All Stars Revision Proposal

This is a proposal for revising All Stars across all three log sites (Warcraft, FF, WildStar).

The current All Stars formula is that a boss is worth N points, typically 100, but some bosses are worth less (e.g., Hummelfaust in FF). To find your score, the result is N * Rank 1 DPS / Your DPS, so if your DPS at rank 2 is 200k and rank 1 is 400k, then you get N * 0.5 points = 50 points for a 100 point boss.

Here is how I propose revising this formula:

(1) Each boss is worth 1.2 * N points (where N is the value in the old formula). You can earn up to N points as before, but now 0.2 * N “bonus points” are available.

(2) Determine the player’s rank percentile (e.g., 0-100), P. The player then gets (P / 100) * N points based off rank position.

(3) If the player achieved legendary (orange) ranking status, defined as P >= 95, then the player is eligible for up to 0.2 * N bonus points. These are computed using the old formula, so 0.2 * N * Rank 1 DPS / Your DPS.

So here’s a concrete example. Suppose rank 1 on Kilrogg for a spec is 400k DPS and rank 2 is 200k DPS. There are 100 total rankings.

Currently the player with rank 2 would get 50 points under the old formula.

Under the new formula, they would get 99 points for having a 99th rank %, and then they would get 10 possible bonus points (20 * 200k/400k), for a total score of 109. Rank 1 would get 120. The top person still gets to pull ahead by ~10%, but the effects of the cheesing were reduced dramatically (as it should be given that the other person has 2nd place on the rankings).

The idea here is to still allow some differentiation, but to prevent it from getting completely out of control. There is also a significant reward for achieving a legendary (orange) ranking, in that you only get the bonus points if you achieve that 95th percentile.


1 Like

For Warcraft only there is a concept of combining similar roles of the same class together, i.e., in addition to your spec All Stars score, you also might have a role All Stars score. The question then becomes “What should your rank position be?”

For example, MM is vastly superior to BM or SV on Iron Reaver, so an MM hunter’s rank percentile would actually go up in the role score if I computed your rank position on the global hunter list using all 3 hunter specs. I don’t think this makes sense when computing the rank percentile, so my proposal for the role rank percentile computation is that the spec with the top DPS score (i.e., the best spec on the fight) is the one whose rankings are used. Your rank % is then inferred by where you fall within that best spec’s ranks.

This maintains the current model where if you play the best spec on a fight, your spec points and role points remain the same.

This only really applies to Warcraft also, but for completeness, you must have a rank in the top 10,000 on a fight in order to get any All Stars points. I may relax this more if I ever move to faster hardware on my job server. :slight_smile:

I’m very much in favour of steps 1-3. Makes a lot of sense.

I don’t think I like the multipliers for role, though. Do you think they ought to be combined? If yes, why? Do the multipliers defeat that purpose? They’re like a ‘middle ground’ whose main purpose is to remove healers from the top spots.

Well it seems to me that the best solution to that end is to simply not compare different roles within the same all-star rankings at all. Poor tanks will never get good combined rankings for anything anyway! :<

So doing away with the combined rankings seems like the most logical thing to me.

This does prevent a problem with regard to Discs, obviously!

Not comparing across distinct roles is a possibility yeah. That would probably force me to move to a tabbed UI for All Stars (with a tab for each role), but that’s probably ok.

Of course that begs the question of, why shouldn’t people who want to be able to see a combined view, be able to see it? But I don’t think it shows them anything useful, and I think adding in multipliers doesn’t necessarily mitigate the obscuring of the usefulness.

I think I am probably just going to eliminate showing tanks and healers by default in the damage view, and omit tanks and dps in the healing view. You can always select “All Tanks” or “All Healers” if you want to see those combined views.

So I thought of something really slick regarding the combined score. One of the issues with the combined score is sometimes people do a cheese parse on a spec that actually is performing worse on average. An example of this is Hellfire Assault, where BM has the top cheese parse, but even at the 99th% MM is outperforming BM.

