This post looks at what DPS Bot has to say about how we perform on this tier’s tank and spank, Ultraxion.
DPS Bot’s rolling two-week sampling window has exited the period of post-patch volatility for Ultraxion, and so it now seems reasonable to take a look at how the different DPS specs are stacking up on the normal modes for this boss. Why Ultraxion? He is the closest thing to a monotonous, egalitarian tank and spank that we have this tier. Why only normal mode? The data still looks a bit too messy on the heroic side of things.
Specifically, this post will look at the rolling median DPS measures of the top 100 parses on normal modes of Ultraxion for the two weeks preceding Dec. 26. I am using median DPS because it is not as easily influenced by the magnitude of extreme outliers as averages are. I have selected the top 100 parses, as opposed to all parses, because what I am concerned with here is class potential (found in the top parses) rather than where the average player stands. I am using a sample period of two weeks because it is the largest sample that DPS Bot allows and gives the best chance for good parses to filter into underrepresented specs like BM.
Before we get into the numbers, there are some general disclaimers that I should go through. For each hunter spec to have an equal shot at a good parse, we would want players to be randomly selected into them so that, overall, each spec would have roughly the same proportions of talented players and theorycrafting interest. But that is not the way the game works. MM has been the favored hunter spec of the expansion and BM has struggled to both keep up and shed its Wrath-era reputation of non-viability. We see this disparity in player preference reflected in the simple number of samples brought in by DPS Bot. Over the last two weeks for normal 25-man raids, World of Logs held 93,987 usable parses for MM, 67,169 for SV and only 12,387 for BM. With that kind of representation the odds are simply against BM when it comes to getting an equal share at theorycrafting attention, let alone an equal shot of nabbing the best hunters. Only sub rogues and frost mages saw less representation.
There is also skewing relating to in-game advantages. Both casters and rogues have powerful legendaries (or pre-legendaries) these days, and the top parses are likely to be filled with players using them. What this means is that comparisons of hunters with rogues and casters are not equalized in terms of gear.
We also cannot be confident that some other selection effects are not going on. Not all raids upload to World of Logs, and we cannot be certain that it is mere randomness that separates those who do and those who do not. There is also selection internal to WoL and DPS Bot. How effectively are erroneous parses being removed? Does DPS Bot still cut off the top 5% of parses to avoid extreme outliers?
Despite having these and other potential concerns, I think we should still press ahead and look at the data. No data is perfect and most of the time the concern is how best to accommodate the imperfections rather than correct them. In the task at hand, the best we can do I think is allow that the data is not ideal and keep its biases in mind as we analyze it.
The Data
In the following tables, “Median DPS” is a spec’s median DPS in the top 100 samples on Ultraxion taken from the two weeks preceding Dec. 26. Rank is where this median DPS is positioned among the 22 different DPS specs.
MSS is the the spec’s spec score based on median dps; that is, it is its median DPS as a percentage of the highest spec’s median DPS on Ultraxion. Unless a spec is the top spec it will not score 100% on this measure. Spec score is a way of measuring a spec against the top DPS (as we tend to do ourselves in game).
FS is the spec’s `Frostheim score.’ It is the spec’s median DPS as a percentage of the median spec’s median DPS. Put another way, FS is how close a spec is to the middle of the pack. Percentages over 100 represent how far above the middle the spec is. Percentages below 100 represent how close a spec gets to (but still falls short of) the middle. The Frostheim score (as I’m calling it) is a way of measuring how divergent a spec is from where we might assume developers might want all specs to be.
I see value in both the MSS and FS measures. One mirrors what players think in the game, that is, they (and their raid leaders) look to how far away from the top they are. The other reflects what we can assume to be a developer goal, that of identifying how divergent from the norm a spec is. Neither score is a be-all and end-all measure and neither should be viewed exclusively. However, they each do a reasonable job reflecting a ratio that players have reason to find meaningful.
