Alright so an update: now that we have explored what builds are good, when they're good, and what legendary interactions we have therein, we will now examine at what point should we consider swapping builds.
For this next bit of analysis we will first define some terms:
Single-target fight: A single-target (ST) fight is a fight with only one target and no cleave components, DoT or clumped.
Cleave fight: A cleave fight is a fight where there are two or more targets stacked on top of each other.
Hybrid fight: A hybrid fight is a fight that combines single-target and cleave together. A hybrid fight contains x% cleave component, meaning that x% of the fight is modelled as a cleave fight, and (1-x)% is modelled as single-target. We allow for x to vary from 0% to 100%. We also allow number of targets in the cleave to vary from 2 to 4.
Now we will establish how we're going to conduct our investigation. The hard part is that SimCraft does not allow for the modelling of a Hybrid fight as defined above, so we need to use some interpolation techniques to model them. Let's first start with the simplest model: linear interpolation, a straight line.
But there's a problem with linear interpolation: we assume that each 10% of cleave uptime gives a static amount of DPS, which in this case is 10% of the DPS difference between simmed 2-target DPS and single-target DPS.
But we know this to be false -- because of AP pooling and delaying the use of cooldowns to line up with known cleave/burst phases, we can get more than 10% benefit from a fight with 10% cleave proportion. So, out goes the idea of linear interpolation.
Instead, we want a concave function, which allows us to model the fact that even small amounts of cleave in a fight can be abused. So we will do the rest of the modelling with nonlinear interpolation techniques (namely quadratic). Some of you may notice that it looks very similar to a diminishing returns (DR) graph -- that's because it is.
Now that we have the basis for our discussion, the rest of the post will look at the break-even points for five different builds, presented in some nice graphs. You can access the Excel spreadsheet that generated these graphs here: https://drive.google.com/file/d/0B3h...ew?usp=sharing
1. Base Case Analysis: No Legendaries
So a little bit about the graphs first: the vertical axis is DPS, the horizontal axis is the proportion of the fight that involves the cleave part. Each build is a separate (coloured) line on the graph, and the part that we're most interested in is where they cross over each other. This implies that the builds are equivalent in DPS for that Hybrid fight type.
In the first graph, we see that the two best builds (WoE/SotF/SS/SD and SL/Inc/BotA/SD) are even at about 20%. This means that for a fight with two targets for 20% of the time and one target for the remaining 80% these two builds will do roughly the same damage. Further to the right (ie. the longer the two target cleave is active), WoE/SotF/SS/SD will pull ahead significantly, but if you go left of the 20% then SL/Inc/BotA/SD is better.
The second and third graphs say the same story, but at around 15% and 10% cleave proportion, respectively. This implies that the best cleave build (WoE/SotF/SS/SD, as expected) scales fairly well with the amount of targets. Noticeably, the builds have separated significantly by the time you have three-targets up for 100% of the fight. SL/Inc/SS/SD is always the second best build by 100% uptime, and the gap between it and SL/Inc/BotA/SD seems to widen while WoE/SotF/SS/SD stays fairly constant. This implies that SL/Inc/SS/SD is a consistent build that scales just as well if not better than its competitors.
Overall, I'm surprised by how early the cleave-style builds take over. I would have eyeballed a 50% uptime on cleave for the cleave builds to dominate, but this is certainly implying as little as 10% is enough with 4+ targets.
2. The Emerald Dreamcatcher Analysis
The Emerald Dreamcatcher graphs tell a slightly different story. It takes a higher proportion of cleave in the two-target case for the cleave builds to take over, at about 30%. This continues for the other two graphs as well: about 25% for three-target and 20% for two-target. This implies that Boomkins with ED should be swapping to cleave builds less often than Boomkins without ED, as the benefit of being able to use ED and surgeweaving during the single-target components of the fight is a significant DPS increase.
Importantly, SD seems to be a core talent even for ED users. You lose the infinite DoTs, but you can always refresh after surgeweaving has ended. Picking up SD allows you to benefit on cleave portions of fights, and SS is allowing for more surgweaving to happen (on 3+ targets it's generating more AP on average than BotA).
3. Impeccable Fel Essence Analysis
For IFE users we would expect that Incarnation builds should remain relevant for longer, and this is what we observe in the data. Like with the ED scenario, the break-even points for the 2, 3 and 4 target cleave graphs are approximately 30%, 20% and 15%, respectively. IFE users benefit from Incarnation moreso than other Boomkins, so considering a cleave-oriented build with Incarnation instead of SotF is a strong option. The gap between the top-performing build (WoE/SotF/SS/SD) at 100% 4-targets and the best build with Incarnation (SL/Inc/SS/SD) is relatively smaller than in the no legendary or ED scenarios. However, CA is still a strong cooldown and benefits from AP pooling quite significantly, which is shown in the WoE/SotF/SS/SD having higher DPS than in the other two sections above.
4. Oneth's Intuition Analysis
For OI, it seems the break-even points are similar to that of having no legendary at all -- about 20%, 18% and 15%. What's interesting about OI is that it doesn't seem to really interact with either Incarnation or SotF much at all. Instead, it interacts with SD much more: OI Boomkins are the only Boomkins where the SL/Inc/BotA/NB and SL/Inc/BotA/SD builds are almost equivalent for ST. This means that SD builds take over NB builds much quicker than non-OI Boomkin builds, and so it's understandable that the break-even points are similar to non-legendary Boomkins'.
Overall, what to take from this analysis? Well there's a few things that pop out:
- SD is very strong, and it is one of the reasons why we're doing so well at cleave in 7.1.5.
- SotF is incredibly strong, but you need to pool AP and use it correctly to get maximum benefit from it in a fight where the cleave is only for a small percentage of the time.
- It seems that there are some magic numbers: 30% for ED or IFE users and 20% otherwise. If you feel that you're cleaving for more than 30%(20%) of a fight, it's likely that a cleave-oriented build is your best option. Any more than that and you'll see some really good DPS.
- We only explored up to 4 targets in this analysis. If you encounter a fight where there's 10 targets up for 10% of the time, you can bet on a cleave build being nuts on it.
- WoE/SotF/SS/SD and SL/Inc/SS/SD are two very strong builds, both with their pros and cons.
- Finally, caution should always be advised. Some cleave will be considered padding and not useful (see: Mythic Botanists), and some cleave will not get the same maximum DPS as seen in the simulations (due to the mobs being spread out too much, or movement, or both). As such, you should always critically think about the fight before committing to a build -- experiment! This analysis was just to help you on your way to making the right decision.
Thanks for reading again! I will be analysing Nighthold trinkets next, so stay tuned. I should have the post up within the week.