Tuesday, March 22, 2016

A Random Update: My Blog, not the Alpha.

Hey guys! First things first, I want to again thank Chaithi for inviting me onto Monk Meditation last night. I had a wonderful time. I'm actually writing this post in part because of that podcast, as I had forgotten to mention something during the Brewmaster alpha news bit. There's also been some interesting changes to brew generation that I will be talking about in a little bit. So without further ado, let's dive right in!

So the latest update to the alpha gave us very little in the way of new stuff to think about and tear into. Relatively. Gift of the Ox now scales with attack power rather than health, Obstinate Determination - which in the previous update was changed so GotO overhealing provided a shield - now guarantees a GotO orb spawn when you drop below 35% health. We also got a new passive, Celestial Fortune, which provides a percent chance equal to our crit to duplicate any heal on you.

To basically reiterate what I said on the podcast, the GotO change is... honestly kinda expected. Overhealing on GotO is VERY easily cheesable so changing it to where GotO is more reliable at spawning when you need it to is a healthy change design-wise. Changing to AP vs health isn't really that big a deal, it just means it scales a bit differently. I still think that GotO as our only emergency ability apart from hard CD's is a bit questionable for players who don't really get the ISB and PB balance and upkeep, as there's more steps needed to save your bacon (noticing health is low and pressing a button (guard) vs noticing health is low, looking around for your orb, and moving over to it). But a lot of that is very much tied into the damage intake model and whether punishing poor play like that is a good idea for a specific difficulty of content, all of which is something that could honestly fill a post on its own. The big thing I want to talk about is what I didn't actually bring up too much on the podcast, and that is Celestial Fortune.

The Brewmaster community at large has really taken to pointing out the similarity of Celestial Fortune to Shining Protector for paladins, only that one is based off of multistrike and not critical hit chance. A lot of people point to how this passive is just complete trash for Prot Pallies and how multistrike is their worst stat. That being said there's a couple key differences between the two.

Firstly, the amount of base crit we have is much higher than base multistike (I believe it's 20% vs 5%) so already we're going to see more use out of this at baseline. Secondly, the potential throughput is also much higher. Shining Protector, when procced, provides 30% additional healing from whatever heal triggered it, including multistrike heals. Celestial Fortune on the other hand effectively provides a secondary stacking source of crit. If a heal crits on you and procs Celestial Fortune, that should effectively mean that crit has crit. Thirdly, the nature of damage intake is going to be very different. Protection Paladins have to deal with a lot of burst damage right now, which makes their passive less appealing because they need to focus more into stats which actually provide solid burst reduction. In Legion, Brewmasters have ludicrous amounts of stagger which means that we should be very smooth on damage intake, but because we lack baseline damage reduction, we'll be taking a lot more straight damage across a fight. This is actually a perfect environment for this kind of passive as over a fight, it makes healing much more efficient... and we're already leaning a bit more towards being friendly for mana-efficient healing. This just makes that even nicer.

The other thing I wanted to talk about was something Hinalover brought to my attention this morning. Namely that we did actually get some haste CDR! But only for Keg Smash. So I decided to model out how that'd look in my usual way (+10% haste rating and seeing the difference of charges generated across 10 minutes) and found some very interesting data as well as some worrying data.

So this is the table from my last post's discussion on haste scaling with our brew generation. Below is the table now, factoring in Keg Smash's CDR.

At first glance, it seems pretty good doesn't it. The baseline amount that haste provides across 10 minutes has roughly doubled. The trouble comes from its scaling. The things that bump up our charge generation all focus on Tiger Palm as the medium for the change. Because while Keg Smash is primarily operating off of its cooldown, it still shares a resource with Tiger Palm. More KS casts inherently means less TP casts as well. This is actually the core problem of Secret Ingredients... because Face Palm and our 4-set both make Tiger Palm more lucrative, KS has an increasingly high opportunity cost. Specifically for this scenario, it appears as though this change starts to trend negative once you get your 4-set and have 4+ ranks in FP.

