The Case for Haste

Warning: Theorycraft inside.

Moderators: Fridmarr, Worldie, Aergis, theckhd

The Case for Haste

Postby Ashaton » Thu Nov 27, 2014 10:16 pm

Hello fellow theorycrafters,

Recently it's become quite clear, thanks to Theck's work, that haste is in a bad spot. Through mechanics and rating conversion changes, all of the other stats have come out ahead. As we used to love haste throughout Mists and it is currently our official attunement stat, this seems like a problem that needs to be addressed. I've created this thread as a hub for sharing and scrutinizing ideas that can help restore it to its former glory.

To summarize Theck's analysis of the problem, I'll start by recapping our competing secondary stats purely from a survivability standpoint (ignoring bonus armor because it's in a league of its own):

  • Mastery benefits Improved Block and Holy Shield because of its impact on block chance, and strengthens Sacred Shield and Seal of Insight because it increases our attack power and thus our spell power.
  • Crit now affects Seal of Insight and Sacred Shield, and adds to our Parry rating thanks to Riposte.
  • Multistrike also affects Seal of Insight and Sacred Shield, and boosts our incoming healing via Shining Protector.
  • Haste reduces the cooldowns of some of our core abilities via Sanctity of Battle, and Sacred Duty gives us more of it.
  • Versatility has no special interactions.

Essentially, all of the stats now improve our survivability and many of them do it in multiple ways. Mastery in particular wears a lot of hats. It's so strong in this respect that some people have suggested we make it our attunement stat instead of haste. Thematically it certainly fits; paladins are block tanks, mastery helps us block, and we get extra damage and healing throughput as a bonus. With all of these ability synergies, it just seems to make sense that mastery would be our stat of choice. Combine this with the fact that haste's conversion rating was severely reduced relative to the other stats in Warlords, and you arrive at our conundrum.

So why is this a problem worth solving in the first place? Why not just officially switch our attunement, forsake haste and be done with it? Personally, I am going to make the argument that haste is a fun stat to stack, and makes the game feel more exciting. I suspect I'm not alone in thinking that the current Protection rotation feels very slow compared to how we played the game a month ago. I constantly feel like I'm waiting for my GCDs to finish. It's purely a subjective issue, but it just feels wrong to me. While I respect that we are in a new expansion and our power levels have to be reset, I'm looking further down the road towards the next few raid tiers. If haste maintains its position as our worst stat, then we'll never get to see those one-second GCDs again. That would frustrate me.

So, in an effort to preserve the feel of our Mists gameplay where haste-stacking was a thing, let's take a hard look at what kinds of changes would need to be made in order for haste to make a comeback. I have volunteered to work on modeling some new mechanics in Simcraft and take some of the burden off of Theck, so I'll be happy to test any and all ideas we come up with and see how they hold up. I will try to maintain a list of them in this post as the thread goes on, along with pros, cons, and sim results of each one.

-----------------------

IDEAS (last edit = 12/1, 10:47 pm)

  • Increase Sacred Duty's bonus
    The bonus would have to be about 40% instead of 5% to make haste compete with mastery. All other specs have a 5% attunement, so unless the devs plan to make them adjustable per spec, this is probably not allowed anyway. That being said, 40% Sacred Duty makes haste and mastery complement each other perfectly as our gear gets better, which extends to higher tiers as well, not just T17. This would be a great solution if it were allowed!
  • Increase Sanctity of Battle's cooldown reduction
    Lowers some CDs and the GCD, but doesn't increase attack speed or self healing rates. This would help the slow-rotation feeling, but it splits haste into two types (melee/spell haste and... CD haste?) and is vulnerable to GCD-capping problems. Once simmed, this essentially acts the same as a buff to the haste coefficient but without the benefit of speeding up SS so it doesn't look like a winner.
  • New passive that gives haste a side-effect like Riposte does for crit
    Easy to implement, and looks good on paper. "Divine Riposte" (haste gets added to dodge or parry) makes haste surpass mastery handily regardless of which avoidance stat it boosts. The parry implementation keeps haste as a valuable stat from heroic gear all the way to mythic gear, so this solution would work throughout this raid tier. After that, for T18 and T19, it scales too heavily and makes haste far too strong. It can also be argued that haste already naturally increases DPS and, thanks to Sanctity of Battle, already increases survivability - perhaps this idea would simply make it perform too many functions.
  • New proc triggered on melee hits that only benefits from haste
    This seems like one of the best options because it's easy to balance since (ideally) it won't be affected by the other stats. Adding a new mechanic like this doesn't happen in the middle of raid tiers though, so this would have to wait for the next one. Still, I'd love to gather ideas for this effect.
  • Buff to the haste rating conversion, to bring it in line with the other stats
    Theck calculated that we'd need the rating conversion to be ~30% higher for it to be competitive. Since the conversion is global across all specs, I can't see this being allowed. We'd need a skill that is unique to us, and one that isn't static at 5% like the attunement. If only attunements were adjustable per spec! That would be lovely.
  • Make Shield of the Righteous reduce damage based on haste as well as mastery
    This sounds like it would make haste scale exponentially, like Armor Penetration back in the day. It's not a buff if we take SotR's base reduction from 25% down to 20%, and for heroic gear levels, it puts haste extremely close to mastery... the TMI and DTPS numbers are almost identical to live mechanics sims. For T18 and T19, this makes haste way too strong though.
  • Increase SotR's duration based on haste
    Same problem as above, and it would have to be done very carefully so that you couldn't reach 100% uptime with enough haste. Maybe worth simming, but probably too good.
  • Add Divine Protection and Guardian to the list of abilities affected by Sanctity of Battle
    Too situational to be of real use, and too powerful when it is.
  • Move mastery's attack power bonus to haste
    This removes mastery's DPS contribution which makes it a non-starter. Haste already contributes to DPS and this would make it double dip.
Last edited by Ashaton on Mon Dec 01, 2014 8:48 pm, edited 12 times in total.
Ashaton
 
Posts: 55
Joined: Fri Nov 21, 2014 4:41 pm

Re: The Case for Haste

Postby Ashaton » Thu Nov 27, 2014 10:27 pm

EDIT: Ignore this entire post, I didn't do a plot for Sacred Duty which reveals a lot more info than I assumed to have here. See post #36 for the actual results. Tl;dr, Sacred Duty should be at 40%, not 75%.

