Competitive Top Tier Armys

[SleeperAgent]

22 minutes ago, Dracan said:

I took the Honest wargamer stats and made a quick graph of their AOS 2.1 vs AOS2.2 comparison for 6 May 2019 table leaving out the "non army" armies and got to this.

One can prob with certainty say Daugters to Idoneth are pretty solidly top tier with win ratios over 60% which is way too high, second tier is probably skaven to Beastmen which might need slight point adjustments, Tier 3 with Bonesplitters to Seraphon who prob need slight point drops, and last tier of armies not even competing, luckily most of them still will get new books but it does show Gloomspite and NH where pretty shafted in their bbooks.

More proof that Nighthaunt needs a buff or rework. We got a 2.0 book before all the good stuff came out. Should be handing out -3 bravery naturally and even more spells and abilities on top of it. 

[BobbyB]

Ah yes that famously second tier and fine army Skaven

[Dracan]

Skaven are very very good... but they haven't gotten the 60%+ win ratio in more than 70 games. I still think that t2 is very good, it is just that t1 is waaaaaaaaay too good, no army should have win ration over 55% in more than 50 games in a "global"  meta.

[sal4m4nd3r]

1 hour ago, SleeperAgent said:

More proof that Nighthaunt needs a buff or rework. We got a 2.0 book before all the good stuff came out. Should be handing out -3 bravery naturally and even more spells and abilities on top of it. 

I think if nighthaunt units was never allowed in a LoN army then more people would have played nighthaunt. And the win ratio would have gone up. I think the better players realized you can basically play nighthaunt with the best parts of Nagash and ROLL. Nighthaunt had a great deal in the starter box, and they are very easy to paint, lending themselves to beginners and less experienced players. 

Letting LoN armies play with nighthaunt, and more specifically,  grimghast reapers...was a very dumb and short sighted decision.  My theory is to drive sales of the Soul Wars box. But why play nighthaunt allegiance.. when for the ridiculous cost of 1 cp.. you can just bring back a full unit of already powerfully efficient ghosts by playing LoN? Because you are new/narrative minded and dont really care or know better. 

Edited May 15, 2019 by sal4m4nd3r

[SwampHeart]

4 minutes ago, sal4m4nd3r said:

My theory is to drive sales of the Soul Wars box.

I try not to jump on board conspiracy theories (especially ones regarding GW's competence) but this feels like what happened. They wanted to give a huge group of people a reason to buy the starter so they took a unit that is clearly balanced for NH and allowed it in LoN. And in doing so they effectively offset every possible weakness of that unit and created a scenario where NH is better in LoN than on its own. 

[whispersofblood]

2 hours ago, Dracan said:

I took the Honest wargamer stats and made a quick graph of their AOS 2.1 vs AOS2.2 comparison for 6 May 2019 table leaving out the "non army" armies and got to this.

One can prob with certainty say Daugters to Idoneth are pretty solidly top tier with win ratios over 60% which is way too high, second tier is probably skaven to Beastmen which might need slight point adjustments, Tier 3 with Bonesplitters to Seraphon who prob need slight point drops, and last tier of armies not even competing, luckily most of them still will get new books but it does show Gloomspite and NH where pretty shafted in their bbooks.

What controls are being used to analyze this data, to minimize the affects of bias? What non-rules factors were considered? 

[sal4m4nd3r]

1 minute ago, whispersofblood said:

What controls are being used to analyze this data, to minimize the affects of bias? What non-rules factors were considered? 

Its just based on objective tournament data. Numbers at the top are games win percentages. Blue bars are percentage of people who bring those armies to events. 

[Ravinsild]

1 hour ago, sal4m4nd3r said:

Because you are new/narrative minded and dont really care or know better. 

stop bullying me. I chose my skeletons for death on purpose. I don't like spooky ghosts I like normal skeletons  lul 

Oof my narrative/this looks cool streak will be the death of me in this game. 

