Going to need some help with how "All Stars" works here, after thinking I've got it figured out it changes randomly again

So i’m kind of confused here, i’ve done the math here for my char and https://www.warcraftlogs.com/rankings/character/11609182/8/#boss=0
Adding this up is 835 but i’m at 806 when i look myself up? And i was at 824 last night. Why is my Marksman “All Star’s” ranking much higher then my overall? I don’t play any other spec why would it compare different specs?

there are 2 different allstar rankings for you as hunter. 1st is compared only mm spek and 2nd compared all speks. and it’s all right with that. why you think that you can have high rank in lowdamage spek and this high rank will be also in overal rank? example:

  1. you play mm and do 80k dps. and it’s top1 mm log - you receive 100allstar points for mm spek
  2. enemy do 100k dps as bm on same boss. so you receive 80allstar point for this boss in overal rank with your top1 mm result.

as for me i don’t understand why this rule do not work when compare classes with each other. there is exact zero point to make table with different classes which count separately… it’s rediculos that every realm allstarts table have dozen discipline priests in healing tables and shamans in damage table. and it’s only because top1 for these classes is so low that everybody can easyly do 80+ allstars points - it’s nature of the class that it can’t do exceptional damage or healing. so for example worldtop20 heal paladin is not even in top10 allstars realm table at all because this table is full of worldtop2k discipline priests…

1st mage in overal player damage ranking have 49th place :slight_smile:
and yes this mage beat all this top1-top48 on every boss.

Keep in mind the All Stars points I show on the character pages is live, i.e., it’s completely up to date based off current rankings. However the actual All Stars rankings only update once every 24 hours. This means the two numbers won’t ever match up exactly. All Stars takes a lot of time to actually compute the totals and get everyone ranked in order, and it’s not really possible to do that live.

As ztn points out, All Stars has lots of problems. I’ve pretty much come to the conclusion that All Stars as a feature should just be removed (or changed to only be per spec with no cross class view). It simply isn’t possible to produce any kind of meaningful ranking that works across different classes and specs.

I can either use absolute DPS, which is of course meaningless across specs, since one spec may be better than another, or I can attempt to normalize, which doesn’t really compare well either. If anyone has any suggestions, I’d love to hear them, but probably the easiest solution is to have only per spec All Star scores and to remove all aggregation.

One idea might be to have a “top 10” view that shows the top 10 in every spec for All Stars.Then you’d still have one page where you could see the best of each spec, but i wouldn’t aggregate them together into a single combined ranking.

  1. i think it’s very bad solution to have only allstar spec ranking. i think that now we have only one allstars ranking that is ok - it’s ranking vs class/role… for example warriors play normaly in arms but several bosses like kormrok or manoroth in fury. they never play this bosses in arms. now warrior who plays in better spec - have better allstar ranking. if you introduce only spec rankings then warriors who play 2 bosses in fury every cd will have only 11 ranked bosses in arms and 2 in fury. so they lose to every noob who have 13 ranked bosses for spec… so every warrior who compete in rankings will forced to be only in one spec for whole raid even if other spec do more damage and all raid and raid leader want warrior to take another spec… all this suited for mages, warlocks, hunters etc… and what next? next level of separation is to separate logs with different tallents etc. why fury with bladestorm compete with fury without? really there is no point of existing allstar spec raiting at all… the only point is class/role rating. the only exception is feral/moonkin spec which must counts as “different classes” for obvious reasons…
  2. we need another allstar rating which counts vs role. so for example for damage dealer role it count all damage classes from top1 result…

example: boss1

  • mm hunter do 100k dps
  • bm hunter do 80k dps
  • arcane mage do 200k dps

mm hunter have 100allstars class rating
mm hunter have 50allstars damage rating

bm hunter have 80allstars class rating
bm hunter have 40allstars damage rating

arcane mage have 100allstars class rating
arcane mage have 100allstars damage rating

this 2 allstars ratings is only that have any meaning. spec rating means nothing at all… if you don’t want to inroduce role rating it’s not a big problem but if you kill class rating then all spec ratings for classes with several specs in one role will be just nothing.

Right, but isn’t that all changing in Legion? With the Artifact, It seems like people may just be locked into a spec and not be switching any longer.

I don’t really agree with just using absolute DPS by role either. Then all we have is an All Stars list filled with Arcane Mages. That seems just as useless as what is there now.

yes but this variant is a little better then right now. yes top10 role table now will be only mages but it’s not true in general. during progression there was real competition betwen classes and mages shine only for fast kills. the main focus of my post is to save class rating and not separate it by several spec ratings. i think it will be even better to delete spec ratings entirely not to confuse people like topic starter. i am sure legion will not change this and players in general will not be forced to one spec. artifacts for another spec will be opened even before 110lvl (blue blizz post) and as we can see on alfa leveling artifact is not super time consuming thing. you can easyly have offspec artifact not far below mainspec.

my main idea is allstar is competition - it’s not funny fake nubmer - it’s real competition and thousands of ppl invovled in it now. im my guild even wf+socket loot distributed according to allstar rating person have.