I've tested out increasing Sacred Duty's haste bonus with the same setup that Theck used in his original post:
  • Using the Paladin_Protection_T17H setup
  • TMI Standard Boss, T17M
  • 2-piece and 4-piece bonuses are disabled

In short, I found that Sacred Duty would have to increase our haste bonus by 75% instead of 5% to make it compete with mastery.

Sacred Duty at 30% bonus:
Image
Source HTML: https://www.dropbox.com/s/uibhudengv67fo1/30p_SacredDuty.html?dl=0
Not much different than where it is today.

Sacred Duty at 50% bonus:
Image
Source HTML: https://www.dropbox.com/s/a9239btt0xwugwe/50p_SacredDuty.html?dl=0
Getting better, at least it overtook multistrike...

Sacred Duty at 75% bonus:
Image
Source HTML: https://www.dropbox.com/s/o394ls3tdxntmb7/75p_SacredDuty.html?dl=0
Now we're talking!

75% seems like an absurdly large bonus and would probably earn us a lot of complaints from players of other classes. It would also lead to serious scaling problems later on in Warlords as gear came with more and more of it.

Okay you can stop ignoring stuff now.
Last edited by Ashaton on Mon Dec 01, 2014 8:54 pm, edited 3 times in total.
Ashaton
 
Posts: 55
Joined: Fri Nov 21, 2014 4:41 pm

Re: The Case for Haste

Postby Ashaton » Thu Nov 27, 2014 11:20 pm

Next up, I tested the "Divine Riposte" idea. That is, haste adds an equal amount of rating to dodge, just like Riposte adds our crit amount to parry. With the T17H gear set, this gives us 502 dodge rating which comes out to a total of 8.32% dodge, up from 5.0% baseline. Apparently that extra 3.32% is a huge amount of extra mitigation, on top of the 23.93% parry that we already have. So much so, in fact, that haste immediately rocketed past mastery:

Image

In the baseline sim, we avoided 20.3% of attacks. In the Divine Riposte sim, we avoided 23.7%. That only resulted in an extra 1.73 holy power gained from Grand Crusader over the duration of the sim, which isn't much at all. I'm not entirely sure if I believe these results because they don't seem to provide that much impact, but I guess the extra 3.4% of attacks that we avoided makes the difference?

EDIT - Plot finished overnight (had to take BA out because the URL was too long when submitting it to the Google Chart API):
Image
It doesn't look like there are any major breaks or jumps like we've seen in the past, but it's not perfectly linear either. I'd like someone to verify these results before I get too excited. SimC output is here: https://www.dropbox.com/s/vus1nig84knidvo/DivineRiposte_Plots.html?dl=0
Last edited by Ashaton on Fri Nov 28, 2014 5:30 am, edited 2 times in total.
Ashaton
 
Posts: 55
Joined: Fri Nov 21, 2014 4:41 pm

Re: The Case for Haste

Postby Finkum » Thu Nov 27, 2014 11:37 pm

I don't think that in and of itself a huge Sacred Duty bonus will earn us too many complaints from other classes (assuming we are appropriately balanced around it) - recall that both BM Monks and Guardian Druids have, in the past, had very favourable rating bonuses to haste, crit etc when in the appropriate stance/form. However I believe these bonuses were scaled back or removed for WoD so the devs may not be willing to introduce a new one.

My thoughts on the proposed ideas:

Boost sacred duty: Straightforward to implement, and as such a good band-aid but probably not ideal long-term (as above and especially if attunements end up being binned).
Increase SoB CD reduction: This would be my preference as it addresses the main issue I have right now, that everything feels so slow and cumbersome. [Edit: Downside would be that you would soft-cap (reach GCD) at presumably quite modest amounts of haste.]
New passive side effect or new melee proc: Difficult to introduce and balance in a timely manner at this point in the expansion, and if delayed until 6.1 the devs may run into the same issue they had in MoP (except then it was parry/dodge) where momentum has moved too far to some other stat (in MoP, haste, but now, presumably mastery).
Haste rating conversion change: I'm pretty sure that your Sacred Duty experiments show that the rating conversion change would have to be massive to make it competitive (which would have big knock on effects for every other class); thus, impractical.
ShoR benefits from Haste via increased DR/duration: Would be interesting to see this modelled in SimC. Intuitively it would make haste/mastery strongly symbiotic, so you might get some weird stat weights where the value of crit/vers/multi actually diminish as you stack more haste/mastery?
DP/GAnK benefit from Haste: Not a fan. Will make us absurdly good for any fights where we can achievably CD every occurrence of [Big Nuke], and will have no value otherwise. All other tanks would need something equivalent for parity.
Last edited by Finkum on Fri Nov 28, 2014 12:00 am, edited 1 time in total.
Finkum
 
Posts: 28
Joined: Tue Dec 22, 2009 9:39 pm

Re: The Case for Haste

Postby Finkum » Thu Nov 27, 2014 11:54 pm

Ashaton wrote:EDIT: I'm running some plots for this sim to see if haste is doing any of those weird breakpoint shenanigans. I'll update this post in the morning with the results.


Heh I was just about to suggest this might be what's happening here. Although if it were, you would expect the value of haste to be high(er) even without Divine Riposte (unless Divine Riposte changes the nature of the breakpoints somehow). I assume the gear-set you used is unchanged from your Sacred Duty tests?
Finkum
 
Posts: 28
Joined: Tue Dec 22, 2009 9:39 pm

Re: The Case for Haste

Postby Ashaton » Fri Nov 28, 2014 12:10 am

Finkum wrote:Heh I was just about to suggest this might be what's happening here. Although if it were, you would expect the value of haste to be high(er) even without Divine Riposte (unless Divine Riposte changes the nature of the breakpoints somehow). I assume the gear-set you used is unchanged from your Sacred Duty tests?


