Fun With Popular Buys: Card Power Levels

In our most recent installment, we looked at the popular buys page to examine the impact of tournament prizes to see which ones were popular, which ones were effective.

While looking at this, I had jumped onto the page, sorted by “Win Rate With”, and then started in on the analysis. I noted that tournament had a win rate with of 1.02 (about 51% in a 2p game), which put it in Torturer / Wharf territory… a good card, but not Witch / Mountebank level. If you only look at this, Tournament looks like a slightly above average card. Unfortunately, it’s watered down because EVERYONE is buying it. When you also include the win rate without, you get a much different picture of the card.

Card Win Rate with Win Rate without
Torturer 1.03 ± 0.01 0.96 ± 0.01
Wharf 1.03 ± 0.01 0.91 ± 0.01
Tournament 1.02 ± 0.02 0.88 ± 0.06

So, based on this, its much safer to ignore torturer than it is to ignore Wharf or Tournament.

Read on for way too much data, and even a chart!

So, I decided to take all of the cards, and calculate their Win Rate Delta = [Win Rate With] – [With Rate Without]

This gave me a pretty wide range of deltas, from +0.73 (Trusty Steed) to -0.38 (Thief).

I then grouped these deltas into a variety of buckets for easy summary. I ended up with the following thresholds (converted to 2p win % for ease of mental digestion)

Win Rate Delta Bucket
X > 10% Amazing
5% < X < 10% Great
2.5% < X < 5% Good
0.5% < X < 2.5% Decent
-0.5% < X < 0.5% Average
-2.5% < X <  -.5% Poor
-5% < X < -2.5% Bad
-10% < X < 5% Terrible
X < -10% What Were You Thinking?

And how did the distribution fall? Not at all like I was expecting.

Below is the full set of cards, with some pithy comments. Please remember a few things before you rage at the placement of your favorite cards:

This is based on all isotropic data since the logs were parseable. This includes your stereotypical “village idiot” plays, and people buying chapels on turn 10 because theory has sea hagged them 10 times already. Just because a card is low on this list doesn’t mean its a bad card, it just means that in the history of isotropic, more games have been won when the card was ignored than when it was purchased / gained. Also, the Cornucopia cards are likely to be skewed because not only are people more likely to play them with other cornucopia cards (instead of pure random) where they may have additional synergy, but also because we don’t yet know how to use these cards effectively, but people are buying them anyway. This new metric is not scientific in any way… remember, the title of this article is “Fun with Popular Buys”. I had fun playing with the numbers, and I hope you will too.

Card Rate With Rate Without Delta Category
Trusty Steed 1.52 0.79 0.73 Amazing
Followers 1.53 0.82 0.71 Amazing
Princess 1.42 0.87 0.55 Amazing
Diadem 1.42 0.93 0.49 Amazing
Colony 1.09 0.63 0.46 Amazing
Bag of Gold 1.35 0.90 0.45 Amazing
Mountebank 1.09 0.83 0.26 Amazing
Platinum 1.07 0.82 0.25 Amazing
Witch 1.09 0.86 0.23 Amazing
Goons 1.04 0.81 0.23 Amazing
Familiar 1.06 0.85 0.21 Amazing

No surprises here.  It’s bad to ignore the power attacks.  It’s damn near impossible to win without Colonies, and getting Prizes means you are buying Provinces.

Card Rate With Rate Without Delta Category
Grand Market 1.05 0.89 0.16 Great
Province 1.02 0.86 0.16 Great
Tournament 1.02 0.88 0.14 Great
Ambassador 1.04 0.92 0.12 Great
Wharf 1.03 0.91 0.12 Great
Hunting Party 1.03 0.92 0.11 Great
Menagerie 1.03 0.92 0.11 Great

Also some familiar faces here, but also some newcomers!  Note that the Cornucopia cards will no doubt move around over time as people learn how to counter them and use them effectively.  My experiences thus far have also put these 3 (Tournament, Hunting Party, and Menagerie) as cards that are very difficult to ignore.