I also think the Idoneth thing is frustrating. As I have said, I love using cool models and spamming 1 model over and over is boring, yet the Idoneth Deepkin thread is literally dead. The army is solved. It's extremely binary "Bring Morsarr Guard eels, or lose." So IDK aren't even good. They just have literally 1 super way too good unit that makes literally everything else (RIP Turtle, Mathalann, Thralls, Sharks, the literal rest of the range) objectively worse than it. IDK players are worried how they will even play when inevitably the Morsarr Guard eels are nerfed. 

Edited May 15, 2019 by Ravinsild

[whispersofblood]

9 hours ago, sal4m4nd3r said:

Its just based on objective tournament data. Numbers at the top are games win percentages. Blue bars are percentage of people who bring those armies to events. 

But... that is statistically unimportant data. It doesn't even evaluate for strength of schedule. Every win is equally valued, regardless of score, player, opposing faction. Faction strengthen is a hugely complicated thing to calculate based on outcomes, and this doesn't come anywhere close to quality being data worth actioning. 

Its not actually objective in even the most generous version of the word, equally valuing every win makes this data essentially meaningless.

[stratigo]

It also takes several months for an army to be “solved” as it were, so I suspect skaven, gobbos, and flaming dwarves are not fully accurate to the quality of their books quite yet. It takes time to both learn how to play and how to play against a new book

 

 

[PlasticCraic]

Does the Skaven figure include a lot of pre-Battletome results pulling the average down? 

I'd be interested to see the pre- and post-Battletome win rates for them and FEC.  

[Dracan]

7 hours ago, whispersofblood said:

But... that is statistically unimportant data. It doesn't even evaluate for strength of schedule. Every win is equally valued, regardless of score, player, opposing faction. Faction strengthen is a hugely complicated thing to calculate based on outcomes, and this doesn't come anywhere close to quality being data worth actioning. 

Its not actually objective in even the most generous version of the word, equally valuing every win makes this data essentially meaningless.

While this is true, one also has too start assuming that after 50-100 games with random people these types of factors are starting to be averaged out, especially when you consider it is a 3 continent spanning data set and not an isolated closed meta of the same 10 players.

It is also the only real data available, it doesn't prove much and is definitely not definitive of the state of any isolated meta, but gives an point of reference as to the average strength in a normal tournament setting for armies on average.

The only real way you are going t get the "scientific" data you are looking for is if you record the 2 same opponents playing every variation of an army vs every variation of another army in each of the 18 missions at least id wager 20 odd times... this is impossible...

5 hours ago, PlasticCraic said:

Does the Skaven figure include a lot of pre-Battletome results pulling the average down? 

I'd be interested to see the pre- and post-Battletome win rates for them and FEC.  

Skaven doesnt include any data from before the book and they have almost 80 games on record so could probably go or down up a few % points but should be pretty stable. FeC and fyreslayers is both pre and post tome so does muddy the data.

Will be very interesting to see what the new Legion of Grief does for Night Haunt...

Just as a side note I'm not involved in any way with the stats i just made a little graph 😜 

[XReN]

1 hour ago, Dracan said:

Will be very interesting to see what the new Legion of Grief does for Night Haunt...

I suspect a nail on the coffin of their original allegiance, to be completely honest.

[Tropical Ghost General]

48 minutes ago, XReN said:

I suspect a nail on the coffin of their original allegiance, to be completely honest.

100% this. But I'll keep on playing pure spooks because I don't like winning games 😋.

[whispersofblood]

3 hours ago, Dracan said:

While this is true, one also has too start assuming that after 50-100 games with random people these types of factors are starting to be averaged out, especially when you consider it is a 3 continent spanning data set and not an isolated closed meta of the same 10 players.

It is also the only real data available, it doesn't prove much and is definitely not definitive of the state of any isolated meta, but gives an point of reference as to the average strength in a normal tournament setting for armies on average.

The only real way you are going t get the "scientific" data you are looking for is if you record the 2 same opponents playing every variation of an army vs every variation of another army in each of the 18 missions at least id wager 20 odd times... this is impossible... 

That is just not true. Its the easiest data available it still doesn't mean you can or should draw any inferences from it. Aggreggated data does not average out bias it usually just reinforces or exaggerates the affects. And, that is assuming that the 50\50 win percentage is even what should be pursued as balanced. 