Same gear set as the baseline (I am using Theck's post-hotfix file to make sure that my baseline is the same as his). The only thing that changed between that and the Divine Riposte run is this:

Code: Select all
double paladin_t::composite_dodge_rating() const
{
   double d = player_t::composite_dodge_rating();

   // Divine Riposte!
   d += composite_melee_haste_rating();

   return d;
}


I'm happy to provide my sim HTML so you can check it for mistakes, but I can't upload them as attachments and don't know have a dropbox so I'm not sure how to deliver it. The plots are still chugging along, at 19/48 now. I have to go enter a turkey-induced coma but I'll be back in a couple hours with the results.
Ashaton
 
Posts: 55
Joined: Fri Nov 21, 2014 4:41 pm

Re: The Case for Haste

Postby Schroom » Fri Nov 28, 2014 2:51 am

Ashaton wrote:
What's the best way to upload the HTML output from Simcraft, in case anyone wants the full report?


I always use dropbox.

makes it looks like this: https://dl.dropboxusercontent.com/u/38739176/simc_report_Schroom_05-09_PTR_EF_noset.html (very old sim, this is only an example)


Finkum wrote:Increase SoB CD reduction: This would be my preference as it addresses the main issue I have right now, that everything feels so slow and cumbersome. [Edit: Downside would be that you would soft-cap (reach GCD) at presumably quite modest amounts of haste.]


I like this idea. in order to circumvent the early capping problem, Blizzard could change our GCD cap from 1s to 0.5s.


and to be honest even 1s GCD felt to slow sometimes, on some situations and I would have wished for it to be lower.

this model (lower than 1s GCDs) also works for DKs. so why not for us :)
User avatar
Schroom
 
Posts: 671
Joined: Mon Aug 15, 2011 2:03 pm
Location: Luxembourg

Re: The Case for Haste

Postby Thels » Fri Nov 28, 2014 4:11 am

Ashaton wrote:Increase Sacred Duty's bonus

All specs receive 5%. Having a single spec receive more than 5% would be really odd.

Ashaton wrote:Increase Sanctity of Battle's cooldown reduction

Confusing. 10% haste means you can use your abilities 10% more. If SoB would reduce twice as much, it wouldn't be 10% anymore.

Also, unless this would also affect the GCD, it wouldn't help as you'd still be GCD capped, unless you choose not to spam your fillers, just so you can press CS just a little earlier, which would be a major DPS loss, and hard to balance around. Not good.

Ashaton wrote:Buff to the haste rating conversion, to bring it in line with the other stats

Except the Haste rating is uniform among all specs, so this cannot or at least should not be done.

Ashaton wrote:Add Divine Protection and Guardian to the list of abilities affected by Sanctity of Battle

This would make Haste pick up Readyness on the side. It might help a lot on some encounters, and not at all on other encounters. There was a reason they ditched Readyness. Having a talent to reduce their cooldowns by a static amount would make more sense. Oh wait, there is such a talent, though it doesn't affect Guardian.

Ashaton wrote:Make Shield of the Righteous reduce damage based on haste as well as mastery

Both complex and lackluster at the same time. It would also cause a problem where SotR reduction could go through the roof.

It would also make Haste boost itself exponentially, which is not a good thing. Remember Armor Penetration?

Ashaton wrote:Increase SotR's duration based on haste

Like above, makes Haste boost itself exponentially, which again, is not a good thing.

Also could lead to situations where SotR is up 100% of the time, in which case you start macroing it to Crusader Strike and Judgment and forget the skill exists.

Ashaton wrote:New passive that gives haste a side-effect like Riposte does for crit

It could do that. This option is viable.

However, Crit and Multistrike didn't have any survivability (outside tiny increments to our small passive heals), so they needed Riposte and Shining Protector to become viable defense stats. Haste already has a viable defense contribution, one they are happy enough with that they gave it to Warriors too. It just needs some tweaking.

Ashaton wrote:New proc triggered on melee hits that only benefits from haste

Sounds like the most perfect way of doing it, though it doesn't have to be from melee hits specifically. It used to get from Crusader Strike and HotR, but then Haste became too dominant, so they reduced and then removed it. Now that Haste needs some love, it seems Blizzard decided to tweak it back in.
ImageImage
User avatar
Thels
 
Posts: 1233
Joined: Tue May 15, 2007 7:30 am
Location: The Netherlands

Re: The Case for Haste

Postby mclem » Fri Nov 28, 2014 6:21 am

Out of interest, is Sacred Shield handled like a HoT, such that Haste makes the ticks come quicker than every six seconds? That seems logical, but the information I'm finding is only talking in the context of DoTs and HoTs, not these, uh, AoTs. That's a benefit not outlined in the OP if so.
Last edited by mclem on Fri Nov 28, 2014 6:27 am, edited 1 time in total.
mclem
 
Posts: 116
Joined: Tue Sep 25, 2007 6:12 pm

Re: The Case for Haste

Postby Ashaton » Fri Nov 28, 2014 6:27 am

The plots for Divine Riposte are done and added to the results post above. It looks like a fairly regular haste curve to me, with the exception of being flat from -1000 to -500 (but our gear only has 502 haste rating anyway).

Schroom wrote:I like this idea. in order to circumvent the early capping problem, Blizzard could change our GCD cap from 1s to 0.5s.


I imagine that a 0.5 second GCD would start running into latency issues for some players and may have odd interactions with spell queueing. That's probably too fast. Buffing SoB essentially makes it act like another haste multiplier, but only for a select few spells and effects. That gives haste this weird result where it's doesn't affect everything evenly and definitely makes it less intuitive. Like the Sacred Duty tests showed, we have to multiply our haste rating by a massive amount before it catches up with other stats, so it seems like any idea that purely multiplies it (by itself, at least) is out. A more reasonable multiplier might work in combination with the other ideas, though.

Thels wrote:It could do that. This option is viable.

