Print

Please login or register.
1 Hour 1 Day 1 Week 1 Month Forever
Login with username, password and session length

[Axxle]

My worst given available by far is Jester.  Which is one reason why it bothers me that O claims it so often! (in mafia)

I've been meaning to ask this forever.. how does one read the councilroom stats (for invidual cards)?

http://councilroom.com/popular_buys?player=MrEevee

[Lekkit]

Apparently the highest effect with for me is Thief. That's kinda cool, I guess. Noble Brigand is up there too.

[Eevee]

My worst given available by far is Jester.  Which is one reason why it bothers me that O claims it so often! (in mafia)

I've been meaning to ask this forever.. how does one read the councilroom stats (for invidual cards)?

http://councilroom.com/popular_buys?player=MrEevee

I know that page.. what I dont know is how to read it. 

[qmech]

It says my worst card is Mine. Which to me makes sense, I don't know exactly how to use it  .

High level summary: don't, unless it's a Colony game.  In that case, you don't always want Mine, but if you do the goal is Mine Gold->Platinum, and that's what you should use to evaluate its effectiveness.  Any of the other transmutations should be considered consolation prizes.  (Corollary: you usually want to upgrade the most expensive Treasure you have in hand.)

I've been meaning to ask this forever.. how does one read the councilroom stats (for invidual cards)?
 

The basic unit of account is "win points": you get one for every player in every game that you win.  Win rate is the proportion of maximum available points that you gain: if you only play 2 player and win half your games your win rate is 1.  Anything higher means you do better than random, anything lower means you do worse.

The "Win Rate given avail" column shows what proportion of games you win given that the corresponding card is in the Kingdom, irrespective of whether you or your opponent actually buy it.  The "Win Rate with" and "Win Rate without" columns split the games between those when you buy (or otherwise gain) the card, and the games when you don't, allowing you to separate the cases where you do well with a card because you know when to buy it and how to use it when you do, and the cases when you know when things can be safely ignored.

The "Effect with/without" columns are based on the same win rate data, but offer a picture of how you do with or without that card compared to some baseline.  I can never remember if the baseline is your performance overall, or how other people perform with the same card, and it would be great if someone could explain here.

The other columns are less interesting, and more self-explanatory.  Win rate given available is probably the most useful statistic (there may be a case for effect with/without, but it would be reckless to claim that when I'm not sure how they're defined).

[ftl]

The "Effect with/without" columns are based on the same win rate data, but offer a picture of how you do with or without that card compared to some baseline.  I can never remember if the baseline is your performance overall, or how other people perform with the same card, and it would be great if someone could explain here.

The baseline is how you perform overall.

A positive "Effect with" means you win more games than your average when you get the card (presumably, that means you're good with it and/or correctly recognize when to get it). A positive "Effect without" means you win more games than your average when you skip the card (means you correctly recognize when the card is skippable).

A negative "Effect with" means you win fewer games than your average when you get the card; you're either buying it too often or not using it right when you do. A negative "effect without" means that you're skipping the card when you should get it.

(Of course, if you haven't played  that many games with any particular card, any of those "Effects" could also just be due to chance, since it's comparing against your average; if you have only a few games where you gained a card, your "effect with" is going to be determined entirely by those few games, and so won't be a reliable indicator of anything. And if the values are close to zero, from either side, positive or negative, that means it's not a big effect. )

[Eevee]

Thanks qmech (and ftl), I get it now.

.. i seem to have 11 cards with negative winrate given available, farming village being the worst of them. i earlier thought i underestimated it, it seems i was wrong.

[Archetype]

It also says my best is Caravan, which is kind of a surprise to me.

That means you are not buying it too frequently. In other words take this as  an indication that you should NOT increase your buying of this card over your typical pattern.

Oh. Woops, must have been reading it wrong  . Thanks pingpongsam.

@qmech: When I play irl that's when I buy it too, but I'm still not so good with it

[verikt]

I'm still not sure I've figured this out. I have high effect with for duchy estate and copper. What does that tell me?

[hdu88]

Mint is the only card where my win rate given avail is <1. What surprises me is that my win rate with Adventurer is 1.48 even though I get it twice as often as the average player.

[AJD]

I'm still not sure I've figured this out. I have high effect with for duchy estate and copper. What does that tell me?

It probably means that you're more likely to buy them in games where you're already winning, or where you have a strategy (such as Duke or Gardens) that hinges on them.

[ftl]

I'm still not sure I've figured this out. I have high effect with for duchy estate and copper. What does that tell me?

Probably not too much. Strictly speaking, it means that when you buy duchy, estate, and copper, you do better than when you don't.

However, given what those cards are, I'm not sure what to make of it. All three of those COULD be cards that you buy in the late-game on a last turn when you're ahead anyway.  Or maybe you only switch to Duchy and Estate after you already have a few provinces, so if you're behind and losing you don't end up getting any, thus making a correlation appear there. 