There are several issues with using this model, the biggest being that metas are not constant and we can't model how quickly they change or how quickly people can get new armies for new books done, and update their armies in respose to new books. How many of the IDK measurment are wins in an unprepared meta, against factions with known issues, againsts players with no experience. All of which have almost nothing to do with the strength of IDK or or Morrsarr.

The truth is you have a graph that doesn't exclude anything so realy doesn't measure anything anyway. 

[Dracan]

13 minutes ago, whispersofblood said:

That is just not true. Its the easiest data available it still doesn't mean you can or should draw any inferences from it. Aggreggated data does not average out bias it usually just reinforces or exaggerates the affects. And, that is assuming that the 50\50 win percentage is even what should be pursued as balanced. 

There are several issues with using this model, the biggest being that metas are not constant and we can't model how quickly they change or how quickly people can get new armies for new books done, and update their armies in respose to new books. How many of the IDK measurment are wins in an unprepared meta, against factions with known issues, againsts players with no experience. All of which have almost nothing to do with the strength of IDK or or Morrsarr.

The truth is you have a graph that doesn't exclude anything so realy doesn't measure anything anyway. 

So just disregard the only real numbers available because it might not be 100% accurate or representative of every meta... Isn't it interesting then how peoples views with no stats correspond roughly with perceived power from the empirical useless data...

Im not saying its perfect never did, but to ignore it completely, is folly.

[whispersofblood]

1 minute ago, Dracan said:

So just disregard the only real numbers available because it might not be 100% accurate or representative of every meta... Isn't it interesting then how peoples views with no stats correspond roughly with perceived power from the empirical useless data...

Im not saying its perfect never did, but to ignore it completely, is folly.

Well in short? Yes. Its probably doing more harm than good, because the data is misleading and not minimally informative. 

Long form. Basically you are measuring the total number of spiders globally seeing that some spiders are growing faster and suggesting that since these spiders seem to be growing faster they must be x because that is what you decided before hand you were looking to find out. But that isn't how stats work, because you haven't generated the data in a way to measure that.

But if you are actually interested in getting results on faction strength you need to have heavily constrained data sets.

Constrained by time period, because the types of competition change drastically over time.

You need to quantify different game results, because if IDK are winning a large portion of their matchs by 3 or less points after 5 turns then the book is probably fine, if they are blowing out all of their wins in three tuns then we might have a problem.

you need strength of schedule to reduce the variance of in game difficulty between quality of players and the quality of factions

you need to know the subfaction lists are using to even begin to have real credibility about the mechanics of the faction. Fuethan play way differently than other enclaves and are more powerful given the culture of the game at the moment

[Mayple]

I agree with @whispersofblood on pretty much all points there. Was going to write a thing, but it's been covered neatly. 

[Joseph Mackay]

i agree that not enough information is being collected or considered with regards to these sorts of stats.

the one that really drives me up the wall is Grand Host Of Nagash, Legion Of Sacrament, Legion Of Night and Legion Of Blood all being recorded just as Legion Of Nagash which is skewing the stats for that book

i also think the sub factions are very important to consider in this regard too

[prochuvi]

On 5/15/2019 at 12:37 PM, Dracan said:

I took the Honest wargamer stats and made a quick graph of their AOS 2.1 vs AOS2.2 comparison for 6 May 2019 table leaving out the "non army" armies and got to this.

One can prob with certainty say Daugters to Idoneth are pretty solidly top tier with win ratios over 60% which is way too high, second tier is probably skaven to Beastmen which might need slight point adjustments, Tier 3 with Bonesplitters to Seraphon who prob need slight point drops, and last tier of armies not even competing, luckily most of them still will get new books but it does show Gloomspite and NH where pretty shafted in their bbooks.

I think is safe say free sumon is umbalanced (as every that is free)

The god tier and 2nd tier are all free sumon armys,only dok and idoneths are there without it(even idoneths have free sumon with trhalls coming back but i wont call them free sumon army).

I dont know nothing about khaos dwarf so i dont know where they are.