However, Crit and Multistrike didn't have any survivability (outside tiny increments to our small passive heals), so they needed Riposte and Shining Protector to become viable defense stats. Haste already has a viable defense contribution, one they are happy enough with that they gave it to Warriors too. It just needs some tweaking.


Thanks for all of your feedback Thels! I want to discuss this point in particular because it really gets down to the heart of the issue. All of the secondary stats are now valuable for us which is a good thing - we don't want to go back to a world where crit doesn't matter at all to us, especially now that reforging is gone. I think the original intent behind Riposte and Shining Protector is that crit and multistrike became viable stats by themselves, but didn't overshadow our "core" secondaries. The problem is that in reality, they do just that.

Right now, haste helps us in two ways. It gets us holy power faster for more uptime on SotR, and it speeds up our self-heals. Theck's work shows that neither of those effects is strong enough to outweigh the benefits of the other stats. When you say that it just needs tweaking, I assume you mean we'd buff Sanctity of Battle so that those two effects start to matter again. As you already pointed out, simply making it multiply haste results in non-intuitive interaction between the stat and the abilities. I don't think a multiplier by itself is the right way to go, so adding a side-effect seems like the next logical step.

Finkum addressed this nicely:
Finkum wrote:New passive side effect or new melee proc: Difficult to introduce and balance in a timely manner at this point in the expansion, and if delayed until 6.1 the devs may run into the same issue they had in MoP (except then it was parry/dodge) where momentum has moved too far to some other stat (in MoP, haste, but now, presumably mastery).


A new proc that only benefits haste would be a nightmare to introduce in the middle of progression, so it would have to wait until the next tier and by then it could be too late. I agree that this is ultimately the best route to fix haste but it's really, really bad timing. The simplest fix I can see for the moment is a subtle one like Divine Riposte. It's easy to add, it seems (from the sims above) to be effective, and it bypasses the intuition problem entirely. Riposte isn't even in our spell book anymore so there's no reason to add a new entry to confuse players; we could just tack it onto Sanctity of Battle as secondary effect for Protection spec. The only thing I worry about is how we would compare to other tanks if we had both a dodge and a parry bonus. Is that 3% extra avoidance too much? I'll probably have to sim again with Mythic gear instead of Heroic to see how it scales.
Ashaton
 
Posts: 55
Joined: Fri Nov 21, 2014 4:41 pm

Re: The Case for Haste

Postby Ashaton » Fri Nov 28, 2014 6:55 am

I tried removing the avoidance proc for Grand Crusader, so it only comes up on CS / HotR hits (basically back to its original form). It does help haste a small amount but, it's still at the bottom. In this configuration we gained an average of 32.43 holy power during the fight, which is way down from the baseline's 43.84 holy power.

Image

Source HTML: https://www.dropbox.com/s/zvb180ry6nzii3w/GC_No_Avoid.html?dl=0
Ashaton
 
Posts: 55
Joined: Fri Nov 21, 2014 4:41 pm

Re: The Case for Haste

Postby Ashaton » Fri Nov 28, 2014 11:52 am

Here are the results for adding haste to SotR's damage reduction, in addition to the boost that mastery provides. The formula is the one Theck suggested in his blog:

SotR reduction = 0.20 + mastery * 0.005 + haste * 0.005

Implementation since this one wasn't just a number tweak:
Code: Select all
double sotr_mitigation;
if (wod_hotfix)
{
   // Convert cached haste back into it's percent form, so it's treated the same as mastery
   double haste_val = (1.0 / cache.attack_haste() - 1.0) * 100.0;
   sotr_mitigation = -0.20 - cache.mastery() * 0.005 - haste_val * 0.005;
}


This was a very interesting sim. It keeps our survivability almost exactly where the baseline is, so it's not the massive buff I was afraid of.
Baseline: 21626 dps, 32232 dtps, 22433 hps (10050 aps), 109.9k TMI
Sotr_Haste_DR: 21630 dps, 31793 dtps, 22332 hps (10062 aps), 109.7k TMI

Our stat weights, however, are right where we want them to be:
Image

Source: https://www.dropbox.com/s/l1l4yum5mths67w/SotR_DR_Haste.html?dl=0

At first glance, this looks like another viable option similar to Divine Riposte. I worry about how haste will scale with this option, though. Like Thels said, it could result in a positive feedback loop where haste's strength ends up growing exponentially. I'll do some plots for this one too, not sure how far out I should let it go though... maybe +1000 won't be enough?

Edit: Plot's done.
Image

Source: https://www.dropbox.com/s/2vl5wkkr67ke5h1/SotR_DR_Haste_Scaling.html?dl=0

Haste seems pretty close to mastery until we reach +300 rating, when it really takes off and beats everything else by a significant margin.
Ashaton
 
Posts: 55
Joined: Fri Nov 21, 2014 4:41 pm

Re: The Case for Haste

Postby Thels » Fri Nov 28, 2014 2:15 pm

Ashaton wrote:Thanks for all of your feedback Thels! I want to discuss this point in particular because it really gets down to the heart of the issue. All of the secondary stats are now valuable for us which is a good thing - we don't want to go back to a world where crit doesn't matter at all to us, especially now that reforging is gone. I think the original intent behind Riposte and Shining Protector is that crit and multistrike became viable stats by themselves, but didn't overshadow our "core" secondaries. The problem is that in reality, they do just that.

Right now, haste helps us in two ways. It gets us holy power faster for more uptime on SotR, and it speeds up our self-heals. Theck's work shows that neither of those effects is strong enough to outweigh the benefits of the other stats. When you say that it just needs tweaking, I assume you mean we'd buff Sanctity of Battle so that those two effects start to matter again. As you already pointed out, simply making it multiply haste results in non-intuitive interaction between the stat and the abilities. I don't think a multiplier by itself is the right way to go, so adding a side-effect seems like the next logical step.


No. Buffing SoB is not the way I'd take, because of what haste represents. 10% haste means you can use abilities 10% more in the same duration. Changing SoB would break that inherent principle. Instead of buffing how much faster we attack, we could tweak the impact of attacking faster. Blizzard took 2 steps in that direction.