Card Rate With Rate Without Delta Category
Masquerade 1.05 0.95 0.1 Good
Sea Hag 1.04 0.94 0.1 Good
Nobles 1.01 0.91 0.1 Good
Chapel 1.02 0.93 0.09 Good
King’s Court 1.02 0.93 0.09 Good
Peddler 1.03 0.95 0.08 Good
Ghost Ship 1.04 0.97 0.07 Good
Vineyard 1.04 0.97 0.07 Good
Torturer 1.03 0.96 0.07 Good
Minion 1.02 0.95 0.07 Good
Venture 1.04 0.98 0.06 Good
Possession 1.03 0.97 0.06 Good

I’m a little surprised to see Sea Hag down here.

Card Rate With Rate Without Delta Category
Young Witch 1.02 0.97 0.05 Decent
Apprentice 1.02 0.97 0.05 Decent
Warehouse 1.02 0.97 0.05 Decent
Laboratory 1.01 0.96 0.05 Decent
Scrying Pool 1.02 0.98 0.04 Decent
Tactician 1.02 0.98 0.04 Decent
Island 1.01 0.97 0.04 Decent
Vault 1.02 0.99 0.03 Decent
Alchemist 1.01 0.98 0.03 Decent
Hoard 1.01 0.98 0.03 Decent
Fairgrounds 1.01 0.99 0.02 Decent
Golem 1.01 0.99 0.02 Decent
Caravan 1.00 0.98 0.02 Decent

Now we’re starting to find some cards that I really like.  I’m a little surprised to see cards like Caravan down near neutral levels.

Card Rate With Rate Without Delta Category
Fishing Village 1.00 0.99 0.01 Average
Apothecary 1.00 1.00 0 Average
Bank 1.00 1.00 0 Average
Bishop 1.00 1.00 0 Average
Cutpurse 1.00 1.00 0 Average
Harem 1.00 1.00 0 Average
Loan 1.00 1.00 0 Average
Salvager 1.00 1.00 0 Average
Upgrade 1.00 1.00 0 Average
Estate 1.00 1.00 0 Average

Now we’re getting into some nice cards, but mostly non critical.  I would have expected to see Upgrade and Fishing Village a little higher on the list.

Card Rate With Rate Without Delta Category
Swindler 0.99 1.00 -0.01 Poor
Bazaar 1.00 1.02 -0.02 Poor
Gold 1.00 1.02 -0.02 Poor
Conspirator 0.99 1.01 -0.02 Poor
Gardens 0.99 1.01 -0.02 Poor
Militia 0.99 1.01 -0.02 Poor
Mint 0.99 1.01 -0.02 Poor
Rabble 0.99 1.01 -0.02 Poor
Trading Post 0.99 1.01 -0.02 Poor
Farming Village 1.00 1.03 -0.03 Poor
Haven 0.99 1.02 -0.03 Poor
Moneylender 0.99 1.02 -0.03 Poor
Monument 0.99 1.02 -0.03 Poor
Duchy 0.99 1.02 -0.03 Poor
Library 0.98 1.01 -0.03 Poor
Hamlet 0.99 1.03 -0.04 Poor
Jester 0.99 1.03 -0.04 Poor
Expand 0.99 1.03 -0.04 Poor
Market 0.99 1.03 -0.04 Poor
Envoy 0.98 1.02 -0.04 Poor

Now here are some real surprises… there are some decent cards down here.  Trading Post most likely because people buy it as a reaction to getting junk in their deck, and don’t get to play it enough.

Remember that this is not factoring in when the cards were bought, so despite the fact that Militia / Silver is a strong opening, Militia is not a good card to buy halfway through a game.