That's a complicated scenario because on one hand, despite this change actually causing an overall negative in brew generation over time, if you look at the one GCD, Keg Smash still provides more brew recharge than one Tiger Palm. But that does go away fairly quickly the more seconds you look at. Another factor is that this again only negatively affects you when you have 4+ ranks in FP and also have the 4-set. It's unlikely that people will be using relics to boost their FP rank because of other more lucrative choices out there. A lot of the problems with the change are that at first glance, you think it's positive. "Less CD on our Keg Smash, our biggest brew generator? Sign me up!" But the truth is really hidden under the inner workings of the mechanics and the relationships therein. That lends itself to the design of the spec causing miscommunication between the designers and the players.

But this ultimately brings up design philosophy and from my personal experience with design and my own philosophies, I'd rather design something so that even if it's not an optimal choice, it's not actively punishing you if you do it. Even if the scenario where this happens is pretty rare, I would prefer to take steps to not let it happen at all if I can't help it. I'm not comfortable assuming that players will just get it, especially when I have experience with Brewmasters in particular who have had trap abilities, trap talents, which have confused a lot of people in regards to why they aren't doing well.

So what is to be done about it? There's a couple things that I think Blizzard could do. For one, either have one or both of FP / 4set affect Keg Smash in some way so the biggest factors to brew generation increases scale with it as well, For another, just have haste provide CDR to the brew charges. I could see them bump up the base CD's of them by a little bit just to be on the safe side for scaling, but even adding a couple seconds on should still result in an overall positive change, especially because haste would become very lucrative in that case. Another idea would be to actually reduce the energy cost of Keg Smash to 25, same as Tiger Palm. The question with that one though is whether there'd be some scaling issues across energy regen, because higher energy regen does favor TP. At the same time, the core issue of the present is that Keg Smash just becomes less efficient in terms of CDR per energy as you get more TP perks, so having it be the same cost as TP would get rid of that. I think further modeling on that idea would be necessary.

In any case, thanks for reading everybody! Feel free to comment below or hit me up on twitter if you want to further discuss the topics above. See ya next time!

~ The Brewing Scribe

Tuesday, March 1, 2016

A Hasty Addendum: Factoring in Artifacts and 4-Set

Okay so... remember when I said that the numbers I posted on the previous blog post were a lowball estimation?

Yeah, I decided to get in some more work on the spreadsheet and they were indeed a fairly lowball estimation.

My process here is basically to multiply the number of tiger palms by 1 or 2 (depending on whether or not the 4-set is marked as active) plus the percentage chance to proc an extra second of cooldown reduction via Face Palm (1 times (% chance) = % chance)). And that kinda really started to escalate pretty wildly. See the nice tables below! Like the original thought experiment, these show how many more charges you would get with 10% haste than with zero haste across 10 minutes, but now factoring in Face Palm and the 4-set bonus alongside that additional 10% haste. So in other words, the benefits of getting all those traits alongside the haste increase.

So already, we're seeing a VAST improvement here, even still at 10% haste. These values are not accounting for haste also providing straight CDR on our brew charges. With just the 4-set, we're already higher than just haste CDR. So let's see how this works out with haste CDR:

So that's actually a lot. Because I used my spreadsheet to get these numbers, I was also able to see the total number of charges you are able to use across that 10 minute fight with these variables. For Light Brewing, with 3 ranks in Face Palm, the 4-set, and haste providing CDR, you would generate 158.36 charges with perfect play.

So remember when I said that 100 charges meant 100% uptime on ISB? This means that you'd have 58 purifies across 10 minutes... which is actually not that bad, at once every 10.3 seconds. Assuming you aren't tank swapping, that is. Still, that's insane, and because everything likes to compound on itself as you get more haste, that's going to rise if you get even more haste. it might become a bit... insane. The only problem is that this is a very extreme scenario which while it shows just how powerful our artifact traits and 4 set are alongside haste, doesn't show us a good representation of how haste works when those perks are a constant, while changing how much haste there is.

