A comparison of how builds function in Median XL from 1-130

Discuss Median XL!
midknight
Pit Knight
106 | 59
Common Popularity Badge
Has a thread with over 10.000 views
Common Love Badge
Earned over 20 cookies
Common Supporter Badge
Donated 1 time
Common Guide Badge
Created a complete character guide
A comparison of how builds function in Median XL from 1-130

Big Thanks to Aahz for all the effort in coding and collecting all the data provided!

Image

Median XL has over 70 builds and creating an environment where all builds have the potential to do 1-130 plus ubers equally as well is incredibly tough. I personally dislike the use of the word balance for me it has a connatation that every thing is equal. The concept of balance is contradictory by nature. Balancing the game around one group such as the average gamer might completely imbalance the game for another group the elite tier speed runners or elite tier lab runner. Whilst balancing the game around the elite tier player or an elite tier speed runners will put off the game for the average gamer or casual gamer for the difficulty being too hard. I think its fair to say that Median Xl wants to maintain the current player base and keep them challenged with a changing meta whilst also bringing in new players at the same time who can enjoy the mod.

People have a tendency of complaining how bad specific builds or their favourite build feel without any real comparison yardstick. Hopefully these metrics can give a point of comparison and people can see there is not such a huge disparity between the builds as it seems or in some case there is a huge disparity :bounce: . It is best not to treat this data as black and white especially those with small sample sizes. Uber bossing are not included which is an integral part of the mod. Skill, knowledge and experience and afking are more likely to affect small sample sizes.

Rather than looking at overall balance, hopefully the data will show the different strengths and weaknesses of each build and with different questions and challenges being posed at different phases of the game. Hopefully with the data collected it can show which builds might need a reduction of its power in a particular phase of the game for being overtuned or potentially looking to buff the undertuned characters that struggle a lot.





The Data
Considering that 70% of people quit the game at level 120. I have grouped up the data into three categories early game will be from 1-100,100-120, 120-130. I have split up the data as best I could whilst trying to keep the integrity of data. I will make the assumption that people reading this have a general understanding of basic statistics and data collection. This data is not independently identically distributed so performing any kind of distribution analysis doesnt really make any sense. I have just used simple robust estimators such as the mean, percentile average to have a look at the data. This data will of course contain sampling bias.

Modern statisticians are familiar with the notion that any finite body of data contains only a limited amount of information on any point under examination; that this limit is set by the nature of the data themselves, and cannot be increased by any amount of ingenuity expended in their statistical examination: that the statistician's task, in fact, is limited to the extraction of the whole of the available information on any particular issue.

I hope some of this is useful to the devs for considering changes and balancing to the game rather than seeing an amazon rework and melee sorc buffs every ladder :bounce: in 2.x .

Note: Bubble charts are included in googlesheets linked at the end.



Ladder Start
One of the most important parts of the ladder where the player base is at its highest and the challenge a fresh ladder brings. Data has been split into three parts: The first three days of ladder from 1-100, The first two days of the ladder 100-120 and 120-130. (Probably my favourite part of the ladder due to the challenges it brings for HC, optimisation with limited resource, plus ubers with unoptimised gear).

Data is split into three parts: Early ladder as a whole, the elite players and the exceptional players groups these players have the greatest influence of how a ladder progresses.

Ladder Start all
1-100 1504 people on ladder start (first 3 days)
► Show Spoiler

100-120 307 players on ladder start (first 2 days)
► Show Spoiler

120-130 46 players on ladder start (first 2 days)
► Show Spoiler



Ladder Start exceptional tier
1-100 Sub 5 hour characters on ladder start 354 players
► Show Spoiler

100-120 Sub 5 hour 40mins 100 characters on ladder start
► Show Spoiler

120-130 Sub 10 hours 40 mins 40 characters on ladder start
► Show Spoiler





Ladder Start elite tier
1-100 Sub 3 hours 30 minutes 51 players
► Show Spoiler

100-120 Sub 4 hours 30 minutes 30 players
► Show Spoiler

120-130 Sub 7 hours 25 minutes 20 players
► Show Spoiler




Top Builds across the whole ladder
Some people prefer to enjoy a long ladder play different build or min max the same one. Most likely tweaked setup and potentially players on sc were using nithogur charm.
Top 100 players across the whole ladder
1-100
► Show Spoiler

100-120
► Show Spoiler

120-130
► Show Spoiler



Top 10 percentile of all players across the whole ladder
1-100 689 characters
► Show Spoiler

100-120 387 characters
► Show Spoiler

120-130 214 Characters
► Show Spoiler



Builds people struggle with
This data looks at the bottom 10% percentile of players and which builds they have struggled on the most.

