[A Suggestion on Balance]

1 hour ago, whispersofblood said:

When it comes to data analysis volume is not depth. Especially in a data set that is constantly changing. Data from the summer is essentially useless beyond the most rudimentary conclusions. In my line of work using data as shown by THG would be laughable, these data sets need to be heavily curated, especially given that specific players are showing up multiple times in numerous data sets disproportionately.  The British data set is dominated by probably less than 30 players, and British data dominates the data set, the Aus/NewZ meta is similarly dominated and disproportionate. There is this high school math belief that aggregation averages out, it is incorrect. That only works as well as the strength of your data collection and robustness of the curation method. As is the current method would increase the representation of any skews in the data, skewing the results to be inline with the preferences of maybe 100 individuals globally. 

So you end up with problems like factions which appear to be good, get picked up by good players, which gives good results, which confirms the original assumption. Especially as a good player doesn't expect to go less than 4-1, which is already an 80% winrate. That's before you get into the context of the event, realm spells, artefacts. An event that uses realm spells and rotates realms, is a vastly different beast from an event that plays the whole event in Aqshy with realm spells, to an event with no realms at all. 

And, that's before we even get started on the assumption that the stats represent the "best build" without ever actually measuring the occasions any variation of what could be considered the best build shows up. Or even an estimation of what the "best build" would be as an archetype. The lack of rigor combined with the hyperbolic level of usage literally does my head in.

For these reasons and more I would heavily recommend people do not use TWG stats as anything more than an inference. 

My favorite quote "there are 3 types of lies... lies , damn lies, and statistics". THW has 13,490 matches recorded from over 69 events since drop of GHB 2019. 

In my line of work we use data sets to actuality make decisions and decide courses of action. What you propose is more of what we refer to as analysis paralysis. wanting a 100% perfect set of data before you can draw any conclusions is impossible in reality as you can never get it from something like this. If it is 80% direction-ally correct you can draw conclusions from the data.

Take storm cast vs skaven

Skaven make up 8.33% of the sample group with a win % of 55.6% 

SCE  are 9.21% with a win rate of 43%. 

We know Skaven have become much more competitive since the drop of their new tome. The data reflects that and the SCE under preform in comparison . sure there is going to be some noise in there but data always has it. 

can do same comparison to Night haunts vs DOK.  AoS is not well balanced for competitive play and the data shows it. 

Source data

https://thehonestwargamer.com/13th-december-stats/

[A Suggestion on Balance]

36 minutes ago, whispersofblood said:

They do not definitively show any such thing, there isn't nearly enough information collected in that data to be able to suggest that is true.  They at best imply imbalance, and at worst suggest there is something going on worth looking deeper at.

Win loss rates and dominate armies in the data make it clear that many armies perform sub par. It is a robust enough set of data that you can draw conclusions from it.  

39 minutes ago, Dead Scribe said:

The flaw with the honest wargamer stats is that they don't show you the context.  They don't show you the skill level of the players involved, and they are heavily weighted and influenced by volume of games played by people bringing what the meta considers OP.  

The honest wargamer stats contextually show what people are bringing to events, and overall how that faction does.  It highlights inferred imbalance.  It definitely points at trends of what the community feels is very weak and what is very strong.  

Player skill will always be a factor and is one with Infinity tracking as well. CB data is collected from tournaments like honest wargamer so it is comparable data. 

as above, if some one is bringing what is considered OP to skew the statistics, it actually validates that AoS is not balanced. There shouldn't be a perceived OP and it should not perform so well vs other armies. The significant win ratio actually means it has a performance advantage over other armies. 

 

[A Suggestion on Balance]

I play both infinity and AoS.

I play infinity competitively and enjoy it very much. I never feel like I do not have a chance to win a game due to army and it is all dependent on me and how i play (well and the dice to a point)

AoS I would never play competitively. It is not balanced nor really tries to be. So I play it just for fun. 

Back to the original subject of the post, GW could make all their games balanced and develop the tools to collect data. The honest wargamer stats show such disparity in win/loss due to the imbalance. GW has created this mess but can clean it up if they chose to do so. They have the resources but lack the will or desire to do so. 

The OP is a good example what can be achieved if balance was a goal.