Those are both the tables for haste with CDR scaling and without, and we're seeing a much more consistent (and reasonable) picture here. To give some perspective, without haste CDR, but with 3 ranks in Face Palm and the 4-set, Light Brewing provides a max of 144.30 charges across 10 minutes (or full ISB coverage + 44 purifies). Compared to with haste CDR and the same parameters, Light Brewing provides the aforementioned 158.36 charges (full ISB coverage + 58 purifies). So adding haste scaling will, for most higher end Brewmasters, provide a max of 14 additional charges over a 10 minute fight, which reduces the average number of purifies from once every 13.6 seconds to once every 10.3 seconds. So it's a pretty significant benefit, though not insanely so.

But let's also think about scaling beyond 10% haste. If we still maintain that haste will provide CDR, going from 10% haste to 20% haste is another additional 21.87 charges with LB, 3 FP, and 4-set, bringing you to a total of 180.2 charges over 10 minutes... which is a bit insane.

Doing so without haste providing CDR but otherwise with the same parameters will only provide an additional 6.53 charges, getting up to 150.83 generated in total, which sounds far more manageable. Of course, to get to 20% haste, that's in live about 1800 haste rating, which is a good deal.

It's hard to decide what would work best... haste providing CDR means that it's MUCH more effective per additional charge gained, making it more lucrative, but it could make the spec super difficult to play (again, this is a VERY hard thing to truly predict without actually playing the spec) or actually overpowered, with the silly amount of charges that are possible for you to generate. On the other hand, without haste providing CDR, with 4 set and 3 FP, you'd still need about 1300 haste rating (at 100) with LB in order to get an additional charge per minute (9/2.45 = 3.673 TP's, 3.673*25 = 91.83 energy, 91.83/60 = 1.53 Er/S, 1.53 *900 = 1377.55 haste rating), which would make haste extremely inefficient. It really does need that scaling to be effective.

If I had to pick either or... I'd go with haste scaling. It might be chaotic, but at least haste would finally see a good use.

Also, as a request by LeBlue: "NEED BREWMASTERS GIF NAOW THX"

See ya next time!


Don't be Hasty: A look at Haste and Brew Generation in Legion (Thus Far)

Hey everybody! So as some of you guys know, I've been working on updating my old Brewmaster Calculations spreadsheet for Legion, which has inevitably gotten me to model out how brew generation works with the different easy to model talents (based on LeBlue's napkin math). So today, I figured it would be cool to go into a bit more detail about them and talk a bit about how the model currently works as far as the current version works in the tooltips in Alpha.

So as of right now, over a 10 minute fight at base regen if you include charges that you have at the start and your starting energy, if you play 100% perfectly you should be able to get 76.31 charges without a talent, 118.04 charges with Light Brewing, and 108.89 charges with Black Ox Brew. I'm not going to talk about Secret Ingredients in terms of this direct comparison because according to the latest script work of the lovely Xiaojin, it's even worse than no talent for brew generation. We'll be talking about this later. Keep in mind though that those numbers I just threw out there were ignoring the artifact bonuses and 4 set. I still need to factor those into the spreadsheet, but they suffice for now. One other little note is that over a 10 minute fight, you need 100 charges to permanently maintain ISB. So just bear that in mind.

Back at the beginning of the Alpha when all we had was essentially the class previews to go off of, it was speculated that haste would finally become a useful stat for us. There's been some relatively recent arguments that such is still the case. Well, part of why I thought this was a good time to talk about haste was that this... really isn't that accurate.

With an additional 1 energy per second... you get a grand total of 1.71 additional charges over a 10 minute fight with no talent and BoB. 2.67 with LB. Which is to say, any real significant improvement from haste would require thousands of haste rating. Keep in mind again that this isn't considering artifact traits... but still, that's very low, across 10 minutes.

This then begs the question "Well hey, so this is bad. Does it matter at all though?" and I would say yeah, it kinda does. Blizzard has put haste onto our artifact, so I would have to assume that this means they want haste to be useful for us. There's some natural scaling, it's just that because lots of haste is required for a small amount of energy regen, a good chunk of energy is required for one tiger palm, and several tiger palms are needed for one brew charge, the efficiency of haste to brew charges is severely diminished. But, let's do a quick experiment to see how changing the cooldown of the brews would improve haste's usefulness!

