I am still ranked 16th on shannox hc with 50k dps, bugged files are bugged!
i5 [email protected] -8gb-7970-Dell u2711-2xM4-WCool next
Last edited by Keosen; 2011-07-28 at 08:09 AM.
But missing one MB means your Empowered Shadow might fall off.
Its probably situational, but on yesterdays raid I went down from 16.7% hit to 15.12% and ended up doing about 2k more dps then usual (granted I switched the shitty TB trinket with Moonwell Chalice so I went up in int quite a bit) with maybe 1-2 misses in general on every fight (Ryolith, Alys, Baleroc and Staghelm). So in general hit doesnt seem so important if you're constantly aware of your DoTs (as you should be) and can quickly reapply them.
Now just need baleroc to drop Necromantic Focus and I'll be a very happy shadow priest.
monster 4pc crits make me wet there is noway id go under hitcap with 4pc
One thing to consider is DPS spread. Historically, SPs have had an extremely narrow spread b/c they always hit capped and stacked haste which minimized the RNG factor on our DPS. As such the width of our DPS bell curve has always been very narrow compared to other DPS that rely on crit and don't hit cap as aggressively. Cata changed things significantly with orbs being fairly RNG, making mastery a broad ranged stat.
Not being hit capped further increases our DPS standard deviation, all the priest high rankings will be from the times those individual shadow priests are on the + side of their bell curve. They'll vulnerable to the - side of it just as often. At the very top of the, statistically you'll tend to see SPs at their 2nd/3rd standard deviation.
It's definitely open to a lot more subjective opinion and preference, but don't over read the WoLs . Personally, I consider reliable DPS on demand a plus of being a shadow priest. I take comfort in knowing a MB will hit. Others will prefer the larger spread in their DPS and ^.^ at the high points. /shrug
edit -- I'm currently 0.01% below the cap, which I personally deem more than acceptable.
Last edited by Themos; 2011-09-02 at 05:33 PM.
What do you mean with 2nd/3rd standard deviation?
Personally, anything below 15% hit is just annoying to play with. It feels like the point where I get enough misses for it to affect my DPS in a negative way compared to the DPS I could gain by sacrificing that hit. I would love to be at 16-16,5% hit but sacrificing Haste for it is not something I will do. Ever.
Before mastery and orbs were introduced, the only RNGish thing we really had to worry about was hit, and that was conquered easily by capping it. Everything else could be mathed out, and variations in DPS were due to movement, mechanics, gear, and skill. Everything was relative to the player and the raid, not inherent in the class. Thus, if you had a particular player with BiS gear on Patchwerk with a stable net connection, you'd have very little difference in min/max DPS as long as they play correctly (priorities and such). That's what he means when he refers to the "DPS spread".
Now consider Cata. Our orbs are pure RNG; we can increase our chances of proccing an orb with talents (18%) and increased haste (more MF and SW: P ticks), but it's still chance. We can also increase the bonus we get from orbs by stacking mastery, but it still relies on the chance of an orb proccing. It's not something we can cap like hit, so in the same situation above (BiS, patchwerk, etc), the spread of DPS varies a lot more than it would have with Wrath spriest mechanics.
So, regardless of what we do, DPS for even the most perfectly geared, talented, secure-net spriest will vary more than it has in wow history. Not by an absurd amount (you won't see a good spriest suddenly plummet below tanks for no reason), but still something that isn't ignorable. It's why we have to pay attention to orbs.
Standard Deviation is a statistical term that Themos is using, though I'm not savvy enough with simcraft to know if it applies so well to this case. Basically, within the spread, you're more likely to get a result that's average than one that's extreme. SD's are divisions of a spread that help you point out how likely it is to get an extreme result in the spread. Getting a result within the 1st SD is a 96% chance (assuming a normal curve, which is why I'm not sure it applies so well). A 3rd SD result is extremely rare, it's about a 1% chance.
Basically, the more things you can control about spriest DPS, the smaller the spread; the extremes aren't so extreme. It's trading the minute possible throughput (2/3/4 SD bonus DPS) for the certainty that you won't get the same minute possibilities of low DPS (-2/-3/-4 SD DPS). If you want to think about it from a healing perspective, it's kind of like trading Int for Crit. Int gives you a surefire increase to healing, but crit gives you a better chance at doubling the heal.
Standard Deviations are basically the mathy way to measure all that. I haven't played enough with Simcraft to be able to see if it actually forms a curve where that would make sense though. I have doubts that hit cap makes as big a difference for control now as it did in Wrath just because orbs mess with it.
The point Themos was trying to make was not that hit is the sole thing affecting the spread of DPS by those who rank; he was pointing out that all of them are on the good side of the spread. They didn't just get lucky with hits, they got lucky with orbs, mechanics, movement, and everything else. While no one thing can make a huge difference alone, all of those things combined can make a significant difference in DPS. Themos prefers more control because of how many things can go wrong, so he prefers at least controlling hit so he knows his spells will hit when he wants them to. He wasn't bashing anyone who doesn't cap.
I agree with Ariadne to a point. I believe dropping hit will affect dps in the same way that dropping haste, crit, or mastery will affect dps. Aside from the fact that it can be capped, it's basically just like any other stat. Having less of it lowers dps by X%, having more of it raises dps by X%. It's just a stat that affects DPS, like any other, and small changes in it only result in small changes to DPS.