Ultraxion N25
| Median DPS | Rank | MSS | FS | |
| MM | 38,628 | 17 | 79% | 95% |
| SV | 39,571 | 16 | 81% | 97% |
| BM | 34,573 | 21 | 71% | 85% |
This table is ordered by the current implicit ranking of specs in player preference: MM first, followed by SV and BM. As we can see, the table upends this presumption a bit, at least for this difficulty level. SV pulls ahead of MM slightly! This lead appears in the heroic parses as well. Whether it will hold up in the coming weeks, well, only time will tell.
Unfortunately, the overall rankings are not that promising for hunters. We do not reach higher than 81% of the top spec’s DPS and that does not bode well for how hunters will look to raid leaders who compare everyone to the top of the meters when considering how to fill raid slots. However, we should keep in mind that not many specs are looking particularly good on Ultraxion spec scores these days because the top spec is something of an outlier. Fire mages are out above the pack, so to speak, with an FS value of 119% (which is high, higher than BM is low) and the next highest spec only getting a 94% spec score on normal 25man. If we decide that fire mages are too much of an outlier and have the spec score refer to the next highest spec instead, hunter spec MSS values look a bit nicer: 86% for MM, 88% for SV, and 77% for BM. Most specs are above 90% on the adjusted measure.
FS measures suggest that hunters are not stragglers so much merely below average. MM and SV hunters are close to the middle of the pack. BM looks a bit worse than MM or SV on both the MSS and FS measures, but we should remember the reasons mentioned above as to why BM might be lagging. BM might have more potential to it than is suggested by the data here.
Ultraxion N10
| Median DPS | Rank | MSS | FS | |
| MM | 35,106 | 17 | 80% | 95% |
| SV | 37,561 | 9 | 86% | 101% |
| BM | 33,488 | 20 | 77% | 90% |
This table reflecting Ultraxion 10man is a bit more promising than the 25man table, showing one spec (SV) making it into the top half of DPS specs. BM also looks better here, nabbing a FS value of 90%. However, the hunter specs are in general not standing out for their DPS. Note that here again Fire mages are ahead of the pack and, in adjusted MSS values, MM would be 85%, SV would be 91%, and BM would be 82%. Even 90, 82 and 85 percent, though, do not look good when most specs are at 90% or higher.
These tables are a bit disconcerting overall, especially considering how little raid utility and survivability we bring to The Dragon Soul. There are few mechanics (none for bosses that I can think of) that cause a raid to distinctly benefit from having a hunter, and there are some (such as on the Spine) where the raid actually hurts for having a hunter. Hunters in general tend to take more damage than all other dps specs this tier owing to our lack of survivability options (we don’t get to Raptor Strike much and Deterrence does not work on many sources of damage). This leaves me thinking that hunters are not looking particularly attractive these days, not when another ranged DPS of equal skill could do more damage, live longer and potentially offer more raid utility. Of course, this is all speaking in terms of averages. Player competence trumps these sorts of odds, and so it it behooves us to have more skill and know our jobs better than other ranged DPS so that we earn a spot despite whatever slight numerical disadvantages we might be operating under.
Euripides
jurgwena
lemli
eidotrope
pherra
femaledwarf
Shadowstreak
Very nice analysis, thanks for putting this together!
What might be interesting is to compare this to a “theoretical” target dummy fight, using Femaledwarf.com. If that were to show say, BM doing significantly better relative to MM and SV, it would strengthen the case you’re making that BM may actually perform much better in the hands of a skilled hunter than these parses indicate.
I think that would require that all specs have the same baseline ratio of attainable to theoretical dps. I like your idea, but I’m just not sure how comparable the specs are.
