[Analysis] Era 11 Patch 2.6.6 Classes Grift Clearing Efficiency Indicator Assessment (Class Power)

General Discussion
1

In this study, I am going to compare the greater rift clearing efficiency of Barbarians to Necromancer and Wizard class using whole leader board of Era 11 Patch 2.6.6 as of 25/08/2019. The analysis shows that Wizard class seems to have broken the balance achieved in the game in S12. Currently their clearing capability is “significantly” higher than the other classes. Buffs for barbarians are once again required to create a logical and reasonable end game balance to make this game an ARPG again (where classes can kill as efficient as each other).

This analysis is exactly based on my earlier analyses except that I am using Necromancer and Wizard as reference and comparing all of the classes (however I am commenting on only the barbarian class as I main barb).

Note: Some reddit community requested me to add data for all classes so I am modifying this analysis to include them as well. I won’t comment on them though as I main barb. Also, before some other community members copies my approach (which may be misused to misrepresent the reality, just as a precaution I would like to add all classes)

Reddit Link:

Background:

In this thread, I will try to compare all of the classes and their power efficiency in clearing high greater rift levels using era 11 solo leaderboards.

The Methodology:

    1. The whole era 11 leader boards have been extracted for the considered classes in the game.
    1. Total number of data points considered are 7000 in total (considering all of the classes)
    1. The parameters used in the analysis are Greater Rift Level (A), Clearing Time (B), and Paragon Level (C)
    1. For each toon, a greater rift clearing efficiency coefficient (P) has been calculated as follows:
      P=A/(B*C)
      The resulting numbers are scaled up with 1M in order to ease the comparisons.

The Explanation of the Formula:

According to this formula, the efficiency in clearing grifts of a toon is defined as the highest possible greater rift level to be cleared as fast as possible with the lowest paragon level. It is a way of normalizing the greater rift clears and a way to observe where would the average player perform.

Performance Indicator Results:

Class Grift Clear Efficiency Comparisons Graph Link
Barbarian-Wizard-Necromancer Imgur: The magic of the Internet
Monk-Wizard-Necromancer Imgur: The magic of the Internet
Crusader-Wizard-Necromancer Imgur: The magic of the Internet
Demon Hunter-Wizard-Necromancer Imgur: The magic of the Internet
Witch Doctor-Wizard-Necromancer Imgur: The magic of the Internet
All of the classes Imgur: The magic of the Internet

Imgur
Barbarian-Wizard-Necromancer

Imgur
Monk-Wizard-Necromancer

Imgur
Crusader-Wizard-Necromancer

Imgur
Demon Hunter-Wizard-Necromancer

Imgur
Witch Doctor-Wizard-Necromancer

Imgur
All of the classes

Analysis of Results:

Note: this assessment considers the expected class averages, which considers every possible build options available in the leaderboards (shown as a continuous line on the charts)

Also note that barbarian leaderboard is biased due to the gimmicky play mechanics (and bugged elite packs) of wall charge in Raekor HOTA build, which may unrealistically skew the barbarian class efficiency to higher levels, which is not the case for all other barbarian builds.

    1. Greater rift clears at the bottom of the leaderboards start at 100 for Necromancers, 105 for Barbarians and 120 for Wizards.
    1. At greater rift clearing efficiency of 90 a Necromancer can clear 102 grift level while a barbarian can clear 105. On the other hand, wizard leaderboards start at 120 so there is no direct data to observe. However, if we project the wizard class average, we should expect 112 as a grift clear to occur for wizards at this efficiency.
    1. At 120 grift efficieny level of an average wizard, an average barb would clear 110 while a necromancer would clear 108.
    1. At 130 grift efficiency level of an average wizard, an average barb would clear 120 while a necromancer would clear 120 as well.
    1. At 140 grift efficiency level of an average wizard, an average barb would clear 126-128 while a necromancer would clear 130-132.

Verdict:

According to this analysis. Barbarian class (its set and/or supporting legendaries) require a buff amount equivalent of 14 greater rift levels. And wizard class and its supporting legendaries must not be touched as the game balance is continuously upset by such unthoughtfull buffs. It turns into a dog chasing its own tail.

Currently Wizards and Demon Hunters are much stronger than the other classes, all of which follow very similar curves otherwise. Other classes are more or less balanced among each other.

Conclusions:

Buff barbarians via supporting legendaries 1.17^14=9x times. The class requires a damage buff of 9x across all its potential builds in order to match wizard performance. In order to preserve the game’s balance, the other classes must be buffed at required amounts accordingly except Demon Hunters (maybe very slightly~4 grift levels).

Average GR cleared by the top 1000 players Average Greater Rift
Barbarian 110
Monk 109
Witch Doctor 113
Crusader 114
Necromancer 108
Demon Hunter 120
Wizard 124

The greater rift difference order displayed by the class mean curves more or less aligned with the hard averages of the top 1000 (although such a comparison is not exactly correct as we cannot average different GRs in this manner).