Card Rate With Rate Without Delta Category
Steward 0.98 1.03 -0.05 Bad
Treasury 0.98 1.03 -0.05 Bad
Embargo 0.97 1.02 -0.05 Bad
Lookout 0.97 1.02 -0.05 Bad
Mining Village 0.99 1.05 -0.06 Bad
Courtyard 0.97 1.03 -0.06 Bad
Smithy 0.97 1.03 -0.06 Bad
Trade Route 0.97 1.03 -0.06 Bad
Wishing Well 0.96 1.02 -0.06 Bad
City 0.99 1.06 -0.07 Bad
Remake 0.98 1.05 -0.07 Bad
Lighthouse 0.98 1.05 -0.07 Bad
Shanty Town 0.98 1.05 -0.07 Bad
Merchant Ship 0.97 1.04 -0.07 Bad
Adventurer 0.95 1.02 -0.07 Bad
Explorer 0.95 1.02 -0.07 Bad
Festival 0.99 1.07 -0.08 Bad
Great Hall 0.98 1.06 -0.08 Bad
Cellar 0.97 1.05 -0.08 Bad
Spy 0.96 1.04 -0.08 Bad
Harvest 0.95 1.03 -0.08 Bad
Royal Seal 0.95 1.03 -0.08 Bad
Duke 0.93 1.02 -0.09 Bad
Stash 0.93 1.02 -0.09 Bad
Copper 0.93 1.02 -0.09 Bad
Pawn 0.97 1.06 -0.09 Bad
Fortune Teller 0.96 1.05 -0.09 Bad

Starting to see some of the underpowered 5’s, villages and slow trashers that people might overbuy as reactions.

Card Rate With Rate Without Delta Category
Potion 0.97 1.07 -0.1 Terrible
Forge 0.96 1.06 -0.1 Terrible
Pearl Diver 0.96 1.06 -0.1 Terrible
Council Room 0.95 1.05 -0.1 Terrible
Feast 0.95 1.05 -0.1 Terrible
Curse 0.92 1.03 -0.11 Terrible
Navigator 0.91 1.02 -0.11 Terrible
Worker’s Village 0.98 1.09 -0.11 Terrible
Native Village 0.96 1.07 -0.11 Terrible
Philosopher’s Stone 0.93 1.05 -0.12 Terrible
Watchtower 0.93 1.05 -0.12 Terrible
Village 0.97 1.09 -0.12 Terrible
Horn of Plenty 0.93 1.06 -0.13 Terrible
Mine 0.93 1.06 -0.13 Terrible
Moat 0.93 1.06 -0.13 Terrible
University 0.96 1.09 -0.13 Terrible
Quarry 0.95 1.08 -0.13 Terrible
Remodel 0.94 1.07 -0.13 Terrible
Tribute 0.93 1.07 -0.14 Terrible
Scout 0.91 1.05 -0.14 Terrible
Chancellor 0.88 1.02 -0.14 Terrible
Baron 0.93 1.08 -0.15 Terrible
Herbalist 0.93 1.08 -0.15 Terrible
Ironworks 0.93 1.08 -0.15 Terrible
Treasure Map 0.93 1.08 -0.15 Terrible
Outpost 0.90 1.05 -0.15 Terrible
Contraband 0.89 1.04 -0.15 Terrible
Bridge 0.96 1.11 -0.15 Terrible
Horse Traders 0.95 1.10 -0.15 Terrible
Woodcutter 0.89 1.05 -0.16 Terrible
Silver 0.98 1.15 -0.17 Terrible
Throne Room 0.95 1.13 -0.18 Terrible

Lots of cards down here that frequently look good to newer players, but aren’t good in actuality unless you’ve specifically planned for it (Throne Room, University, Mine).  I’m surprised to see potion down here.

