you go Haste>Mastery>Crit.

I don't really trust in the Crit>Haste>=Mastery, because the 2pc was improperly calculated (as 15 like the WoWHead toolkit says, instead of 6 which is accurate), and I prefer a faster playstyle.

Share this post


Link to post
Share on other sites

The reason that there are so many different opinions on the reforging discussion about assassination rogues is that there is too much of a balance between all of the three stats.

 

On your item level, mastery turns out to be the "most" valuable" stat, but I personally would reforge based on my personal preference. If you like a spammier build, you should reforge into haste and if you like your poisons to tick even higher than they are right now, you should go for mastery. The only secondary stat I don't quite like is crit, but that's just personal preference...

 

The thing that I recall about the reforging is that the legendary meta gem and the cloak define a lot of your reforging. As soon as you get these two items, haste becomes the most valuable stat because of their proc chance.

Share this post


Link to post
Share on other sites

Secondary stats change value depending on the value of the other stat along with many other factors such a the ledgendary cloak and gem and set pieces. A priority isn't set in stone, it will change. The best way to  find out what your priority is for any rogue spec is to use shadowcraft. It is dps modelling program that has been in use since wrath and it is very accurate.

 

Here is a basic guide.

 

As for a general hand wavy explanation, the stat values of assassination are usually so close in value that they can and often do change the order of priority. This also means that the impact of reforging and gemming is going to be minimal. For the most part, after hit and expertise caps, the vast majority of any gear optimization that can be done by shifting secondary stats isn't going to matter much.

Edited by WaveFunctionP

Share this post


Link to post
Share on other sites

A lot of people come in touting ShadowCraft as the number one way to reforge and gem... But why?

 

The way it's currently designed is actually kind of poor, and it relies on SimC to provide it results each time you audit, but not for each change.  So for something as close as Assassination, it can recommend full Crit, then once you gem it, it'll recommend full haste.  It's not smart enough to see the breakpoints at which haste plateaus, and instead looks at how more crit would affect your current situation.

 

If you want to use ShadowCraft to it's fullest, I'd make single changes, update, reaudit each time, or else it'll give some really funky results.

Share this post


Link to post
Share on other sites

A lot of people come in touting ShadowCraft as the number one way to reforge and gem... But why?

 

The way it's currently designed is actually kind of poor, and it relies on SimC to provide it results each time you audit, but not for each change.  So for something as close as Assassination, it can recommend full Crit, then once you gem it, it'll recommend full haste.  It's not smart enough to see the breakpoints at which haste plateaus, and instead looks at how more crit would affect your current situation.

 

If you want to use ShadowCraft to it's fullest, I'd make single changes, update, reaudit each time, or else it'll give some really funky results.

 

If you come from another class, I can see why you'd want depend solely on simc. Because its all you have.

 

SC is essentially a spreadsheet. It doesn't depend on SimC. Both are best used in conjunction with each other, but for most player's purposes, shadowcraft will be easier to use and accurate enough.

 

If you see oscillating priorities, it means that the values are so close that it doesn't really matter with priority you go with. The general advice is to optimize gems and then reforge. You shouldn't see much oscillation in gem recommendations, but it's fairly common for reforging. Just take your preference, or go with the highest value.

 

The reason it does this is because the optimizers use the stat weights it has at the time of optimization. Each time you optimize, it will change those weights, so you can run into cases where more of one stat makes another stat more valuable, and so you end up with oscillating stat priorities. you can see those weights on the side of the application. It is also fairly common to see oscillation cause by hit/expertise caps, which often have discontinuous break points.

 

The key point, and one that I try to make in my guide is that you should be looking at the significance of the numbers, not just their values.

 

As for accuracy, the SC models are in fair agreement with simc at significantly less computation cost, which allows the model to run on a website used by thousands of players. Discrepancies between simulations and modeling are going to happen, but SC does a good enough job. Consider for a moment that the website is maintained by the rogue theorycrafting community and is based upon some of our oldest tools. It is a well establish, easy to use and accurate model that all rogues should be using.

 

That's why it is recommended so highly.

 

Should you find an error in the model, be sure to tell us, but I think you'll find that most discrepancies are minor when you run the numbers.

 

You mention gemming priorites for assassination, have you look at what % impact the difference makes. That should put it into perspective.

Edited by WaveFunctionP

Share this post


Link to post
Share on other sites

If you come from another class, I can see why you'd want depend solely on simc. Because its all you have.

