State of Set Balance Report

General Discussion
1

Hello Friends,

While we wait for the upcoming patch notes, Rage and I want to share the State of Set Balance. This is a report that organizes the Best, Worldwide Greater Rift clears by their Adjusted Clear, (a normalized GR tier score based on a player being 5,000 paragon and using all 15 minutes to clear the rift.)

Our report shows Peak Clear, Averaged Top 10 and Average of the Top 100, coupled with other metrics that spotlight the competitiveness and certainty of each build. To learn more about our methods, please read this doc.

The State of Set Balance represents clears from S26, S23, E16, E15 and E14. It paints a picture about set balance that isn’t easily gleaned by opening up the in-game leaderboards and flipping through class sets. We’ve accounted for nerfs, buffs and other changes along the way.

When the patch notes eventually arrive, we hope you’ll have a better understanding of how your favorite build may be impacted.

Did this report change your perspective on a particular set?

Please share your thoughts in the comments below.

18 Likes
2

I would say it looks great but I think I am blind now… (only seeing rainbow blur of colors everywhere I try to look) teehee.

Seriously tho, great job all.

One suggestion, maybe add a key to explain the meaning of the colors…
Nevermind, I see it now.

1 Like
3

I told Rage that I felt like a deer in headlights after staring at it. Not from all of the colors, but from the information! If you check the right column, we prepared an abbreviated appendix that explains it in text. (Ahh glad you found it!)

3 Likes
4

Gives a pretty clear picture of which classes need the most help too, very well done, thx so much for the work.

1 Like
5

Will you share it on other forums where devs are more active?

1 Like
6

We’ve got it Reddit and I’ve personally started notifying some ‘important’ people. It will likely find its way to the devs.

3 Likes
8

I haven’t truly digested it all, but I have to say, just on first glance, this is an impressive level of research you’ve given us. Thank you for the hard work and detailed report!

1 Like
9

Personally I believe your calculations are a bit skewed. In most calculations to achieve an average you must eliminate the highest and the lowest. In achieving a correct balance of sets, you should sample at least 100, but preferably 500 players. Then remove the top 10% and the bottom 10%. Then average out the rest.

Your calculations, although interesting and I’m sure you put a lot of work into them (kudos by the way), are artificially skewed by the high paragon players on the top of the leaderboards and not a true average. It would truly be interesting of what the true Paragon Clear average would be.

Just a suggestion.

2 Likes
10

There are certain sets that manage several skills quite effectively. Have they been taken into account ?
I am thinking for example of MOTE6 Earthquake or Seismic Slam, Sunwuko Tempest Rush or Wave of Light, etc.
I have the impression that no but I want to have a clarification to get a better idea.

1 Like
11

I would add the colors to the legend on the right. Reading out the color doesn’t give me an immediate idea of the tier list.

Otherwise excellent work! Hope the devs take all this into account.

2 Likes
12

Hi Argos, we feel we’ve addressed much of the paragon issues by normalizing the clear to 5k paragon @ 15 minutes. If you take a peek at any set tab in the spreadsheet you’ll see the compilation of clears. Unless the set was recently updated, there’s over 600 clears per set, they are sorted and we evaluate based on the peak, top 10 and top 100.

Yes, paragon is still a factor due to the defense which allows for more aggressive play, however, low paragon players are still able to compete with the 15k+ paragon behemoths.

For the purpose of balancing sets, IMHO the best data point is the Averaged Top 10. Why? Because most standard deviations are low across all sets in the Top 10, and we can assume the most talented players and hardest pushes are represented, where as when you get to the Top 100, there is a steep drop in the adjusted clear in less competitive sets due to the reality of this game, some sets are just unpopular and the data suffers because of it. Take the Necromancer’s Trag’oul set for example:

PEAK: 141.4
TOP 10 AVERAGE: 136.6
TOP 100 AVERAGE: 105.3

No one should make set balance decisions according to a GR105.3 Adjusted Clear. That’s why it is not recommended to sacrifice the top and bottom 10% when looking at the whole picture. This type of approach only works on more popular sets like Inna’s and Marauders.

3 Likes
13

We didn’t splice out individual skills. In a previous post, we did this for Marauder, Firebirds, etc. Given the broad scope, working with data from multiple seasons/eras across 42 class/set combinations, we had to keep it simple and just say Marauder will represent Cluster Arrow AND Multishot skills.

1 Like
14

I hate to disagree with you dmkt, but if you’re basing your entire calculations on the top 10, then your averages are only representative of the top 10 players of each set, which some are the same player with behemoth paragon points. It’s not a true representation of the players base nor skill. It’s a snap shot of the highest clears, but certainly not a true balance if you take into account of the average paragon player in the game.

As for this statement: I believe it’s the opposite. They should make decisions to the top 100 players and not the very elite ones with 15K paragon. It’s the paragon points that is the problem and not a true representation of each sets capabilities.

15

You’re not understanding.

Averages, in the way that you’re thinking of them, don’t mean anything in this game.

