Win ratios - Overall and faction specific

Here have some pretty numbers from 5 months ago: http://forums.poxnora.com/index.php?threads/2015-03-to-2015-09-player-one-wr-map-data.18728/

 

Yeah I saw that back in the day - so I know this data is available somewhere.

Without taking away from Sok's great work, I think I could do a lot better After all this is far from his primary responsibility whereas it's how I make my living!

And again I think the biggest improvement I'd like to make is releasing it in a user-queryable form - think less spreadsheet and more interactive reporting tool. Exponentially multiplies the amount of insights you can glean from data - especially with a userbase as analytical as Pox's.

 

I am not sure that is really something that would be useful or helpful. I mean it would be cool, and could potentially be very useful internally, but I am not sure making this public-facing would be beneficial for anyone involved.

 

Why's that? For instance take the thread Markoth linked - my impression of that is it led to many interesting and informative discussions from players re balance, considering factors that we hadn't previously weighted as appropriately, such as P1/P2 over faction/map weightings. And as discussions went on, updated reports allowed participants to directly assess the truth of their predictions, and inform future decisions.

Some potential benefits, in my opinion:

• Figures to decisively answer whether X rune/map/IS is OP
  • Better for players - less reliance on anecdotal his-experience-vs-mine arguments
  • Better for DOG - less reliance on the same (of course they should segment for higher-ranked players when informing balance changes. And of course consider input from the Council as there will always be a human element)
• Detailed metagame analyses - again this has already been done but I hope to do it in a way that is more useful to all parties, and easier to keep updated.
• Faster validation of patch impacts - especially targeted nerfs/buffs
• Crowdsourcing of analysis - 10s of people getting + sharing answers to their own questions, leading to others building off original questions and exploring, etc. instead of 1-2 guys releasing sporadic static reports.
• Enriches the appeal of Pox to hardcore competitive strategy gamers - which let's face it is what will keep this playerbase from going under in the present and future.

Some potential drawbacks:

• Data confidentiality (if this is a concern player data could be anonymised, perhaps remove identifiable data for all players eg w/l records if DOG has serious concerns)
• Poor methodology leading to skewed findings (but as long as all analyses are shared and transparent, the result can never be worse than no data whatsoever)

To clarify: I do not plan on releasing anything along the lines of a paper saying "X is OP, here are my calculations proving this." Instead I hope it would be more like "Here is a tool where anyone can drill into any rune and see the W/L % of all BGs containing it, and how that varies when combo'd with other runes or against certain factions, etc." Not pre-calculated readouts but dynamically generated database queries. I would ideally minimize my own opinions and release the data (again, scrubbed of sensitive factors) to the playerbase.

 

My major concern (and I have a bunch of concerns) is that something like this would result in people taking results that are not statistical significant and use them as a basis for an argument that has no real statistical basis. Over and over and over again. To be quite honest the population of pox is small enough, and the number of people using any given combination of runes is small enough that I would be impressed if anything at all of statistical significance could be derived from the data.

 

You're absolutely right - sample sizes are small especially for individual non-ubiquitous runes.

If this were going to an academic journal I'd be much more hesitant to participate. I also wouldn't want to see random %s cherry-picked and slung left and right. But it'd be better than the nothing (or near nothing) we have to go on right now, beyond personal anecdotes. And transparency in methodology goes a long way in keeping yourself and others honest.

Way I see it, endless shitposts and back-and-forth zealotry demonstrably happens anyway. It is true that adding data to the equation might not even dent this in future. But it certainly shouldn't exacerbate it - it just changes "IS is OP" to "IS is OP because Dumb Statistic X". But aside from that it would be contributing great value to those who can interpret with a clear head. And in that light any value add at all makes this worth it, considering it costs no one anything

With or without data there are still going to be dummies and zealots, pros and analysts. Dummies gonna dum regardless, so let's see what we can do for the other side instead.

 

Much appreciated Sok, that would be amazing. I've been playing on and off since 2010, never top ranked but I do feel some attachment to Pox and would love an opportunity to give back

Happy to lock down the data, run results by you first, whatever else you think is prudent. Just wanna get my grubby hands on something and start messing with it.