 

SC is essentially a spreadsheet. It doesn't depend on SimC. Both are best used in conjunction with each other, but for most player's purposes, shadowcraft will be easier to use and accurate enough.

 

If you see oscillating priorities, it means that the values are so close that it doesn't really matter with priority you go with. The general advice is to optimize gems and then reforge. You shouldn't see much oscillation in gem recommendations, but it's fairly common for reforging. Just take your preference, or go with the highest value.

 

The reason it does this is because the optimizers use the stat weights it has at the time of optimization. Each time you optimize, it will change those weights, so you can run into cases where more of one stat makes another stat more valuable, and so you end up with oscillating stat priorities. you can see those weights on the side of the application. It is also fairly common to see oscillation cause by hit/expertise caps, which often have discontinuous break points.

 

The key point, and one that I try to make in my guide is that you should be looking at the significance of the numbers, not just their values.

 

As for accuracy, the SC models are in fair agreement with simc at significantly less computation cost, which allows the model to run on a website used by thousands of players. Discrepancies between simulations and modeling are going to happen, but SC does a good enough job. Consider for a moment that the website is maintained by the rogue theorycrafting community and is based upon some of our oldest tools. It is a well establish, easy to use and accurate model that all rogues should be using.

 

That's why it is recommended so highly.

 

Should you find an error in the model, be sure to tell us, but I think you'll find that most discrepancies are minor when you run the numbers.

 

You mention gemming priorites for assassination, have you look at what % impact the difference makes. That should put it into perspective.

See, I was instead under the assumption that SC used SimC's models, sort of using their code in a refined implementation.  And I'm not coming from another class; I consider myself fairly accomplished on my rogue, and I've personally run into these issues.  And a LOT of weirdness.

 

For example, I just loaded my character (Valakondria-Mal'Ganis) and it's telling me to reforge/gem for more Agi and more crit, but both are DPS losses, sim and real world.  Sub/Combat, it has always worked fine.  But Assass, it has never been correct or offered me a DPS gain, and I guarantee that if I was to change my gems/reforge to that, I could 'reoptimize' and it would tell me to change it back.

Share this post


Link to post
Share on other sites

I don't know how you can know real values without having spent hours running perfect rotations on a test dummy with full raid buffs. Simc is always going to diverge slightly given different fight lengths in the model.

 

I put your rogue into sc and tested assassination, and the variance between optimizations was 0.3%, which is expected given such close stat weights. Optimizations for subtlety resulted in a similar outcome, variance of around 0.3%.

 

I'm not sure what you hope will be accomplished by these tools. Those variances are likely insignificant, and certainly imperceptible in regular play.

Share this post


Link to post
Share on other sites

It still doesn't explain why it optimizes me downwards.  I know it's a small variance, but any variance downwards is incredible to even see.  Why would it 'optimize' me even 0.00001% lower?

 

I think I'll continue to recommend AMR w/ custom fixed stat weights based on SimC reforge plots and stat weighting.

Share this post


Link to post
Share on other sites

It still doesn't explain why it optimizes me downwards.  I know it's a small variance, but any variance downwards is incredible to even see.  Why would it 'optimize' me even 0.00001% lower?

 

I think I'll continue to recommend AMR w/ custom fixed stat weights based on SimC reforge plots and stat weighting.

 

So to you it's easier to install and setup simc properly, and then manually input those values into mrrobot rather than press a button a couple of times?

 

I just don't have little faith that the average player can utilize simc properly, nor do I believe it is necessary.

 

Nevermind the issues that players have with downloading the wrong software and getting viruses. I know thats not a simc issue. It's a player issue and that's my point. If you goal is to help inexperienced players, you would do well to direct them to shadowcraft rather than confuse them with simc.

Edited by WaveFunctionP

Share this post


Link to post
Share on other sites

It still doesn't explain why it optimizes me downwards.  I know it's a small variance, but any variance downwards is incredible to even see.  Why would it 'optimize' me even 0.00001% lower?

 

I think I'll continue to recommend AMR w/ custom fixed stat weights based on SimC reforge plots and stat weighting.

 

This has been extensively discussed in other places but apparently not over here.  Basically here is the deal, ShC takes a snapshot of EP values and passes those to the reforger which then returns a set of reforges based on those EP values.  Due to synergistic stat scaling (for example: haste increases the value of mastery more than it increases its own value) the change is stats from the reforge, that appeared optimal when you started the reforge ceased to be optimal somewhere in the middle.  This can cause ShC to show a dps loss when you optimize reforges.

