Dominion Strategy Forum

Please login or register.

Login with username, password and session length
Pages: [1]

Author Topic: Interpreting CouncilRoom stats: Huge 'Effect With(out)'  (Read 3552 times)

0 Members and 1 Guest are viewing this topic.

theschizm

  • Pearl Diver
  • **
  • Offline Offline
  • Posts: 14
  • Respect: +8
    • View Profile
Interpreting CouncilRoom stats: Huge 'Effect With(out)'
« on: January 10, 2012, 11:18:30 am »
0
Long time lurker, first time poster. I took a look at my CouncilRoom stats and I noticed some strangely high values in my Effect With and Effect Without columns. It says I do 26.99 standard deviations better than normal when I buy Woodcutter (a card I really dislike); nothing is even close to that, with Farmland in second with 4.45. I have three cards with a huge Effect Without, King's Court, Fishing Village, and Tunnel are all in the 20s, and Forge with 5.22 is next.

What's causing those huge numbers? Am I that much better at playing Woodcutter and ignoring those other three than most people? I doubt it.
Logged

WanderingWinder

  • Adventurer
  • ******
  • Offline Offline
  • Posts: 5275
  • ...doesn't really matter to me
  • Respect: +4386
    • View Profile
    • WanderingWinder YouTube Page
Re: Interpreting CouncilRoom stats: Huge 'Effect With(out)'
« Reply #1 on: January 10, 2012, 11:24:20 am »
0
Quote from: theschizm on January 10, 2012, 11:18:30 am
Long time lurker, first time poster. I took a look at my CouncilRoom stats and I noticed some strangely high values in my Effect With and Effect Without columns. It says I do 26.99 standard deviations better than normal when I buy Woodcutter (a card I really dislike); nothing is even close to that, with Farmland in second with 4.45. I have three cards with a huge Effect Without, King's Court, Fishing Village, and Tunnel are all in the 20s, and Forge with 5.22 is next.

What's causing those huge numbers? Am I that much better at playing Woodcutter and ignoring those other three than most people? I doubt it.

Your numbers are all pretty extreme, as a function of having not played many games. Even numbers up in the 2s are pretty big, and you have loads of these.
But the reason for all those cards in the 20s can be explained by looking at the win rates in those situations. You've won every game where you had woodcutter, and you've also won every game you haven't gotten KC, FV, and tunnel. Those aren't many games, but 100% is pretty warping to things.
Logged

theschizm

  • Pearl Diver
  • **
  • Offline Offline
  • Posts: 14
  • Respect: +8
    • View Profile
Re: Interpreting CouncilRoom stats: Huge 'Effect With(out)'
« Reply #2 on: January 10, 2012, 11:27:21 am »
0
Okay, that makes sense; I guess with so few games in those categories, it is only logical for the stats to be way off.

But is playing over 800 games really "not many games"? I guess not for this forum, but wow.
Logged

theory

  • Administrator
  • *****
  • Offline Offline
  • Posts: 3603
  • Respect: +6125
    • View Profile
    • Dominion Strategy
Re: Interpreting CouncilRoom stats: Huge 'Effect With(out)'
« Reply #3 on: January 10, 2012, 11:38:51 am »
0
Not that many in that particular situation.  So given that you hate Woodcutter, it'd make sense that your numbers for when you do buy Woodcutter are extreme.
Logged

WanderingWinder

  • Adventurer
  • ******
  • Offline Offline
  • Posts: 5275
  • ...doesn't really matter to me
  • Respect: +4386
    • View Profile
    • WanderingWinder YouTube Page
Re: Interpreting CouncilRoom stats: Huge 'Effect With(out)'
« Reply #4 on: January 10, 2012, 11:45:15 am »
0
Quote from: theschizm on January 10, 2012, 11:27:21 am
Okay, that makes sense; I guess with so few games in those categories, it is only logical for the stats to be way off.

But is playing over 800 games really "not many games"? I guess not for this forum, but wow.
Well, it's less than 80 games for each and every kingdom card. But what I really meant was in those categories, i.e. games which you bought woodcutter, didn't by KC, didn't by FV, didn't buy tunnel. It's 7, 5, 4, and 2 games respectively.
Logged

rrenaud

  • Administrator
  • *****
  • Offline Offline
  • Posts: 991
  • Uncivilized Barbarian of Statistics
  • Respect: +1197
    • View Profile
    • CouncilRoom
Re: Interpreting CouncilRoom stats: Huge 'Effect With(out)'
« Reply #5 on: January 10, 2012, 11:45:19 am »
0
It's basically the statistical equivalent dividing by 0, since you don't have any games that you lost when you bought woodcutter.  I should possibly fix the estimation so it's not so extreme.