First, by providing GC procs off CS/HotR again, there's more HoPo to be gained from attacks, thus faster attacks lead to more HoPo gained. Second, they buffed the starting value of SotR. The more SotR reduces, the stronger Haste becomes, obviously. Admittedly, at decent Mastery levels, it becomes a nerf.

Not saying it's there, but it's a step in the right direction.

Having Haste provide a secondary effect like Riposte/Shining Protector might end up the way to go, but I'd prefer it if they could solve the issue with something that's actually related directly to Haste.
ImageImage
User avatar
Thels
 
Posts: 1233
Joined: Tue May 15, 2007 7:30 am
Location: The Netherlands

Re: The Case for Haste

Postby Ashaton » Fri Nov 28, 2014 2:49 pm

Thels wrote:No. Buffing SoB is not the way I'd take, because of what haste represents. 10% haste means you can use abilities 10% more in the same duration. Changing SoB would break that inherent principle. Instead of buffing how much faster we attack, we could tweak the impact of attacking faster. Blizzard took 2 steps in that direction.

First, by providing GC procs off CS/HotR again, there's more HoPo to be gained from attacks, thus faster attacks lead to more HoPo gained. Second, they buffed the starting value of SotR. The more SotR reduces, the stronger Haste becomes, obviously. Admittedly, at decent Mastery levels, it becomes a nerf.

Not saying it's there, but it's a step in the right direction.

Having Haste provide a secondary effect like Riposte/Shining Protector might end up the way to go, but I'd prefer it if they could solve the issue with something that's actually related directly to Haste.


Ok, I just misunderstood what you meant by tweaking it earlier. I've added your feedback to the OP under each idea.

I agree that messing with SoB's coefficients would be awkward and split haste into multiple different entities - one for reducing CDs, and another for increasing attack speed and SS refreshes / SoI procs. That's probably not the way to go. I also agree with the "step in the right direction" comment. It would certainly feel the best if haste brought its own value to the table rather than being artificially boosted by Divine Riposte, and a melee proc like Grand Crusader does fill this role. It sounds like the ideal solution, but in actuality this particular implementation ended up making haste even worse than it was before. We'd need something in the same vein as Grand Crusader, but it'd have to be immune to interference by the other stats.

I also agree with your last point. Adding a proc that only benefits from haste (like GC but done correctly) is the solution I want to see the most but I don't think it can be done right at the start of a new raid tier. Adding something like Divine Riposte could probably be implemented before the next tier but it feels like a crutch. Honestly, I'm okay with that as long as it means I can get a faster rotation without negatively impacting progression.
Ashaton
 
Posts: 55
Joined: Fri Nov 21, 2014 4:41 pm

Re: The Case for Haste

Postby Ashaton » Fri Nov 28, 2014 4:02 pm

This is the sim for inflating SoB's coefficient, so we get 1.3% of CD and GCD reduction per 1% of haste. Long story short, it keeps our TMI and other metrics the same but doesn't seem to help haste very much.

The codes:

Code: Select all
  void update_ready( timespan_t cd = timespan_t::min() )
  {
    if ( hasted_cd )
    {
      if ( cd == timespan_t::min() && ab::cooldown && ab::cooldown -> duration > timespan_t::zero() )
      {
        cd = ab::cooldown -> duration;
        double augmented_reduction = 1.0 / (1.3 / ab::player->cache.attack_haste() - 0.3);
        cd *= augmented_reduction; //ab::player -> cache.attack_haste();
        cd *= cooldown_multiplier();
      }
    }

    ab::update_ready( cd );
  }


Code: Select all
  virtual timespan_t gcd() const
  {

    if ( hasted_gcd )
    {
      timespan_t t = action_t::gcd();
      if ( t == timespan_t::zero() ) return timespan_t::zero();

      double augmented_reduction = 1.0 / (1.3 / p()->cache.attack_haste() - 0.3);
      t *= augmented_reduction; //p() -> cache.attack_haste();
      if ( t < min_gcd ) t = min_gcd;

      return t;
    }
    else
      return base_t::gcd();
  }


The charts:

Image

Image

The source: https://www.dropbox.com/s/dxw2k64mxs1znyg/SoB_Coef_Boost.html?dl=0

This is a wacky mechanic. The average CS interval was reduced from 4.00 sec to 3.93 sec, so it managed to fit an extra 1.9 into the fight. All told, it generated 4.32 more holy power with the SoB change over the duration of the fight. Even though this change doesn't affect our base attack speed, we also procced SoI 2.8 times more than the baseline. Sacred Shield remained the same between this sim and the baseline with 148.7 ticks for the encounter, because SoB doesn't affect it. This idea just seems like a less-promising version of the 30% buff to Sacred Duty.
Ashaton
 
Posts: 55
Joined: Fri Nov 21, 2014 4:41 pm

Re: The Case for Haste

Postby Ashaton » Fri Nov 28, 2014 8:48 pm

I took a look at Divine Riposte again, but this time I made haste add its rating to parry instead of dodge. So in this sim, both crit and haste contribute to parry while dodge remains unbuffed at 5%. The previous version boosted our dodge by 3.32%, but this version only boosts parry by 2.87% because of DR. That doesn't seem to matter much, because once again haste jumps way past mastery.

Image

Image

Source: https://www.dropbox.com/s/pprp8prtf8rl32s/DivineRiposte_Parry.html?dl=0

Avoidance obviously sims extremely well. The parry version of Divine Riposte does slightly worse than the dodge version because of the DR with parry, but we're talking about 108.6k TMI vs 108.4k here. This certainly looks like an effective way to bump haste's value... but it almost feels like cheating. If enough of haste's benefit simply comes from the free avoidance it brings to the table, then it loses a lot of its flavor. That seems incredibly ironic, doesn't it? Dodge and parry coming to the rescue of haste, after their status for the entirety of Mists? It definitely bugs me. I was curious if these avoidance results held true for crit as well - how bad would it be if we didn't have Riposte, so its only benefit was a boost to our self healing? So I ran a sim with the baseline configuration, but Riposte disabled.