If we take the top 1000 of each set from the base leaderboard, and just average them out (or if we drop the top 100 and bottom 100 and average the remaining 800, or whatever) then the result we get from that is basically just a bunch of gibberish.

Why? Because some sets get played more than others. For something like LoD HotA, only a relatively small number of people (~25) play it really competitively. And for some other builds, like, say, Necro Pestilence, that number is even smaller.

Let’s say we were going to take your suggestion, and determine the strength of builds based on an average of the sort you’re suggesting. Such a thing might show a build like Pestilence to be 30 GRs behind Raekor, or Inna. But if you then added 30 tiers of power to Pestilence to balance that out (an extra 111x multiplier), you would likely have people blowing through 150s in 3-5 minutes, because the low value of that average is only due to the fact that nobody is playing the build. So once the build got buffed, people would return to playing it in a flood, and it would suddenly “become more powerful”.

So, I’ll say it again: averages of that sort are worthless.

You also seem pretty confused about the paragon issue. Many of the sets included here have their top 10 drawn largely, or entirely, from players with paragon around 5000 (Raekor, LoD HotA, Marauder, Inna, etc), or sometimes lower, not some huge number. For others, more of the represented players have big paragon… and that’s fine. If we’re correctly adjusting for paragon (which I think we are), you’d expect to see examples of both bigger and smaller paragon on the boards, and that is indeed the case.

9 Likes
16

I think it is correct to use the top 10, because that represents the true strength of a set if used to its full potential. There isn’t much reason to go lower because then we don’t get an accurate view of what it’s fully capable of.

It would be better to make a different spreadsheet for a lower paragon cap than to omit the top 10, but that would be a huge amount of work that isn’t worth the time.

4 Likes
17

Done! Thanks for the suggestion.

3 Likes
18

Thank you for your work and involvement, Rage and dmkt :slight_smile: .

2 Likes
19

NO! I beg to differ. Obviously you’ve never ran a probabilities or an statistical reports before. It’s like telling me that if you take the 10 richest men in the world, you’ll get an average of what 500 random people make each year.

The way you’re account for averages not only skews your numbers but makes you report nothing but a bias attempt to showcase the Barbarian as having the best sets and making the numbers work in that favor.

Look, I’m a lifelong Barbarian fan. You can look at my profile and you’ll see that my favorite builds are barbarian. But your attempt to showcase a fallacy just to justify your belief, does not make your numbers correct.

Sorry Rage, I respect you. You’ve done wonders for the Barbarian community and I love your builds, but this is just wrong.

I will leave it at that.

20

That reading of the situation is… odd.

For one thing, I didn’t particularly set out to show that “Barb was the strongest”. I mean, the fact that Raekor is the strongest build in the game right now is not exactly breaking news. Raekor’s got the lowest paragon 150, and the lowest time 150.

Furthermore, personally, I’ve never really cared what build (or class) is strongest, weakest, whatever. My favorite build, which is still the thing I play the most, is Leapquake, a middle-of-the-road B tier build, which I’ve continued to play for years, no matter how weak or strong it is- because I enjoy it. And, I’ve never put much effort into trying to get it buffed, because personally, I really don’t care.

Also, that reading is super insulting to DMKT… you think: I steamrolled him into just agreeing with me about everything? Or, he’s also a “Barb partisan”… despite the fact that he doesn’t play Barb?

That’s… unlikely…

3 Likes
21

Hi Argos, first I want to make clear – there is no agenda other than pursuit of the truth. Rage and I only wish to put helpful information in front of the community.

The 10 richest representing the wealth of 500 random people example is too loose of an example to be a fair one due to population size being in the billions.

I recommend you go through the same journey and see what happens when you start averaging the Top 1000, with and without the top+bottom 10%. Do this for a popular and unpopular set. The results will contradict the top clears to the point where the report is deemed unhelpful, then not credible. Top clears are highly influential, they are the proof that a build is performing within acceptable limits, they cause nerfs, they heavily impact the decisions of players.

Rage and I recognized that the average metric deteriorated very quickly even before the Top 100, so a stop was put there, though we draw and sort our data from the Top 200 in various eras.

What is the point of all of this anyways? To keep entertained while helping others. Take this example, which thousands of players rely on to learn about the best builds:

The Maxroll people organize their own tier list season-after-season, it impacts players decisions – especially those who cannot afford the time to setup all builds. And while Maxroll doesn’t claim their tier list as an absolute truth, people trust and use it regularly. But what is it based on? Player opinions, collaborative spreadsheets like ours, a proprietary Maxroll algorithm, secret sauce? Who knows right?

Here we have created a parallel list. We’ve exposed all of the data behind it, the calculations, the methods with explanations. Our results are similar to Maxroll’s, but not exactly the same. Our list was derived AFTER the patch, theirs coincided with its release.

I’d say our work product was inspired by this Maxroll tier list. Without structuring the averages as we’ve done it — the tier list would look very different. Maxroll’s Tier list is helpful for competitive players, as is ours. Form your opinions using the Average Top 100 since the Top 10 is bothering you – that’s why the column is there.

I hope you’re still interested in telling us the specific barb set results you disagree with and why?

4 Likes