Card Rate With Rate Without Delta Category
Secret Chamber 0.88 1.08 -0.2 What were you thinking?
Black Market 0.92 1.13 -0.21 What were you thinking?
Bureaucrat 0.84 1.06 -0.22 What were you thinking?
Transmute 0.87 1.10 -0.23 What were you thinking?
Workshop 0.83 1.06 -0.23 What were you thinking?
Smugglers 0.89 1.13 -0.24 What were you thinking?
Talisman 0.87 1.12 -0.25 What were you thinking?
Coppersmith 0.80 1.06 -0.26 What were you thinking?
Counting House 0.76 1.04 -0.28 What were you thinking?
Saboteur 0.83 1.11 -0.28 What were you thinking?
Pirate Ship 0.85 1.18 -0.33 What were you thinking?
Thief 0.71 1.09 -0.38 What were you thinking?

Most of these shouldn’t be a surprise.

Stay tuned for an upcoming feature where we look at the same type of analysis for a top ranked player, and how they differ from the standard player.

This entry was posted in Articles. Bookmark the permalink.

62 Responses to Fun With Popular Buys: Card Power Levels

  1. chwhite says:

    One observation: there a lot more bad cards than there are good cards- the *median* (if I’ve done my math correctly) is between Market and Envoy, at the low end of “Poor”. Cards like Trading Post and Militia are, in fact, better than over half the cards in Dominion.

    Definitely some surprises (and disappointments) on this list; in particular, it pains me to see Festival and Watchtower so low, and it’s amusing to see Gold classified as a “Poor” buy. I’m not surprised that Menagerie and Hunting Party are so high; Menagerie is absolutely explosive if you can build your deck right, and HP is proving itself to be better than Lab the vast majority of the time.

  2. Fuu says:

    “I’m a little surprised to see cards like Caravan down near neutral levels.”

    I suppose that (and I know I am guilty of this) sometimes people buy it, maybe over and over, assuming it is just an easy way to improve any deck …

    • Mean Mr Mustard says:

      I think a lot of these cards are in the neutral are because the likelihood that both players are buying them and there is usually a loser. Almost everyone buys at least one caravan and gold. The slightly negative overall number may represent the rare times when they are not optimal buys and the winner realizes it and somehow works there deck around them.

      A gold buyer may lose a fast gardens deck or a game where his opponent can somehow leapfrog straight to Platinum or perhaps against a hoard player.

      • Epoch says:

        Yeah, MMM nails it. Caravan is a “decent” card because if it’s in the game, almost invariably both players buy it, and almost invariably, one of them loses.

        • AndrewH says:

          Given the data, it should be possible to re-calculate these percentages, eliminating cards from games where both players bought them, or providing a weighting based on the quantity bought.

  3. LastFootnote says:

    To me this list mostly represents cards that take skill to use effectively and cards that don’t. Also, there are some cards near the bottom of the list that are more powerful in games with more than two players, which are less common on isotropic.

    It’s definitely interesting data, but I have to remind myself that correlation does not imply causality.

    • tlloyd says:

      I think there are actually three effects going on here, apart from the effect we’re trying to isolate (the card’s causal effect on winning).

      First, as you say, some cards take greater skill to use effectively, or are strong only in particular circumstances. For example, take City. If I told you I played a game where I bought several Cities but my opponent bought none, the odds are that I fell for the “City trap” and lost big time. The Tactician is ranked “Decent”, but I would suggest that in the hands of an expert the Tactician is at least “Great” if not “Amazing”. One more very simple example is the Chapel. Among readers of this blog Chapel’s power is no surprise. But most new players – and we were all new players at some point – just don’t see how to use Chapel effectively. Hopefully the upcoming Top Player stats vs. Standard Player stats will eliminate a lot of this effect.

      Second, some of the higher-ranked cards are simply harder to obtain. If you gain prizes, that means not only that you’ve already bought a Province, but that your deck is probably gaining Provinces and other good cards at a faster pace than your opponents–that’s a pretty good sign that you’re on your way to victory. As another example take Markets and Grand Markets. The only difference is +1 Coin, yet Grand Market is “Great” (almost “Amazing”) while Market is “Poor” (almost “Bad”). What gives? Well, it takes a much better deck to gain a Grand Market, since you can’t use any Copper to buy one. So gaining a Grand Market is a great indication of a powerful deck, while gaining a Market doesn’t say much.