But popular buys might die, and be replaced by the thing I am currently working on.
Logged

theschizm

  • Pearl Diver
  • **
  • Offline Offline
  • Posts: 14
  • Respect: +8
    • View Profile
Re: Interpreting CouncilRoom stats: Huge 'Effect With(out)'
« Reply #6 on: January 10, 2012, 11:52:18 am »
0
Thanks for the answers guys, it makes sense now that you say it that way.
Logged

chwhite

  • Saboteur
  • *****
  • Offline Offline
  • Posts: 1065
  • Respect: +442
    • View Profile
Re: Interpreting CouncilRoom stats: Huge 'Effect With(out)'
« Reply #7 on: January 10, 2012, 11:59:57 am »
0
I have a -76.81 Effect With for Noble Brigand, having never won a single game with it in my deck.  I wonder if anyone has a more extreme number for any card.
Logged
To discard or not to discard?  That is the question.

WanderingWinder

  • Adventurer
  • ******
  • Offline Offline
  • Posts: 5275
  • ...doesn't really matter to me
  • Respect: +4386
    • View Profile
    • WanderingWinder YouTube Page
Re: Interpreting CouncilRoom stats: Huge 'Effect With(out)'
« Reply #8 on: January 10, 2012, 12:03:03 pm »
0
Quote from: rrenaud on January 10, 2012, 11:45:19 am
It's basically the statistical equivalent dividing by 0, since you don't have any games that you lost when you bought woodcutter.  I should possibly fix the estimation so it's not so extreme.

But popular buys might die, and be replaced by the thing I am currently working on.
Interesting....
Ballpark time frame?
Logged

DrHades

  • Conspirator
  • ****
  • Offline Offline
  • Posts: 206
  • Respect: +79
    • View Profile
Re: Interpreting CouncilRoom stats: Huge 'Effect With(out)'
« Reply #9 on: January 10, 2012, 12:05:28 pm »
0
Quote from: chwhite on January 10, 2012, 11:59:57 am
I have a -76.81 Effect With for Noble Brigand, having never won a single game with it in my deck.  I wonder if anyone has a more extreme number for any card.

I looked at your stats: -76.81 is a nice statistic error, but what is even more interesting - you have -4.11 effect with DUCHY!!! What the hell? ;D
Logged

WanderingWinder

  • Adventurer
  • ******
  • Offline Offline
  • Posts: 5275
  • ...doesn't really matter to me
  • Respect: +4386
    • View Profile
    • WanderingWinder YouTube Page
Re: Interpreting CouncilRoom stats: Huge 'Effect With(out)'
« Reply #10 on: January 10, 2012, 12:07:33 pm »
0
Quote from: DrHades on January 10, 2012, 12:05:28 pm
Quote from: chwhite on January 10, 2012, 11:59:57 am
I have a -76.81 Effect With for Noble Brigand, having never won a single game with it in my deck.  I wonder if anyone has a more extreme number for any card.

I looked at your stats: -76.81 is a nice statistic error, but what is even more interesting - you have -4.11 effect with DUCHY!!! What the hell? ;D
He's an engine-builder. Also, I've got worse with colony.
Logged

chwhite

  • Saboteur
  • *****
  • Offline Offline
  • Posts: 1065
  • Respect: +442
    • View Profile
Re: Interpreting CouncilRoom stats: Huge 'Effect With(out)'
« Reply #11 on: January 10, 2012, 12:10:38 pm »
0
Quote from: DrHades on January 10, 2012, 12:05:28 pm
Quote from: chwhite on January 10, 2012, 11:59:57 am
I have a -76.81 Effect With for Noble Brigand, having never won a single game with it in my deck.  I wonder if anyone has a more extreme number for any card.

I looked at your stats: -76.81 is a nice statistic error, but what is even more interesting - you have -4.11 effect with DUCHY!!! What the hell? ;D

And 5.49 Effect Without, too!  If I'm buying Duchies, chances are something's gone wrong. 
Logged
To discard or not to discard?  That is the question.

rrenaud

  • Administrator
  • *****
  • Offline Offline
  • Posts: 991
  • Uncivilized Barbarian of Statistics
  • Respect: +1197
    • View Profile
    • CouncilRoom
Re: Interpreting CouncilRoom stats: Huge 'Effect With(out)'
« Reply #12 on: January 10, 2012, 12:14:08 pm »
+1
Logged

chwhite

  • Saboteur
  • *****
  • Offline Offline
  • Posts: 1065
  • Respect: +442
    • View Profile
Re: Interpreting CouncilRoom stats: Huge 'Effect With(out)'
« Reply #13 on: January 12, 2012, 03:11:15 pm »
0
Quote from: chwhite on January 10, 2012, 11:59:57 am
I have a -76.81 Effect With for Noble Brigand, having never won a single game with it in my deck.  I wonder if anyone has a more extreme number for any card.

Ah, sadly this is about to be no longer the case, as I finally managed to win a game with the Brigand; it gave my opponent a Pool-clogging Copper twice, which is really all I asked of it.  Also, the Duchesses ran out, you don't see that every day! http://dominion.isotropic.org/gamelog/201201/12/game-20120112-120250-d42514e7.html
Logged
To discard or not to discard?  That is the question.
Pages: [1]
 

Page created in 0.11 seconds with 21 queries.