*Edit: Hard averaging different GR levels is wrong. Varying performance levels clearing different GR values cannot be logically compared. such an averaging is just a description of the player base populating the data. These averages are by no means the sign of any thing meaningful.

Thanks for reading

Prokahn

References
Season 11 Analysis:

Season 12 Analysis:

19 Likes
2

Imgur

Conclusions:

The strongest class in terms of greater rift efficiency is wizards (represented by orange dots and lines), a class that many think is vastly overpowered. Many have called for nerfs to this class.

The weakest class in terms of greater rift efficiency depends on the greater rift level analyzed. The two weakest classes for grifts considered below 113-114 were necromancer (the worst) and monks (second worst). For greater rifts above 114 monks had the lowest greater rift efficiency.

Monks are represented by red dots and line.
Necromancers by gray dots and lines. The odd shape of the necromancer curve relates to its thorns vs. other builds.

Barbs are blue dots and lines.
Demon Hunters are brown dots and lines.
Crusaders are green dots and lines.
Purple are purple dots and lines.

Prokhan also showed a more simple comparison of the average grift clears of the top 1000 players. Each > represents a 1 greter rift difference.

Wizards>>>>Demon Hunters>>>>>>Crusader>Witch doctor>>>Barbarian>Monk>Necromancer

I would encourage everyone when they are comparing classes to link this thread/post. I think it would help to dispel misperceptions that occur on the forum.

@ greater Rift efficiency of 50 Greater rift clear
Necromancer 111.0
Monk 111.7
Barb 113.0
Crusader 114.5
Witch Doctor 116.0
Demon Hunter 117.3
Wizard 123.5

EDIT:
The revised OP now includes data on greater rift efficiency for all classes. The data informs and may dispel some misconceptions.

All classes in comparison to wizards need a buff; however, there appears to be other classes besides barbs who can make a compelling case for buffs.

From Prokhan’s analysis, it seems that Free considers mid-tier for classes to be higher than greater rift 120 (if not 122). Although this may make sense for the 2 overperforming classes, this is not true for the other 5 classes. I wonder the rationale why Free did not share this analysis?

EDIT: For those who want a way to reproduce this calculation for themselves:

  1. Request each leaderboard using Blizzard’s API
  2. Open the html file, copy and paste into Microsoft Word
  3. In Microsoft Word, delete the first and last part of the file that does not include player information

Replace “}]},{”;player" with “;}]}^p{”;player" (Reason: Used to insert line breaks between players)
Replace all , with ^t (Reason: For use as tab-delimnited file)
Replace all : with ^t (Reason: For use as tab-delimited file)
Delete all " (Reason: For prettiness)
Delete all [ (Reason: For prettiness)
Delete all { (Reason: For prettiness)
Delete all } (Reason: For prettiness)
Delete all ] (Reason: For prettiness)

  1. Save as plain text file
  2. Import plain text file into Excel
  3. Use the sort function and insert appropriate spaces to re-align columns for players with missing entries such as clan name
  4. This excel (and of course the original json file from Blizzard) file includes the grift clear level, hero’s paragon at the time, how long the clear took (in milliseconds). Use excel to make the greater rift efficiency calculations.
  5. Use Excel for simple graphs
  6. Line fitting is more sophisticated and something like SigmaPlot, OriginPro, or R/SAS may work better than excel.
5 Likes
3

The power over Necro is most likely thanks to bug packs and Wall charge mechanics. If Raekor is disabled, one can be sure the curve will shift left (slightly). Overall Necro Barb balance from S12 has been kept at close levels. However, once again Wizard class seems to have received some buffs upsetting S12 balance levels in favor of wizards (once again).

Barb is slightly more efficient than Necro upto ~116 afterwards Necromancer becomes more effective (slightly). However, at top pushes Necro performs better than barb top clears.

Also note that now there are players in Wizards and Necro clears with +10000 paragon. So paragon cap apparently is not true.

I am not going to do this analysis for other classes.

2 Likes
4

Quick question, did you use the top 1000 or 2000 data for each class? I assume that you grabbed the data using the API. I may repeat what you did for all classes.

4 Likes
5

API gives the 1st 1000 ranks. The ranks you see in game leaderboards. The paragon data is registered to these APIs. If one does similar analyses using web leaderboards, the observable paragon will be the current paragon values instead of the paragon at the time of the clear.

1 Like
6


Thanks. When I downloaded era 10 from the API, I had the top 2000 clears. I have not tried for era 11 solo.

4 Likes
7

Blizzard developer APIs are only giving the 1st 1000. If website APIs are giving more, those will not have the paragon at the time of the clear but it will list the current values, which would be wrong.

1 Like
8

I have only been using the developer API.