Since of course any spec can cheese HFA, it’s just coming down to “what spec happened to do it”, so we need to work to eliminate this nonsense. My idea for this is to base the rank position for a spec off the spec that performs best at the 95th percentile. In other words, the bonus point cutoff is based off the highest performing spec at the 95th, and then both rank position and bonus points are computed using that best performing spec.

WIth this example, rank #1 MM would get full 120 points for HFA, but then every BM hunter that cheesed and beat the top MM parse would also get 120 points. Rank positions and the legendary cutoff for bonus points end up being determined by MM ranks and not BM. This ensures that the best spec is really used for these points and that coincidental cheesing by a spec that is objectively worse does not negatively impact the better spec.

That is how I read it to work the first time anyway haha

I’m not sure how I feel about encounters being “weighted” differently, or if this is even an issue, but I thought I’d bring it up. Using fflogs as an example: About 2000 DRGs have parses for A5S. This means the top 100 parses are eligible for bonus points. Then we look at A8S; 125 parses. Only the top 6-7 players are eligible for bonus points. It seems much more difficult to be in the top 7 in A8S, as it is a smaller number, making this encounter worth more. I’m not great at math tbh, so I thought I’d bring it up to see if it actually is an issue.

I have actually thought about making end bosses worth more. I will probably do this going forward but leave this tier alone.

Actually I think this might be worth re-visiting. Here are suggested point totals for both FF and WoW. I don’t have enough knowledge to do Datascape in WildStar, although I know people wanted one of the minibosses to be discounted completely:

Suggested totals below are the base points. You’d earn 0.2 * base points bonus points in addition.

Hummelfaust - 25 points
A5S - 50 points
A6S - 75 points
A7S - 100 points
A8S - 150 points

Hellfire Citadel
HFA - 50 points
Reaver - 50 points
High Council - 25 points (Fel Rage randomness makes this fight something that should be discounted)
Kormrok - 50 points
Kilrogg - 25 points (Visions ruin parses for this one, best strat now is to skip visions, should reduce weight of this one)
Gorefiend - 75 points
Socrethar - 75 points
Iskar - 75 points
Fel Lord - 50 points
Velhari - 75 points (last phase annoyance for speed runs, otherwise would make 100)
Xhul’horac - 100 points
Mannoroth - 100 points
Archimonde - 150 points

HFA reduced since it’s a clownfest where rankings are inversely proportional to raid quality?

I understand the idea behind weighting and I think it’s one of these things that seems decently fair on the surface, but honestly I think it introduces all sorts of new biases that hurt or benefit different classes and specs, and it’d just be opening up a whole new can of worms. There’s no perfect system–you’re always going to be certain people that might, “undeservedly” have like 5-6 points extra because they got lucky. That’s just something that is part of the rng nature of many raid encounters in World of Warcraft (again speaking purely from knowledge of wow logs, not ff or wildstar).

Some examples/thoughts, and these are just some more concrete examples, but there are plenty more on a more specific class/spec basis that you could dig for:

  1. Like, on HHC yes it’s true that something like Fel Rage is random, but (a) that matters more numerically for certain specs than others and more importantly, (b) a huge part of that issue is already solved by what you’ve done in basing the first 100 points directly on the percentile value. You can easily net the majority of points (and eligibility for bonus points) without having ever gotten fel rage.

  2. People that are ultra into parsing and speed kills for the sole purpose of showing up as Rank 1 on Warcraft Logs will go to whatever lengths they can, as a guild or raid, to stack specs for 15 second Iron Reaver kills etc. But that also doesn’t have to be limited to purely kill times, there are all sorts of things people do to parse on HFA, Iskar… whatever it might be. Obviously Paragon isn’t ( wasn’t now I guess…) a bad guild, but someone there has Rank 1 on HFA. On the other side of things, of course you have people that might love padding and are just in a bad guild and so they’re given the opportunity to grab high ranks on HFA, but I mean honestly who cares. At this point with valor and everything, the only guilds where that’s probably happening alot are guilds that aren’t even done with mythic or something, and in that case those people won’t even have any points for the last x number of fights.