Just to keep things simple we're going to go with 10% haste, ignoring artifact and 4 piece, since that's the same thing as 11 er/s, our previous benchmark for haste benefit.

So using the formula CD = BaseCD/(1+(haste%/100)), we get X1 = 14/1.1 and X2 = 9/1.1 for no talent / BoB and LB respectively. X1 becomes 12.73 and X2 is 8.18.

To not bore you with the math, my spreadsheet essentially calculated that at 10% haste affecting cooldowns, sans any modifiers other than the talents, you get 9.22 additional stacks over 10 minutes with no talent, 14.34 with LB, and 9.47 with BoB. So yeah, having these CD's scale with haste actually really dramatically improves its usefulness.With current ratings values at 100, you still need a couple hundred in order to get a single additional charge per minute, but it's far better than the couple thousand it is without. Furthermore, you have to bear in mind that anything that improves the charge efficiency of tiger palm will also significantly buff haste's value to Brewmasters. So in reality, what is here is in fact a low estimate.

Edit: So I actually did do some math and modeling on this, find it here!

That's actually super great, overall, with some caveats. For one thing, and this is a bit of a potential worry that might not actually be one at all, valuing haste greatly might mean that the spec feels more frantic. Because we no longer have a filler ability, we can't just use that less in favor of more energy-spending abilities. A lot of our core abilities are now on short cooldowns, which means we need to be aware of those while still balancing energy. I do want to stress though that this is a potential worry that requires actually playing with the spec at varying haste levels to determine its reasonability.

The other problem (which segues into the next part of this post) is that prior to the recent patch which further reduced the CD on our brew charges, BoB was the superior choice for brew generation. Which made sense because it was an activatable ability. Mastering that, a more challenging ability, should give the player a higher payoff. As of the latest alpha build, that is no longer the case. The passive LB actually provides superior charge generation. Adding haste scaling widens that gap, as BoB barely scales at all, while LB's effectiveness is compounded by it.

This, combined with the results of Xiaojin's work for SI, makes me feel that the entire tier of talents needs to be reworked, even ignoring the exacerbation that haste scaling for CD's would provide. Balancing is all kinds of crazy... completely understandably, mind you. Blizzard was pretty clearly trying to get the base Brewmaster to look good before tackling talents, and if you don't change talents while you change core mechanics, you're inevitably going to get some unintended balancing snafus. But ultimately you have two passives... one which is extremely effective and one that is worse for brew generation than no talent at all, and one active that is good... but is too much work for relatively little payoff.

So what could be done about it? Well again I posit that haste affect the cooldown of our brews just so that there is actually a use for haste for once, but change SI so that instead of refreshing the CD on Keg Smash, you cause the next one to be free and proc extra brew. This means it doubly scales with haste, just as LB would. Maybe give KS a bit more damage, with a bit less CDR, so that SI becomes a stronger choice for higher damage. Secondly, keep LB more or less the way it is, perhaps ease up a little bit on the CDR, to keep it competitive but still the go-to for just raw brew throughput. Lastly, redesign BoB so that it maintains a higher skill-usage, but perhaps for a more specialized situation. Maybe have it so that for some number of seconds, PB doesn't use up brew charges. Give it a decent CD and it becomes a higher skill-use ability to help deal with larger hits. Or have it so that it instantly regenerates all charges, but then for 20 or so seconds after, PB doesn't have a charge cap, so it scales with haste a bit more. In either case, the changes suggested would give that tier three good choices depending on your situation: one for high brew throughput across a fight, one for extra damage, and one for better management of burst. Each has some tuning knobs so that you can ensure that they are fairly even, providing some defense but having their own advantages.

Thank you guys once again for reading! I'd like to once again point out that this is all discussion based on the current alpha build for Legion and thus could all easily become obsolete within the next couple weeks or months! As always, if you want to discuss anything you read here, feel free to post a comment below or tweet at me. Until next time!