I don’t know if you ever used Pro Raiders, but they actually did both, i.e., you could view both an absolute DPS ranking and a normalized one. I don’t know if that would be too confusing to have both options or not.

The main problem i have is some random player who just goes down on killrogg every week suddenly has as many all star points as me, i’ve only gone down once and even though its a high 99% its only worth 42 points. Its impossible to get rank 10 on any fight unless your guild lets you pad or sets a strat around you. Kinda makes everyone who has a semi casual guild feel like shit even though you play well.

and whats your suggestion? you want to have 100allstar points without any effort by just pressing your rotation? then everybody will have 100points and where is competition then? if you don’t envolved in high rank competition then you can operate your 99% orange number and be happy. why you want to ruin competition for every other person?

Naw what i’m saying is its not competition to be the only dps playing the game on Hellfire Assult and getting a high rank because of that. Its not competition to AoE the adds on iskar while the rest of your raid is on the boss. Those should be logs that should be removed imo.

There’s nothing I can really do about this. Sure, I could come up with some completely arbitrary set of rules like “You can’t do more than X% of the damage to a set of adds,” but that’s really problematic and starts feeling a bit tyrannical to me. Farm is farm and people are going to do silly things. I can’t really be in the business of special casing every single fight to try to stop cheesing.

I don’t think its meaningless at all. The best suggestion I’ve heard so far is your previous suggestion of just rating against the 99% mark, rather than 100%. This just served to take out the cheesed rankings and gave a better sense of how people were doing.
Hell, after you made that suggestion, I manually do this regularly to get a better sense of where people on my team actually are.

I think your current approach of lumping the specs of a class together is not as meaningless as is suggested either, once you take out these cheesed attempts.

To continue with the hunter example, look at say, Iskar and Assult. The top few parses are BM as the attempt has been set up to cheese the rankings and give them a constant supply of adds to cleave. Apply the 99% measure, these get filtered out, and you’re left with a better indication of how you’re doing on this fight in a normal application against the top hunters.
You want to avoid the “yeah im ranked super high because im the top survival hunter, never mind that i just did 23k on archi!” situation.

Also it looks like people will still be changing specs all over the show in legion with the way gathering and spending artifact points are shaping up so far, itll just take a month or two to bed into it.

Except it’s easy enough to get 99 that it would remove all competition among the high end players. There needs to be more of a spread.

Do you not get that if you allow those who are over and above that 99% line to get more than 100 points? That would allow those racing for the top to still have a differentiator.
From my experience, it does create a tight/small spread for those who are close to the 99% line - the thing that then separates them is which fights they have over that 99% mark, and by how much.

shrug - I’ve found limiting it to that lets me see who is actually doing pretty damn well (e.g. arcane mages who arent in a guild that is chucking a dozen arcane mages into a 15-30 second kill look a bit shit, even when theyre extremely good) - while still allowing for spread/providing scope for rankings competition. It absolutely needs those players above that 99% watermark to be able to achieve additional points for it.

Still, its just one possible solution of many - I only thought to try it after you suggested it, but was quite pleased with the outcome.

Yes, I agree that with a normalized system the goal should be to reduce the spread (see the other thread on this subject). I think letting people’s points go above 100 is a poor way of doing that though. The goal should be to allow for differentiation without allowing it go get too out of hand. I think that can be accomplished while still using a 100 point system per boss. The goal should be to guarantee that 99th percentile still gets a certain minimum amount of points.

I suppose one advantage of going above 100 is it lets people know that anyone with a score higher than N is doing great. In that sense, I can see the advantage of treating anything above 99th percentile as bonus points. You’d still have the issue of cross class comparisons being lame, but you would at least be able to look at an absolute score that doesn’t go down when people cheese.

this is fake advatage. the real disadvantage that allstar number become fake number which means nothing. you receive for example 170 and what it means? what number have top1? i don’t know - may be top have 300 and my 170 is just nothing special even its 70% higher then 99%base number. now i receive 95 and i know that i must up my result by 5% and i will be top1. when i receive 170 i don’t know anything. it’s all ok now with class allstar ranking. it’s not ok with role ranking because it’s not count from 1 top1 base results for all classes. it’s not ok with spec rankings because it confuse people.

p.s. sorry i know i repeat this already many times :frowning:

I think the current parsing and ranking system is fine.

While it would be nice to have spec based comparisons, atm it does not make sense, as a lot of pures (rogues,warriors,warlocks) tend to switch specs between bosses and would be at disadvantage.