1-100 bottom 20 builds
► Show Spoiler

100-120 bottom 20 builds
► Show Spoiler

120-130 bottom 20 builds
► Show Spoiler




Notes on the data collected:
sum is the number of players doing that build
I have not done an analysis on endgame so builds could function completely differently post 130.
Bottom 10% of time were removed to remove the impact of people afking or just being idle.
Some overlapping data, Warlock and Rising Dawn from 100-120, 120-130. Hammerzon, Stormzon, Bloodzon, MMzon and throwsin.
Anything less than 10k kills were considered to be tweaked or team runs and were not included from 1-100
Anything less than 16k kills were considered to be tweaked or team runs and were not included from 100-120
Anything less than 30k kills were considered to be tweaked or team runs and were not included from 120-130
Trophy farming from 1-130 can have a potential impact on the data. Although most people tend to farm trophys post 130 apart from TA.
Nephalim paladin is probably an outlier due to nithogur charm and nithogur charm could influence softcore time, although looking through the data it did not seem to have a huge impact imo.
Any transitions in data i.e claw to nagi have been treated just as nagi there were a handful of these and made the data rather convoluted to look at.
People might be change builds in the run most notably due to 0 skills or speed running optimisation.
Top 100 times are heavily influenced by Lynderika with 5 out the top 10 times. Lynderika's times are outliers.
Ladder start is the first three days of the ladder for 1-100 because it is not so heavilty influenced by tweaked gear
Ladder start is the first two days of the ladder for 100-120,120-130 because of the influence of tweaked gear on day 3

Thanks to Lynderika and Knack for some input and things I overlooked!
Thanks to Devs for this great mod!

https://docs.google.com/spreadsheets/d/ ... edit#gid=0
https://docs.google.com/spreadsheets/d/ ... =631941428
https://docs.google.com/spreadsheets/d/ ... d=54765529
Edited by midknight 10 months.
midknight
Pit Knight
106 | 59
Common Popularity Badge
Has a thread with over 10.000 views
Common Love Badge
Earned over 20 cookies
Common Supporter Badge
Donated 1 time
Common Guide Badge
Created a complete character guide
Comparison data of the meta for each class in different level range
Some interesting data although unsure how relevant it is.

Amazon
► Show Spoiler


Assassin
► Show Spoiler

Barbarian
► Show Spoiler

Druid
► Show Spoiler

Necromancer
► Show Spoiler

Paladin
► Show Spoiler

#BuffMeleeSorceress
► Show Spoiler



Absolute times ladder start 1-100 and 100-120
► Show Spoiler
Edited by midknight 10 months.
midknight
Pit Knight
106 | 59
Common Popularity Badge
Has a thread with over 10.000 views
Common Love Badge
Earned over 20 cookies
Common Supporter Badge
Donated 1 time
Common Guide Badge
Created a complete character guide
I refrained from analysis earlier to let people look at the data themselves. These were the interesting point for me. I am sure I have missed other interesting points as well :( feel free to add comments.

Overall Analysis:

  • This is NOT a tier listing, the fact that nature was the top build in both exceptional and elite ladder start shows some of the problems with treating the data as so. Lynderika's performance outlied the data so much that it pushed its performance significantly above the rest of the other builds. Especially with small numbers these numbers are likely to be exaggerated.
  • People on the whole dont seem to follow trends as much as anticipated, the three main builds that were changed the most this season were melee sorc, bowazon and incarnation. Absolute numbers of these three classes were 329/ 1504 people on ladder start 1-100. Although the other popular classes that had some kind of rework since 2.0 still remain popular: Summons, Shaman, Bloodzon, Javazon 420/1504 people. (Its great to see that people enjoyed these changes and came back :D .) In my opinion people tend to favour specific classes or specific builds they enjoy on ladder start.
  • On ladder start the slowest player in the 20th percentile of 1-100 take 5hours to get from normal to hell. On ladder start the slowest player in the 33rd percentile from 100-120 takes around 5 hours 30 minutes including 105/110/115 charm farming or not.
  • Absolute times from 1-100 and 100-120 show that the data is actually pretty well normally distributed with a right skew, this is probably due to the affect of afking.
  • The bubble graphs in google docs show that most builds perform within a pretty similar range from 1-100 and 100-120. :D
  • Some data points might not show up like poison druid because they are not popular.
  • People tend to stick to one build.
  • People have very different experiences of Median XL.
Specific Builds:

