Recommend
6 
 Thumb up
 Hide
6 Posts

Android: Netrunner» Forums » General

Subject: [OCTGN] ~72k anonymized games dataset rss

Your Tags: Add tags
Popular Tags: [View All]
Konstantinos Thoukydidis
Luxembourg
Esch-sur-Alzette
Luxembourg
flag msg tools
badge
Avatar
mbmbmbmbmb
Crosspost to the main site
6 
 Thumb up
 tip
 Hide
  • [+] Dice rolls
Anthony Giovannetti
United States
Woodinville / Bothell Area
Washington
flag msg tools
Avatar
Thanks for posting these!
 
 Thumb up
 tip
 Hide
  • [+] Dice rolls
Guido Gloor
Switzerland
Ostermundigen
Bern
flag msg tools
The statement below is false.
badge
The statement above is correct.
Avatar
mbmbmbmbmb
Ooh, neat site. I've added it to my feed reader
 
 Thumb up
 tip
 Hide
  • [+] Dice rolls
Frederic Bush
United States
Narberth
Pennsylvania
flag msg tools
For some reason I had trouble downloading the whole file; neither of my database programs would handle something that big.

As a result I started with "only" 65k games, and further confined myself to games where the new NBN identity showed up, making it only about 8600 games to analyze.

There are a couple issues with this dataset -- it doesn't adjust the corp score to account for flatline wins, and it also tracks "excess" score, such as when you score 8 or 9 points.

I adjusted things to look at tourney scoring -- any win was counted as 10 points. I used that to look at what I'll call the "tourney delta" for each faction -- If you win and your opponent has 5 agenda points, your tourney delta is 5 points, 10-5.

So:

H-B: Engineering the Future 53.3% win .53 tourney delta
H-B: Stronger Together 36.3% win -2.5 tourney delta
Weyland: Building A Better World 54.1% win .54 tourney delta
Weyland: Because we Built It 36.4% win -2.51 tourney delta
NBN: Making News 50.3% win -.18 tourney delta
NBN: The World is Yours 53.2% win .34 tourney delta
Jinteki: Personal Evolution 51.9% win .04 tourney delta
Jinteki: Replicating Perfection 43.1% win -1.41 tourney delta

overall corp 50.4% win -.11 tourney delta


Anarch: Noise 47.5% win -.47 tourney delta
Anarch: Whizzard 47.3% win -.33 tourney delta
Shaper: Kate 41.1% win -1.43 tourney delta
Shaper: Chaos Theory 43.0% win -1.14 tourney delta
Criminal: Gabe 54.5% win 1.2 tourney delta
Criminal: Andromeda 60.0% win 2.0 tourney delta


On the corp side, there's a couple clown IDs and everything else is competitive, but on the runner side, Criminal just has a colossal edge.

Also note that the corp is winning more matches but losing the war -- there are more runner blowouts than corp blowouts.

I am not a stats guy, so I can't calculate confidence intervals here, but all of the runner IDs and non-clown Corp IDs have 770+ games.


 
 Thumb up
 tip
 Hide
  • [+] Dice rolls
FirstName LastName
msg tools
mbmbmbmbmb
That's an interesting idea, but is the right dataset for this analysis? If you're building a tournament deck, you have to be conscious of the "tournament delta" behavior, but this is a sample of mostly non-tournament games right? So we can see the "accidental" tournament delta behavior of each faction, but perhaps not their fullest potential.
 
 Thumb up
 tip
 Hide
  • [+] Dice rolls
Frederic Bush
United States
Narberth
Pennsylvania
flag msg tools
DrTall wrote:
That's an interesting idea, but is the right dataset for this analysis? If you're building a tournament deck, you have to be conscious of the "tournament delta" behavior, but this is a sample of mostly non-tournament games right? So we can see the "accidental" tournament delta behavior of each faction, but perhaps not their fullest potential.
The raw winrates are up there as well.

It's possible there's some variable potential in how each ID can approach tourneys, but it's not immediately apparent to me that that's true. And unfortunately, without a comparable large db of recent tourney games, I don't see a way to examine your suggestion.
 
 Thumb up
 tip
 Hide
  • [+] Dice rolls