No Riposte
Image

Image

Source: https://www.dropbox.com/s/15ifzcoo0qnkk87/No_Riposte.html?dl=0

It gets absolutely crippled. Riposte is more than tripling crit's value. Adding Divine Riposte in as a savior for haste is certainly a viable option for either of the avoidance stats, because they are just so powerful now. It seems like we live in a world where dodge and parry are better than haste. This is madness!

Divine Riposte might serve as an effective stop-gap to save haste until a better idea can be implemented, one where haste itself brings its own value rather than piggybacking off another stat. That's personally the direction I prefer, but like I said before, I'm willing to deal with something like this in order to have an excuse to stack haste again.
Ashaton
 
Posts: 55
Joined: Fri Nov 21, 2014 4:41 pm

Re: The Case for Haste

Postby Finkum » Sat Nov 29, 2014 3:46 am

I agree that it's downright weird for parry/dodge to be so desirable.

In one of your earlier sims, you turned off the GC proc from parry or dodge and yet crit was still significantly more valuable than multistrike/haste, which in combination with your final sim suggests that it really is just the raw avoidance that forms the bulk of the stat value.

I'd guess avoidance's increased TMI value is due to the (comparatively) larger health pools present in WoD in combination with our much weaker self-healing. But in any event, it does suggest that the simplest and easiest fix is to add Divine Riposte, as you propose. A more interesting/nuanced fix can be rolled out later in 6.1.

Edit: Actually I'd be interested to see how the stat weights change based on the incoming damage profile in the simulation. Intuitively I would expect parry/dodge to be at least somewhat devalued in fights with non-trival magic or unavoidable physical damage; probably a good idea to check that the stat value of haste with Divine Riposte doesn't tank in such scenarios!
Finkum
 
Posts: 28
Joined: Tue Dec 22, 2009 9:39 pm

Re: The Case for Haste

Postby Ashaton » Sat Nov 29, 2014 10:41 am

Finkum wrote:Edit: Actually I'd be interested to see how the stat weights change based on the incoming damage profile in the simulation. Intuitively I would expect parry/dodge to be at least somewhat devalued in fights with non-trival magic or unavoidable physical damage; probably a good idea to check that the stat value of haste with Divine Riposte doesn't tank in such scenarios!


I'll scale up the damage the T17M boss does by a few different multipliers to see how increased damage impacts dodge and parry. Should have some results back in a little bit - I don't think we need to do scaling plots for these so it will go much faster.
Ashaton
 
Posts: 55
Joined: Fri Nov 21, 2014 4:41 pm

Re: The Case for Haste

Postby Ashaton » Sat Nov 29, 2014 3:22 pm

Ok, in total I ran 9 sims: the baseline (stats and mechanics as they exist today), Riposte disabled, and Divine Riposte (parry). For each one, I simmed with the boss doing 1x, 1.5x, and 2x normal damage. The profiles are the same aside from those differences, and each sim was done with 25000 iterations. I apologize for the messiness of these charts, I'm not really a statistician so I don't have much practice making numbers look pretty. If someone with more experience in interpreting results can suggest a different way to represent them or something, I'm happy to take that advice!
Here's the source sims and Excel sheet for all of these: https://www.dropbox.com/s/fjphi65exveog2u/boss_damage_scaling.zip?dl=0

Image

Image

Image

These charts show the normalized values of each stat (basically, how strong they are relative to stamina). The first thing to note is that stamina starts eclipsing all of the other stats very quickly because in each trial, every stat drops in relative value. I'm pretty sure we've already seen this kind of behavior before when damage starts getting crazy in previous sims, so no surprise there.

Armor stays pretty consistent, but bonus armor drops off quickly. I wonder why that is?

Crit ends up surpassing AP in the baseline at 1.5x damage. From 1.5 to 2x it scales about as well as the rest of them. Without Riposte, crit is terrible and continues to be terrible as damage increases.

With Divine Riposte, haste starts off very strong. From 1 to 1.5x, it evens up with mastery and it seems to maintain that position from 1.5 to 2x.

While this was a fun exercise, I'm going to start looking seriously at Divine Riposte next to see if it makes us overpowered as our gear gets better. Right now we only have a single point of data for it: T17H vs a T17M boss. If it scales too well, then we'll have to adjust it in order to make it a viable solution.
Ashaton
 
Posts: 55
Joined: Fri Nov 21, 2014 4:41 pm

Re: The Case for Haste

Postby Ashaton » Sat Nov 29, 2014 6:45 pm

Okey dokey, Divine Riposte sims are in.

These results consist of 2 sets of 6 sims each. The first set has our 2pc and 4pc tier bonuses turned off, because that's how Theck did his original sims in the post that started this all. The second set has the tier bonuses turned back on. In each set, I ran the Live configuration, then Divine Riposte (Dodge), and finally Divine Riposte (Parry). These were simmed with the T17H gear profile and with the T17M profile. I haven't touched the normal or LFR profiles.

Because there are 12 sims, I don't want to put the scale factors for each on on here - it's just too many pictures. I took some of the commonly discussed metrics from each one and added them to a spreadsheet, so you can compare the numbers yourself: https://docs.google.com/spreadsheets/d/1C6u_c3ojZP63pdIwCWqPRFC6CYpSZfNAGbBR09p1Qj4/edit?usp=sharing

You can grab all 12 source files and see them for yourself here: https://www.dropbox.com/s/rgstm0bsvdi0lm6/Divine%20Riposte.zip?dl=0

Let's talk about the Tier Bonus enabled set, because I think it's the closest to what we can actually expect in-game. The first thing I want to check is how much of a buff this mechanic is, because if it's too strong then we can't use it in its current incarnation. The following graph shows the TMI for the two gear sets and compares Live, Divine Riposte (Parry) and Divine Riposte (Dodge) mechanics:
Image

From a TMI standpoint, both versions of Divine Riposte are almost identical in either gear setup, and both are a buff to our current live scores of course. For the sake of trying to keep the scores as close to live as possible, I'm going to work with the parry version of Divine Riposte for the rest of the analysis.