Or it could be meaningful. Are you particularly good at games where you might want to get Duchies, Estates, and Coppers? Alt-vp games like gardens or SR, maybe IGG rushes, etc.

It's really hard to tell with the basic cards...

[WanderingWinder]

Once more, people are misunderstanding what effect with and effect without measure.
They don't just measure how you do with(out) a card compared to how you do on average. If that were the case, you would just look at your winrate with and winrate without.
They actually measure how well you do with(out) a card (controlled for how you do on average) in comparison to how well the average player does with(out) the same card.
This really needs to get in an FAQ somewhere.

Anyway, example to help illustrate.
My overall winrate is 1.19. If you look at how I do with smugglers, you see that my winrate with smugglers is only 1.16 (so sorta significantly below my average winrate), whereas my winrate without smugglers is 1.26 (significantly above my average). But my effect with is a positive 1.02, and my winrate without is a negative 1.24 (that's -1.24). What gives? Well, the average player, if you look at the overall numbers, has a winrate with of 0.90 and without of 1.12 (obviously, the average player has an overall winrate of 1). So while I don't do great with the smugglers, and I do pretty well without them, this is not as much the case for me as it is for the average player. The exact numbers also have to do with the standard deviations of these values (which is why you get massive massive numbers for the effect things when you have like 0 wins or 0 losses with or without some particular card. Which I don't have on anything anymore, after my half-billion games).

Hope that helps.

And please, FAQ this question. With a good explanation (I would imagine someone can do it more clearly and/or succinctly than I just did).

[verikt]

WW They actually measure how well you do with(out) a card (controlled for how you do on average) in comparison to how well the average player does with(out) the same card.

So I'm doing better when I buy duchies and coppers than the average player? And duchies by 5?

[Captain_Frisk]

Most of my "bad" cards are from hinterlands.  Like WW, i've been mostly playing high level opposition lately, so my average win rate has gone down.  (I typically play +/- 20 automatch).

My worst card(s) that wasn't Hinterlands, were Alchemist and Lab.

[WanderingWinder]

WW They actually measure how well you do with(out) a card (controlled for how you do on average) in comparison to how well the average player does with(out) the same card.

So I'm doing better when I buy duchies and coppers than the average player? And duchies by 5?

Well, 'duchies by 5' seems like a weird way to phrase it. But yes. You're ~5 standard deviations relatively more skilled when buying duchies than the average isotropic-er.

[gman314]

My worst card is Forge which really comes as no surprise to me. How do you play Forge?

What did surprise me is that my best card given availability is a tie between Governor and Fortune Teller. (Governor with less uncertainty.) I really like Governor and generally do well with it, but Fortune Teller is definitely a surprise as I rarely buy it.

[verikt]

Then if I'm reading my stats right my worst card is trusty steed.

[ycz6]

Then if I'm reading my stats right my worst card is trusty steed.

Uh, you are not reading your stats right. That is an impressively low Effect With though.

[verikt]

Uh, you are not reading your stats right. That is an impressively low Effect With though.

Yeah, isn't it? I guess at least half the time I take ts the other guy takes followers.
Then where am I supposed to look?

[WanderingWinder]

Uh, you are not reading your stats right. That is an impressively low Effect With though.

Yeah, isn't it? Then where am I supposed to look?

If you want to just know what you're personally best/worst with, the best number to look at is generally win rate given availability.

[verikt]

Thanks WW
then that would be fools gold at 0.77 Maybe I'm not enough of a fool?

[Obi Wan Bonogi]

Possession... 

[methods of rationality]

Cellar   at .81 effect given available and city is close at .83. Someone else mentioned philosopher stone. I think the number one rule to philosopher stone is buy it if and only if its a cursing game

[Davio]

Envoy with 1.07.
While still > 1 I disregard Envoy too often in engine games.

Swindler is second with 1.09, but that's probably due to the inherent luck factor.
Swindling an Estate is sooo much worse than Swindling a Copper into an Estate.

My highest is Stables with 1.48. What can I say? I'm a Stable Boy.
Behind that is Jack, gotta love good ol' Jack with 1.46.

But to my surprise Horn of Plenty is up there with 1.46 even if I'm always troubled by it. Apparently less so than I thought.

And #4 is Develop, ha! Mostly because I recognize it's suckiness, my gain percentage is only 18.9%, lower than Herbalist!


So the worst card is Envoy (actually Archivist, but it doesn't count and I only used it in solitaire) as I'm too dismissive of it too often.

[qmech]

Memory stirs: is this characterisation of Effect With correct?  When I buy King's Court my win rate is 1 standard deviation above my average.  When other people buy King's Court, their win rate is 2 standard deviations above their average (which I suppose must be 1).  Then my effect with King's Court would be -1?