For me it is so easy how:

get back free sumon only to opens play as allways have been

Nerf every dok aura to wholy within. 100 points hag queen and 120 wytches

Nerf idoneths eels

 

With those easy fix the meta gonna be so much balanced 

[Dracan]

1 hour ago, whispersofblood said:

Well in short? Yes. Its probably doing more harm than good, because the data is misleading and not minimally informative. 

Long form. Basically you are measuring the total number of spiders globally seeing that some spiders are growing faster and suggesting that since these spiders seem to be growing faster they must be x because that is what you decided before hand you were looking to find out. But that isn't how stats work, because you haven't generated the data in a way to measure that.

But if you are actually interested in getting results on faction strength you need to have heavily constrained data sets.

Constrained by time period, because the types of competition change drastically over time.

You need to quantify different game results, because if IDK are winning a large portion of their matchs by 3 or less points after 5 turns then the book is probably fine, if they are blowing out all of their wins in three tuns then we might have a problem.

you need strength of schedule to reduce the variance of in game difficulty between quality of players and the quality of factions

you need to know the subfaction lists are using to even begin to have real credibility about the mechanics of the faction. Fuethan play way differently than other enclaves and are more powerful given the culture of the game at the moment

I do not get the point of your argument on how this pertains to tiers in AoS in a competitive tournament scene... other than a lecture on statistics which tbh i agree with on the integrity of the data... but what is your point though?

We are discussing the fictional tier system of little toy soldier fantasy armies with almost infinite combinations of variables when you include the human factor.  Do you agree sortof with the order in which these armies fall?  if not other than saying stats are made up and points dont matter how do you rank the armies?

You must also remember that data isnt based on some backwater basement games, most if not all the tournaments they count are pretty darn big affairs with serious gamers trying to do their best, thus we can also assume they are mostly using the top ends of the book. and yes assume and stats and integrity dont go well together but again it is the only numbers we have.

[XReN]

1 minute ago, prochuvi said:

I think is safe say free sumon is umbalanced (as every that is free)

Can you point me what armies have free summon?

[prochuvi]

1 minute ago, XReN said:

Can you point me what armies have free summon?

God tier:

Dok(havent)

Fec(best example of umbalanced free sumon with a 200 points hero bringing free 200 points of ghouls)

Slanesh(free sumon)

Idoneths (i put them as non free sumon army because it isnt used in tournaments,but they can free sumon with revive thralls)

 

2nd tier:

Skavens(free sumons)

Khaos dwarf (i havent idea)

Legion of nagash(best example of broken umbalanced free sumon together with fec)

Tzenth (free sumon)

Beastmens(free sumon)

 

Every army  in the game that have in the game units that havent paid points for them is free sumon

[Mayple]

17 minutes ago, Dracan said:

You must also remember that data isnt based on some backwater basement games, most if not all the tournaments they count are pretty darn big affairs with serious gamers trying to do their best, thus we can also assume they are mostly using the top ends of the book. and yes assume and stats and integrity dont go well together but again it is the only numbers we have.

Incorrect. I was at one of those events, and it included everything from bottom to top tier players, which all counted towards the statistic. 

Not to say none of the events are "top tier competitive only", but it certainly does not apply to all of them, which skews the statistic if you treat it as a representation of that. 

[XReN]

2 minutes ago, prochuvi said:

Every army  in the game that have in the game units that havent paid points for them is free sumon

False, free means free, FEC pay a hefty battleline tax and have overpriced Serfs and Knights units and sacrifice first turn of moving a valuable support hero(es) which can also be killed before they summon, LoN is not summon, it ressurrecting that costs CPs and have a hell of a lot restrictions that can be worked around to prevent using it. Skaven have to take specific units to summon anything, same goes to Seraphon who can summon an insane amount of units but not anywhere high on tournament placings, only Nurgle, Slaanesh, Tzeench, Khorne and Beastmen are rewarded with summoning points no matter how you play the game, but I believe there is tax somewhere. Gloomspite Gits can ressurect half a unit a turn, potentially, whille IDK need to take specific hero to return models and that is this hero's only purpose.

It's also worth mentioning that scraping reinforcements points came with GHB where points were changed accordingly to how GW saw fitting with new edition and rules. 

So in my opinion free summoning is rare and not-that-good.