Divine Riposte buffs our TMI by 1.3k and 1.4k for the T17H and T17M sets, respectively. The difference between T17H and T17M in live is 11.9k. That means Divine Riposte is about 11% as valuable as the difference between the gear. Is that too much? I don't think so, but I am certainly not as experienced as some of the more seasoned theorycrafters here so if someone tells me differently then I will believe them.

Assuming it's an acceptably small buff, let's look at the stat scaling results.

For review, here are the current live scaling results for T17H gear (but this time, with the set bonuses):
Image

Right, haste sucks. Let's see if it does any better with T17M gear:
Image

Oh my. It seems the stronger our gear gets, the worse haste gets by a huge margin! This is really a problem! We already know that Divine Riposte makes haste extremely valuable for the heroic geared sim, but here is T17H gear with the tier bonuses turned on:
Image

Great! Haste is on top again. And now, the moment you've been waiting for, let's see how it does with T17M gear and set bonuses enabled:
Image

Haste has taken a back seat to mastery, but it's still above everything else. I'd call that a win.

To recap my work in this thread so far:
  • Divine Riposte with parry provides a small buff but makes haste a very desirable stat for both T17H and T17M gear sets. This looks like a good option to fix haste (unless I made a mistake in my work somewhere, which is always possible).
  • Adding haste to Shield of the Righteous's damage reduction, but reducing its base reduction to 20%, seems viable after initial sims. I will work on this in detail next.
  • Buffing the haste rating conversion or Sacred Duty would make us special cases when compared to other classes and attunements, and the buffs would have to be massive for haste by itself to be beneficial. These options don't look as promising for scalability and political reasons.
  • Buffing Sanctity of Battle to provide 1.3% CD reduction for each 1% of haste doesn't help enough.
  • Avoidance is really good now, ironically.


EDIT: If Divine Riposte is too much of a buff to be justified, then I thought I'd try disabling Grand Crusader from proccing on avoidance. The results keep roughly the same scale factors while bringing our TMI closer to the live values. It's still a buff, but much less of one. I've added the details from both sims to the spreadsheet linked above.

Source for T17H: https://www.dropbox.com/s/iak8jp914wzy9ov/T17H_TB_DRP_No_GC_Avoid_Proc.html?dl=0
Source for T17M: https://www.dropbox.com/s/y4ccjvb6j085cfq/T17M_TB_DRP_No_GC_Avoid_Proc.html?dl=0

T17H: Live TMI = 107.1k, DR (P) = 105.8k, DR (P) with GC changed = 106.6k
T17M: Live TMI = 95.2k, DR (P) = 93.8k, DR (P) with GC changed = 94.6k

Disabling GC's avoidance proc reduces our total holy power gained from it by 13 and 14.5 per gear set over the course of the fight, which seems like a lot! This dropped the number of SotR casts from 89.9 to 85.1 for the T17M set. Our DTPS goes up by about 600 as a result. However, the TMI only drops by a small amount. It's possible that Divine Riposte could be added along with disabling GC's avoidance proc to make haste valuable without buffing us too much.
Ashaton
 
Posts: 55
Joined: Fri Nov 21, 2014 4:41 pm

Re: The Case for Haste

Postby Ashaton » Sat Nov 29, 2014 9:20 pm

Here are the results for haste being added to SotR's damage reduction bonus. As a reminder, the formula is the one Theck suggested in his blog:

SotR Reduction = 20% Base + Mastery Percent * 0.5% + Haste Percent * 0.5%

Again, I am leaving the T17 set bonuses active for these tests.
T17H Gear: https://www.dropbox.com/s/cxzh5kkii5jqeef/T17H_TB_SotR_Haste_DR.html?dl=0
T17M Gear: https://www.dropbox.com/s/lzc8lmq4sl22czn/T17M_TB_SotR_Haste_DR.html?dl=0

Recall from my initial test that at with the heroic gear profile, this looked like a real contender. It had almost identical DTPS and TMI results as the live mechanics sims, but made haste just as valuable as mastery. The resulting sim with the tier bonuses enabled is largely the same when compared to the live results, so I won't add the charts here for the sake of brevity (but you can see them from the link above).

Unfortunately, when I ran this with the mythic gear profile, haste drops back to the bottom again (but not nearly as bad as it is in the live sim).

Image

It started out so promising, but as gear scales up, haste takes a back seat to the other stats. The resulting metrics for this sim are very, very close to the T17M live sim (for example, live = 95.2k TMI, this = 94.9k) so it's not like this ends up being a huge buff. It just makes haste good early on, and bad later on. Looks like this idea won't work but it was worth a shot!
Ashaton
 
Posts: 55
Joined: Fri Nov 21, 2014 4:41 pm

Re: The Case for Haste

Postby Ashaton » Sun Nov 30, 2014 12:31 am

Just for fun, I added dodge and parry to the scaling calculations. With the massive boost that they gave to Divine Riposte and the huge impact they have on crit, I figured they would be pretty impressive and might rank quite highly. Here's a test with regular old T17H gear (set bonuses enabled).

Image

They don't.

I don't understand these results though. Why does dodge perform worse than parry, when we naturally have more parry (thus making it more susceptible to diminishing returns)? Do dodge and parry have different rating to percentage or DR calculations? That would make sense. Either way, they both end up being the worst stats which I found surprising given their impact on crit and haste with Riposte and Divine Riposte.

Anyway, I tried Divine Riposte added to armor instead of parry this time around. The first test was just a flat 100% haste rating gets added to armor rating, which was hysterically overpowered. Haste ended up scoring as high as stamina. Even at 50% it was too much. After trying a bunch of different coefficients, I settled on 30% as the best balance. It reduces our TMI by 1.0k when compared to live and haste comes in second place, a bit behind mastery. This is for the T17H gear set.