As Ariadne says, this type of thing won't show up on WoL in a significant enough way to allow us to draw some kind conclusion. In general I suspect the only real conclutions we can draw from WoL data are things like this:
1. Better gear does better dps.
2. Poorly reforged gear does worse dps.
3. Poorly itemized gear does worse dps.
But in general two priests with virtually identical gear, one who has reforged a small percentage of hit to some other stat, another who has not, they are both going to do about the same amount of DPS. And both will be capable of ranking on WoL with the same consistency and potential.
u can rock 16% and be just fine.
You can rock -2% Mastery for the same exact mathematical reason and be "just fine".
Standard deviation is indeed a statistics term. You can look at the plot in the upper right corner of the wiki page.
In a nutshell, when you have a bell curve (it most certainly doesn't need to be perfect) various calculations are used to understand the data.
The 'number average' is the most popular form and works really well when the data is symmetric. But it doesn't tell you anything about the width of the data. And that's where 'standard deviation' comes from. You measure how fat a distribution is by splitting into pre-defined % values and seeing how far away from the number average a specific sd lies.
1 sd = 68.2% of the data closest to the 'number average' evenly split. +34.1% and -34.1%
2 sd = 27.4% further out from 1st sd, again evenly split. The top end is +13.6% and the low end is-13.6%
3 sd = 4.2% further out from there, again evenly split. +2.1% and -2.1%
Accounting for the first three standard deviations corresponds for 98.2% of the total data generally speaking is usually enough. You can go further out, but that far out, it's statistically irrelevant to the number average.
There's a lot more to statistically analyzing a distribution, especially when it's not symmetric around the number average (using weighted averages and the like), but that's very tangential to the point I'm making. My point is that those topping the logs are a very minor statistical minority of a wider distribution, which is a result of going for secondary stats that are susceptable to RNG.
You can chose to gear in a way that has a wider spread but the chances of hitting the favorable numbers are on the order of 1% even if you due everything perfectly due to RNG. It requires consistent orb acquisition, crits happening on higher orbed MBs, spell misses below your gearing, and so on. It's a dice roll, imo. You're just as likely to get really low DPS as you are really high DPS for the same statistical reasons.
Those topping the logs are not there because they go for a secondary stats that favors RNG or because their DPS is at the 2nd or 3rd standard deviation, they are just good players performing well. Another major reason for why good players will rank is because most people can't do proper DPS even close to the 1st standard deviation even if they wanted to.
Skill is most certainly a factor as well. So is buff feeding. It's a combination of all of the above that nets someone a #1 slot, if you want to get a better understanding of what works you need to look at a lot more than just a couple of the top logs. Especially considering how borked some of the reports are in FLs. You're better off picking random ranked people from the top 50-100, imo.
wrt to going for hit cap, I'm not pushing some agenda. Merely pointing out what happens if you don't. There is a broader statistical spread. You can verify it by running a bunch of different simcrafts with various stats. Look what happens to the spread when you ease off hit and add another stat. The number average may stay the same or even go up ever so slightly, but you also broaden the distribution.
I'm not telling anyone which is better, but I do have my personal preference. Either works fine for the content.
Your first post was about hit cap and how it affects our spread and you go on implying that people who are ranked are lucky somehow and they choose luck dependant stats. That is not true. Being lucky ahs nothing to do with getting ranked. Buff feeding gets you high ranked but it is usually in combination with legendary staff at the moment. To get ranked you don't need to get buff fed, you don't need to be lucky with your hits, you don't need to be lucky with orbs. The spread is not that significant. Even if you miss a lot and even if you get some unlucky orbs, you will still get ranked if you play well. Conversely, if you don't play well and get lucky with hits and orbs you still won't get ranked most likely.
Every thing you say about bell curve and how people are as likely to do good DPS as bad DPS does not matter. Of course it is a dice roll if you have RNG in your stats but that dice roll doesn't mean anything for your DPS in a raid. It is not of importance. Performance is.It requires consistent orb acquisition, crits happening on higher orbed MBs, spell misses below your gearing, and so on. It's a dice roll, imo. You're just as likely to get really low DPS as you are really high DPS for the same statistical reasons.
This is false. In 10 man it is all about having sufficient gear and skill. It has nothing to do with going for stats/gear that are susceptible to RNG. Sure, if you lack Dark Intent it might be hard. In 25 man, getting top ranked you need to have legendary at the moment. If you aim to just get ranked in 25 man, the requirements are the same as 10 man.My point is that those topping the logs are a very minor statistical minority of a wider distribution, which is a result of going for secondary stats that are susceptable to RNG.
Last edited by Ariadne; 2011-09-03 at 01:08 PM.
I think you both have valid points, but I feel that Ariadne is "more right" in the sense that gear, buffs, and skill play a much larger role than RNG, so much larger that it makes RNG mostly irrelvant. This doesn't mean that some of the people at the top of the WoL haven't gotten some lucky RNG, it just means that most of them are there because they have a combination of good gear, good buffs, and good skill first and formost. If RNG affected them at all, it was so small of a factor that it's barely worth mentioning.
Skill AND Gear are by far the biggest contributors to your dps. RNG plays such a small role that it's not even worth mentioning. Ofc buffs matters aswell, but most SPs just have the standard raid buffs + DI.
Basically what Kilee said. The top parses probably had good RNG, but that's not the reason why they are on top, it's gear and flawless play that matters.