 

If you really want absolute highest possible dps from ShC you have to do some optimization by hand, basically try and change reforges from the lowest EP value to the highest EP value until the EP values equalize.  In practice this generally isn't worth doing because the net dps gain you get out of it is on the order of 0.1-0.3% which so small as to not be noticeable and most people recommend you just use the higher of the two reforges you get out of the optimizer if you do experience that ringing behavior.

 

Its also worth noting here that AMR and SimCraft technically suffer from this same issue however neither makes it quite so clear.  When AMR says something is guranteed optimal, what they mean is guranteed optimal assuming the stat values provided are correct however those stat values may, after the reforge is complete, may not in fact be entirely correct.  SimCraft can technically help here with 3 dimentional stat delta plots but that is time consuming and generally overkill.  All of this is mostly academic because we're talking about dps deltas of tenths of a percent of dps which are unlikely to ever become relevant in the real world.

Edited by Fierydemise

Share this post


Link to post
Share on other sites

This has been extensively discussed in other places but apparently not over here.  Basically here is the deal, ShC takes a snapshot of EP values and passes those to the reforger which then returns a set of reforges based on those EP values.  Due to synergistic stat scaling (for example: haste increases the value of mastery more than it increases its own value) the change is stats from the reforge, that appeared optimal when you started the reforge ceased to be optimal somewhere in the middle.  This can cause ShC to show a dps loss when you optimize reforges.

 

If you really want absolute highest possible dps from ShC you have to do some optimization by hand, basically try and change reforges from the lowest EP value to the highest EP value until the EP values equalize.  In practice this generally isn't worth doing because the net dps gain you get out of it is on the order of 0.1-0.3% which so small as to not be noticeable and most people recommend you just use the higher of the two reforges you get out of the optimizer if you do experience that ringing behavior.

 

Its also worth noting here that AMR and SimCraft technically suffer from this same issue however neither makes it quite so clear.  When AMR says something is guranteed optimal, what they mean is guranteed optimal assuming the stat values provided are correct however those stat values may, after the reforge is complete, may not in fact be entirely correct.  SimCraft can technically help here with 3 dimentional stat delta plots but that is time consuming and generally overkill.  All of this is mostly academic because we're talking about dps deltas of tenths of a percent of dps which are unlikely to ever become relevant in the real world.

And I understand how that works, but it still should have checks in place to not optimize me downwards.  And SimC can tell me these things with reforge plots.

Share this post


Link to post
Share on other sites

And I understand how that works, but it still should have checks in place to not optimize me downwards.  And SimC can tell me these things with reforge plots.

 

SC's combat model, which spits on the dps number, is server side. The script to optimize gems and reforges is a browser script. In order for the reforger to optimize perfectly, it would need to poll the server many times to update the dps number for each possible forge. Enough so that it could lead to performance issue for the server.

 

Mrrobot is easy to use, but it's stats weights are inaccurate without simulation or modeling data. It does however do a good job of taking care of caps, which is the vast majority of the optimization available via reforging, so I'll give it that. It also can't tell you the magnitude of any change you made, which shadowcraft can do, since it is an actual model, not just an optimizer.

 

Each tool has it's uses, and it is my opinion that simc is an advanced tool, and should only be used by those that take the time to learn to use it, and it overkill for the vast majority of players. Meanwhile, there is nothing that mrrobot can do that shadowcraft can't do. You can manually input stat weights into shadowcraft if you really want to bother. Most people don't, because the calculated weights are plenty accurate enough. Certainly accurate enough to be within error of anything simc can spit out.

 

To each his own, but I'd rather have new players sent to any easy to use and accurate tool that lets them play around with gearing so that they can learn about stats intuitively, rather than have them run simulations, which can be time consuming, and not to mention difficult to use and easy to mess up.

Share this post


Link to post
Share on other sites

I understand the complications of computing, but why couldn't it run simplified scenarios, such as, if the result is a negative delta, then throw an error, or try to artificially weight the value of another stat to see if it results in a positive delta?

Share this post


Link to post
Share on other sites

I understand the complications of computing, but why couldn't it run simplified scenarios, such as, if the result is a negative delta, then throw an error, or try to artificially weight the value of another stat to see if it results in a positive delta?

 

You do get an error. The dps shows a negative value. To compute a delta, one would need access to the output of the combat engine, which is on run on the server, as I said. If it were simple it would have been done already.

Share this post


Link to post
Share on other sites

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!


Register a new account

Sign in

Already have an account? Sign in here.


Sign In Now
Sign in to follow this  
Followers 0

  • Recently Browsing   0 members

    No registered users viewing this page.