Image
Source: https://www.dropbox.com/s/xszsh6r5pl0gtnh/t17h_30p_armor.html?dl=0

It's not as strong as Divine Riposte (Parry), but it's still something. Unfortunately, it doesn't hold up as our gear gets better. In the T17M sim, haste drops down to the bottom again (but it's still much better off than it is with live mechanics).

Image
Source: https://www.dropbox.com/s/3gnk4i6x3ynk71r/t17m_30p_armor.html?dl=0

So it looks like either dodge or parry is probably the way to go if the devs wanted to implement Divine Riposte. I think the choices of dodge, parry or armor made sense to test because we can't directly stack any of those stats. I don't like the idea of Divine Riposte converting haste to a different secondary that we can directly stack, like bonus armor, because that seems like a very weird interaction. Doesn't it just become a strictly better version of bonus armor at that point?

Anyway, with those results out of the way, I want to bring up something that Theck mentioned in his blog.

theckhd wrote:In any event, my point is that I think the solution to our haste “problem” lies not in a raw mechanics change, but in a subtle, “tack some extra value on to haste” change.


After running the sims for all of the ideas presented here that pass the giggle test, I agree with his point. Ideally there could be a new proc that only benefits haste but after thinking about it, I'm not really creative enough to come up with an interesting one (and I definitely wouldn't know how to go about balancing it). If anyone wants to propose one I am happy to implement and sim it, but until then, I think I've done all I can here. Divine Riposte (with dodge or parry) looks like the best option.

  • Assuming 1.3k TMI isn't a major buff, then it doesn't make us much more powerful than we already are (this can be offset by a carefully-targeted nerf elsewhere anyway)
  • It keeps haste as a valuable stat all the way up to mythic gear
  • It's very easy to implement, doesn't require a massive overhaul like a new melee-based proc would
  • It has precedent, because crit already does this exact same thing for parry via Riposte
  • It can be hidden from the spell book in the same way that Riposte is currently, to avoid confusing new players

At this point I've presented everything I sought to at the start of the thread, so I will stop until I hear back from you guys. I would love it if someone would double-check my data and either confirm or refute the results. Thanks for reading all of it so far, I hope you found it useful!
Ashaton
 
Posts: 55
Joined: Fri Nov 21, 2014 4:41 pm

Re: The Case for Haste

Postby Jackinthegreen » Sun Nov 30, 2014 2:01 am

Yes, dodge and parry have different DR rates. Take a look at http://www.sacredduty.net/2014/08/06/tc ... ns-in-wod/ to see what's going on there.

While we're at this though, do we have rough guesses for how items now scale based on item levels and slot? Both Blizzard and us players would likely prefer a long term and stable solution to stat scaling so figuring out how things might go into T18 and perhaps even T19 could be interesting, assuming there's enough support for it all.
Jackinthegreen
 
Posts: 55
Joined: Thu Aug 27, 2009 4:22 pm

Re: The Case for Haste

Postby Thels » Sun Nov 30, 2014 3:14 am

I was thinking the same thing. Haste and Mastery buff each other. The more Mastery we have, the stronger Haste becomes and vice versa. The other stats don't have this synergy. Part of the reason Haste is so weak right now, is because AM itself got nerfed. More Mastery will make Haste stronger again, except they just nerfed Mastery's effect on SotR.
ImageImage
User avatar
Thels
 
Posts: 1233
Joined: Tue May 15, 2007 7:30 am
Location: The Netherlands

Re: The Case for Haste

Postby Ashaton » Sun Nov 30, 2014 7:02 am

Jackinthegreen wrote:Yes, dodge and parry have different DR rates. Take a look at http://www.sacredduty.net/2014/08/06/tc ... ns-in-wod/ to see what's going on there.


Thank you for finding this, the results make more sense now. I wonder how the differences between them would factor into Divine Riposte for parry vs dodge.

Jackinthegreen wrote:While we're at this though, do we have rough guesses for how items now scale based on item levels and slot? Both Blizzard and us players would likely prefer a long term and stable solution to stat scaling so figuring out how things might go into T18 and perhaps even T19 could be interesting, assuming there's enough support for it all.

Thels wrote:I was thinking the same thing. Haste and Mastery buff each other. The more Mastery we have, the stronger Haste becomes and vice versa. The other stats don't have this synergy. Part of the reason Haste is so weak right now, is because AM itself got nerfed. More Mastery will make Haste stronger again, except they just nerfed Mastery's effect on SotR.


This is a good question. According to my earlier tests, T17M has 1885 mastery and 576 haste. T17H has 1676 mastery and 502 haste. That's a difference of 209 mastery and 74 haste. Despite the increase in both stats, haste's value dropped tremendously between these two sims - from 3.02 down to 1.74.

To pursue the question of scalability further, here's what I propose. I will look at 3 different gear sets: the current mastery-heavy set, a set with a balance of haste and mastery, and a haste-heavy set. I'll make iLvl equivalents for T17H and T17M, and my best guess at T18H, T18M, T19H and T19M. Obviously I can't account for new set bonuses or trinkets, but multiplying everything by some constant should give a rough estimate of the stats on improved gear. I'll compare the differences between iLvls of existing gear to figure out what the best multiplier is.

For all 18 of those gear sets, I'll run sims with live mechanics, then Divine Riposte (Parry) and if I have time I'll do Divine Riposte (Dodge) as well. That should be enough data to get a feel for how haste scales currently with gear, and how Divine Riposte does in the long run.

36 sims will take a while to run but I have a lot of hardware I can throw at them once I figure out the gear sets and scaling multipliers. Let me work those out and report back when I have something before I start the sims.
Ashaton
 
Posts: 55
Joined: Fri Nov 21, 2014 4:41 pm

Next

Return to Advanced Theorycraft and Calculations

Who is online

Users browsing this forum: No registered users and 1 guest

Who is online

In total there is 1 user online :: 0 registered, 0 hidden and 1 guest (based on users active over the past 5 minutes)
Most users ever online was 380 on Tue Oct 14, 2008 6:28 pm

Users browsing this forum: No registered users and 1 guest