  3. Visions on Kilrogg, again, this far into farm, you can just have people go into visions or draw out the fight for the specific purpose of ranks.

In addition to all this, it’s like again, I don’t think it’s a good idea to arbitrate specific values per boss. You’re never going to get values that most people agree upon, particular if it’s stepped in 25 point increments. At the end of the day all I’m saying is, you’re never going to have a perfect system for determining like “True DPS Skill”, and I think the current revision is incredibly solid and mostly “fair” to everyone.

Just my thoughts.

if you give council and kilrogg 25 points because reasons you should also give 25 points to assault(yay half raid afk) and 25 to iskar (kill boss in 1:00 - 1:30 and bomb adds)

Yeah this relates to what I’m saying exactly. We can go down that rabbit hole forever though, in my opinion it’s somewhat pointless and just introduces a whole load of new biases. With the current system, you still retain the vast majority of points that would otherwise be lost to people doing things like that (under the old system) by just playing well with a relatively “normal” set of conditions.

This makes sense, your original proposal in this thread seems the most fair to me. It also actually makes me wonder why hummelfaust is included at all in all-star points. Over time, it becomes 100% ridiculous cheese parses, and it is just trash after all.

as for me all these changes is only for one purpose - noobs now look better and there was exactly 0 reasons to change anything. dividing skill of top players by x5 just for noobs don’t cry is a joke… and this weighting shenanigans is out of my mind entirely. warcraftlogs is not wowprogress. this was already well balanced. noobs looking their % and happy entering 95% and top players looking allstar rating which proportionaly ranking them. why you want to merge 2 separate things into one i don’t know. it was absolute clear system and after all this changes it will be just fake rating showing nothing…

The weighting idea is just under discussion. Nothing has been decided there. What do people think about only counting the end boss as more points (and leaving the others alone)?

As for “dividing skill of top players by x5 just for noobs don’t cry”, I’m not sure why you see it that way. It isn’t skill that is causing one person to do 2x the damage of the next person. Being allowed to solo adds on Hellfire Assault is not an indicator of skill.

At the moment I have the new points showing in the character pages while keeping the old system in place for real All Stars, and for the most part, the relative positions of people at the top seem to be staying the same. The ones dominating the top of All Stars are staying right at the top where they belong, so I don’t think shrinking the weight given to cheese is going ruin anything here.

I’ve actually contacted several of the top All Stars for some of their respective classes and received an overwhelmingly positive response regarding the revisions in the first post. This is even coming from the people doing the cheesing. They all liked the idea of legendary rank bringing in the bonus points and liked the idea of reducing the cheese.

Anyway, there is no way you will convince me that a person having a 99% rank while being rank #2 and only getting 30 points out of 100 for their boss is a fair system. That’s the world we’re in right now. If you think that’s actually fair, then we’re just never going to see eye to eye here.

Another way of thinking about it is right now you can beat someone on 12 out of 13 bosses by 3-4% DPS, but have one boss where you have way fewer points (e.g., Hellfire or Kilrogg). Someone can then be ahead of you purely because they cheesed one boss and caused you to lose 50+ all star points on that one boss alone.

If you don’t reduce the weight given to cheesing, then a cheese performance on a single boss can catapult you ahead of someone who did better than you on every single other boss. That’s ridiculous.

There’s no question something has to be done to combat this, since no single boss performance should give you so many points that you can perform worse on many others and still be ahead. I think the proposed revision does a good job of that, but maybe people would like to see a different amount of bonus points.

In my mind there are two variables here:
(1) When to apply the bonus points. I suggested 95%, but any value from 0 to 100 could be suggested. They could just always be earnable, they could be earnable only at 99%, etc. I think legendary rank earning bonus points is a fun idea in that you can basically think “If I can get an orange rank on this boss, I’m going to collect the bonus.”, which is why I have proposed 95%.

(2) The multiplier on the bonus points. A bigger multiplier will magnify cheesing. Smaller will reduce it. I chose 20% since that allows someone who performs significantly better to get 5-10 points ahead. If they cheese multiple bosses, that will still add up and keep them at the top. I am open to suggestions regarding this number though.