I checked the barb solo era 11 leaderboard goes to 1000 (PrreciousRoy @ 1000). Era 10 goes to 2001 (The last entry is: AID)

4 Likes
9

Your table is wrong. It is not grift clear at equivalent paragon. It is grifts at the same clear efficiency.

Also I would be glad if you could remove that table from this thread and not re-use my data as another form of post etc.

3 Likes
10

You should add a similar table to your original post with the labels that you feel best describe the data. As you know, the quote below is a lot of words and can be summarized in a simple table.

One conclusion according to your analysis is that in terms of greater rift clearing efficiency of a wizard 130 level that barbs and necromancer have identical efficiency of 120.

4 Likes
11

Great work, Prokhan. Rage and I took a more conservative approach to ask for 2x more damage on top of the needed intra-class buffs, so that meant some builds need +7 GRs, and some builds need only +4. Even though we used a different method, it’s interesting to see we arrived at some eerily similar numbers.

I certainly wouldn’t turn down a +9 GR buff, though :smiley:

Just wanted to touch on a few things:

This is true, but it’s not only for R6 HOTA, and it may not be as slight as you think.

Without wall-charging, Vile Charge would lose a few GR levels against RGs without adds, and would suffer a slightly larger loss against RGs with adds. It’s difficult to calculate how much damage wall-charging adds up to in that build, but I think it’s fair to say it would take a 2-4 GR hit in clear potential.

For R6 HOTA, it’s much worse since you need those Raekor stacks and wall-charging allows you to exploit Endless Walk’s 100% multiplier. That build would likely lose 3-5 GRs of clear potential.

Though we didn’t publish it, we came to the same conclusion on an analysis of low and mid-end clears between classes. It was worth noting that several other classes perform better than Barb in those ranges, though.

3 Likes
12

HI Prokahn,

I now have used your method to look at all 7 classes and associated graphs (7000 data points). It took me a bit to figure out how you “scaled” your greater rift clearing efficiency coefficient but I used your necromancer data points to back calculate it based on grift clear 100 clears for the couple necromancers with the highest greater rift clear efficiency.

If you want to post your data for all classes in general discussion that you indicated that you were not planning on doing, you are more than welcome to. I probably won’t have time to show my work based on your method until mid-week after doing appropriate line fittings and making the visuals look pretty.

At this point, I could quote the number of posts on the forum and people would accuse me of having a secret agenda. P.S. I stopped posting the weekly forum topic/thread numbers after week 5. :sunglasses:

4 Likes
13

If you read 4) it says resulting numbers are scaled up by 1000000 (multiplied). This creates a large interval to ease comparison.

You are welcome to do it as long as you refer to these analyses and apply the method exactly. If you try to use a different formula express them in your methodology so that people can cross check it.

14

I initially converted clear time to minutes ( rather than milliseconds that is given in the API) that meant the scaling factor was off.

I think if I present this myself, I will change the scaling factor. When you made your inotial post, the numbers for grift levels and grater rift efficiency are too similar and can lead to confusion, especially quick reads of points 1-5.

4 Likes
15

I have no comments on barb. (I guess Don Vu has already been sent to the D4 group. I have absolutely no hope for barb. I have no confidence that the current dev has deep knowledge about statistics).

However, I do have some comments on the Necromancer curve.

The shape of Necros is significantly different than barb and wiz. Mainly because of the build they use. Thorns-Necros. Yes, the dev didn’t fix the “exploits” of skeleton thorns, hence make thorn dmg to mobs time based (Yes, this is unique among all builds). It’s like u have a BotS. What’s the result? As grift increase, say, 17% on mob health, for normal builds, clear time increase by 17%. However, for necro, clear time is increased by sqrt(1.17) - 1, around 8.5%.

If u consider RG, the thing is similar. Normal build is around 8.5%. But Necros, cubic root of 1.17 - 1. Nearly 17%/3.

Conculsion:

They can never match the three curves by just changing numbers I mean there is a big balance issue among especially the three classes, wiz(too strong), barb(also maybe monk, too weak), necro (game mechanic broken. I don’t mean everything is broken, I mean, there is some failures in the design.)

3 Likes
16

Yes Justin they cannot. Those curves will never match, but just a little approximation about where to go with buffs.

17

Sure. Move barb curve to the right. Where all three curves intersects at the same point. :slight_smile:

I mean actually, they should also have some modification on wiz curve and necro curve. Not buff or nerf, but mechanic changes.

1 Like
18

On second thought I edited my OP to include all the classes. However, I am not commenting much on them.

19

Included all the classes now and moved to General discussion.

1 Like
20

I tried to follow your method to reproduce your calculation. I have not done any serious line fitting; however, your findings are in agreement with mine.

From looking at your graphs, wizards are the '95 Chicago Bulls. Dependent on the grift level, monks or necromancers are the worst in terms of greater rift efficiency. Excluding wizards, the class balance is not as bad as I thought (all within 5 grifts).

Probably a wise decision, since your data show that barbarians are not the worse class.

4 Likes