  • Skill/Experience builds: From 1-100 and 100-120 naginata, warlock, ritualist warlock, claw, werebear, melee necro look to be more skill/experience base build. It looks like the weaker player overall dont perform particular well but for a person who knows what they are doing it performs reasonably well in comparison to other builds.
  • On ladder start malice, crossbow, elementalist barb and traps do not perform particular well at both the 1-100 range and 100-120 range.
  • On ladder start ritualist warlock (my league starter) had a 47 player count 1-100, one of the slowest builds with a large sample size. With the exception of Usling and I none of those 45 remaining players managed to make it to 120 on the second day. I personally dont recommend it for new players unless they are playing hardcore since I dont think this build is easy to play for new players. Pure ritualist without warlock struggled even more.
  • Bowazon was a dominant force this ladder from the 100-120 and 120-130 range.
  • Melee sorc dominates the sorc tree in every aspect from 1-100, 100-120, 120-130 and kpm its dominance grows at every phase in the game. Sorc looks to be the most out of sorts class :(.
  • Fire,poison, arcane(low numbers),cold sorcs all look to perform pretty weak from 1-100 on ladder start and bottom 10% of builds, Although from 100-120 cold sorc looks like it performs significantly better than the others.
  • Barbarian looks to have the highest proportion of 0 skill builds probably guard tower players.
Edited by midknight 10 months.
User avatar
Lynderika
Grubber
315 | 160
Legendary Popularity Badge
Has a thread with over 250.000 views
Great Love Badge
Earned over 100 cookies
Great Guide Badge
Created 3 complete character guides
So much more people than I expected level as throw sin, and so few as earthshake barb, holy weird

On another note... #buff melee sorc and bowa
They were new buff this patch, nobody was used to it, many tried and yet despite their high playcount their average performance stick out. Truly stronk
User avatar
Lynderika
Grubber
315 | 160
Legendary Popularity Badge
Has a thread with over 250.000 views
Great Love Badge
Earned over 100 cookies
Great Guide Badge
Created 3 complete character guides
Another thing I find surprising is that the free set had pretty much made no difference in players choice, and there's no sign of anyone doing something to 120 to respec to the set build for 120-130. Java and nagi were super good sets too. holy melee pal, werebear or shaman even lost playerbase, though for thoses it's understandable
BearBearderBarb
Pit Knight
115 | 7
A lot of sets are shit even at level 100, let alone 120+. So a build which wasn't interesting in the first place isn't going to be more attractive with a guaranteed set drop even with a great set. Not to mention running the nemesis level is annoying as fuck. I guess I was lucky to run it in total 2 times with both tries having the nemesis present, because apparently it only has 20% chance to spawn. So it's quite an investment of time, crystals and leveling another char just to potentially maybe not having a mediocre result. I'd rather spend the crystals on jewel rerolls.
User avatar
Sardonic
Gravedigger
194 | 8
Lynderika wrote:Another thing I find surprising is that the free set had pretty much made no difference in players choice, and there's no sign of anyone doing something to 120 to respec to the set build for 120-130. Java and nagi were super good sets too. holy melee pal, werebear or shaman even lost playerbase, though for thoses it's understandable


I agree, I find it surprising that the data doesn't show as much correlation to the free sets. One point I will say is that, I struggled to kill the assassin as nagi before 130.
midknight
Pit Knight
106 | 59
Common Popularity Badge
Has a thread with over 10.000 views
Common Love Badge
Earned over 20 cookies
Common Supporter Badge
Donated 1 time
Common Guide Badge
Created a complete character guide
Added absolute time graphs for ladder start 1-100 and 100-120 added at the end.
Added analysis at the end.
bohoTV
Invader
8 | 2
Nice info.

midknight wrote:[*] On ladder start ritualist warlock (my league starter) had a 47 player count 1-100, one of the slowest builds with a large sample size. With the exception of Usling and I none of those 45 remaining players managed to make it to 120 on the second day. I personally dont recommend it for new players unless they are playing hardcore since I dont think this build is easy to play for new players. Pure ritualist without warlock struggled even more.
[/list]


I mentioned this in the tierlist section, but UHM is a genuinely terrible class, softcore or hardcore, in desperate need of a rework. The fact it's recommended to new players is criminal: it's slow and painful from 1-100 and just gets worse from there. But why elaborate when we can let your numbers speak for themselves: if 2 of 47 at ladder start made it to 120, that means 96% of (ladder start) UHM players who made it to 100 abandoned the class before 120.

Yikes.
Bliyadolubov
Grubber
314 | 10
bohoTV wrote:Nice info.

midknight wrote:[*] On ladder start ritualist warlock (my league starter) had a 47 player count 1-100, one of the slowest builds with a large sample size. With the exception of Usling and I none of those 45 remaining players managed to make it to 120 on the second day. I personally dont recommend it for new players unless they are playing hardcore since I dont think this build is easy to play for new players. Pure ritualist without warlock struggled even more.
[/list]


I mentioned this in the tierlist section, but UHM is a genuinely terrible class, softcore or hardcore, in desperate need of a rework. The fact it's recommended to new players is criminal: it's slow and painful from 1-100 and just gets worse from there. But why elaborate when we can let your numbers speak for themselves: if 2 of 47 at ladder start made it to 120, that means 96% of (ladder start) UHM players who made it to 100 abandoned the class before 120.

Yikes.

May be UMP is bad build, but i personaly see no isue in it, i played it from level 1 and didnt sweat compare to HMP. The build tree didnt change so much since sigma release. If you dont know how to not struggle it doest meat that build is bad, may be it bad only for you