This has come up before, but to reiterate, the top 100 is BS. If you want to look at elite players on Ultraxion 25N, use all parses at the 90th percentile, as per this link.
http://raidbots.com/dpsbot/Ultraxion/25N/all/14/30/p90/
To expand a bit (I know I posted this on your WOW insider pieces, but maybe not on your blog) the main problem with the top100 is its miniscule sample size. It’s the top 100 parses per spec. If you select “samples” for Ultraxion 25N, you’ll see there are 7,850 survival samples listed. This is a much more significant data set.
I can’t think of any scenario where the top 100 should be used rather than high percentile all parses. It honestly shouldn’t exist; raidbots maintains it because people got used to using it during the stateofdps days.
Also, if you want to focus on where hunters fall behind, look at 25H. Survival is perfectly fine in 10N, 10H, and 25N. It’s terrible in 25H, and it’s not entirely clear why. Raidbots only has 127 survival samples for ultraxion 25H, so the content isn’t being fully explored yet, but it’s not looking great.
To clarify, I don’t write for WoW Insider. That’s Frostheim.
I did not even know to look for the 90th percentile option since it was not there when I did my initial poking around at DPS Bot. I will keep it in mind for the next time that I write one of these posts.
Ack, sorry! My bad on the mixup there.
Yes, the higher percentiles are very useful. The 90th represents seriously elite players, particularly on heroic modes, and a sufficiently large sample size to be worthwhile talking about.
Nice job Eido. I have to say its disheartening to see classes with a legendary (even a nerfed one at this point) able to out dps me simply because of gear. Its very nerve wracking that a raid leader would make the call, mainly because I cannot blame them! Who wouldn’t make the decision to simply take 1 hunter to your raid, and then have all mages, spriests, boomkins, etc.
This being said, the devs have a knack for balancing things out eventually. Perhaps our gear upgrades across T13 will be equalizers? Here’s hoping.
T13 gear is designed to buff sv over mm, with the result that mm is effectively and seriously nerfed (us buffing everyone elses’ dps while we get nothing only adds insult to injury).
So the ONLY reason a raid would want an mm hunter now is because of trueshot aura giving the entire raid except that hunter a 10% dps buff, at the cost of having one chronically nerfed player in the group who also needs a relatively high amount of healing to stay alive (and thus keep the buff actve).
Testing my dps with 4.2 and 4.3 justice gear (which is what most starting raiders will be wearing, with maybe 1-2 valor pieces), my dps in 4.3 (thus 378 ilvl) gear is actually lower than it was with the equivalent pieces of 4.2 (359 ilvl) gear when specced as mm.
To say we’ve not been nerfed is therefore incorrect. It may not be a direct nerf in as “we’ve reduced the damage of arcane shot by 10%”, but gear stats (and that’s true for all 4.3 hunter gear I’ve found so far) being such that despite higher ilvl and agility your dps is lower than it was in older gear means we have indeed been nerfed (that 10% is about the nerf I notice).
This is in part compensated by the 378 bow that drops in the new dungeons, and some of the valor gear (especially the cloak), but until you get that you’re looking at increased stats leading to decreased dps.
`Frostheim score’, seriously? The Frosthiem worship is getting a little weird. You have officially overreached.
I don’t worship the guy, far from it, but I believe in giving credit where it’s due. He proposed the measure, so when I used it I gave him credit by applying his name to it. How that’s overreach (and offical overreach at that), I have no idea.
So, someone with a name that’s obviously a reference to Frostheim (the phrase “the greatest form of flattery is mimicry” comes to mind here and isn’t flattery just another way to worship someone?) is criticizing a blogger who gave credit to where credit was due to Frostheim, eidotrope’s fellow hunter blogger and friend, because he believes that blogger is too worshiping and overreaching? The irony here is so thick, I can just taste it in the air. Also, your use of the term “officially” implies that you have any authority or experience here, which quite frankly makes me chuckle at your ignorance and/or naivete.
Thankyou for this. Its concerning.
Hunters in PVE are heading towards the same position as they are in PVP.
They can do an OK job but bring no special abilities and have disadvantages that mean you can replace them with other classes and be better off.