      Third, some cards tend to get bought out or at least work best en masse (Fishing Village, Minion, etc.), while other cards you rarely need more than one or two of (Chapel, Tactician). If the whole pile of cards is likely to get bought out, then every card you gain is effectively a card denied to your opponent. Note that this applies with particular force to the Prize cards – you either get the only Followers or you don’t. I think this would tend to overstate the productive value of these cards.

      Of course, you could argue that the ease with which you can obtain and use cards is relevant to how good they are (or at least relevant to whether they will help you win). You could also argue that cards which are likely to split between you and your opponent are extra important to get if you want to win. So it’s not that there’s anything wrong with these rankings, it’s just that they speak to how likely an average player is to win with one of these cards in his deck, which includes more information than the “strength” of the card in the hands of an expert.

      • chesskidnate says:

        I will note that two coin is much better than one and I’d wager that most of the time a grand market is about as good(but obviously slightly worse in some situations) as two regular markets

      • timchen says:

        In my rough observation of the data, it is best understood as follows:

        At the extreme, there are cards which are almost good and almost terrible on every board. The data does show that.

        In the middle, however, are cards that are situational, and more importantly, there are cards which are good, but would be unwise just to spam it without support. For these cards the stats will be skewed toward the terrible side, as (i) for a situational card, there bound to be players not knowing when to use it (ii) for the spammable card, there bound to be players spamming it unwisely. Examples like conspirator and NV would be the first kind; upgrade, village, bazaar, festival would be the second.

    • captainfrisk says:

      Your point about 2p vs. more is very valid.

      Couple of notes on the differences:

      * Reactions are more valueable in multiplayer games with attacks
      * Thief and Pirate Ship have a greater chance of a positive outcome.
      * Non cursing attack cards such as militia have a difficult to quantify reduction in power because the odds of some opponents not being impacted increase (you don’t hurt all opponents evenly) – the average impact is the same, but the odds of a devastating hit on all opponents goes down as you have more opponents. (Example: Swindler Hits your opponents first 5 buy and turns it into a duchy should happen on the order of 10% of the time. Hitting all 3 opponents 5 buys? 0.1%)
      * Cursing attacks are weaker in 4p games because of the way curse scaling works. Ignoring cursing no matter what will get you 10 curses, but if 3 other people are also cursing (evenly), the curses should split 6/7/7/10, so you’ve only given up 3 curses by ignoring Sea Hag.
      * Tournament gets substantially weaker as more players have opportunities to show a province and ruin your chain. DO NOT UNDERESTIMATE THIS EFFECT.

  4. DRG says:

    “I’m a little surprised to see Sea Hag down here.”

    Sea hag has some hard counters, and situations where it is an inferior curse giver. If you buy it mindlessly (and many do because they think it’s really powerful and don’t always consider the rest of the cards) it racks up extra losses, while the person buying masquerades, witches, lookouts, etc not only beat the person with the sea hag, they don’t buy one in the first place.

    A lot of other cards are bought by players that don’t use them properly (especially saboteur) and get lower rates than they deserve. Have to admit, black market in the very bottom category surprises me though.

    • captainfrisk says:

      Re: Black Market.

      I was also shocked, but then I thought about the first time I saw it, and how excited I was at the possibility of getting top secret cards that no-one else had access to.

      I frequently buy black market, but I’m very careful to make sure that I don’t over terminal, and I check the deck first to make sure there’s something worth reaching for.

      I think this is a card that new players gravitate towards and overuse.

      Black market is bad when:

      1. There are other power cards already on the table (you get no huge advantage of pulling a mountebank out of the black market if your opponent is going to sea hag you all day.
      2. If there is a power card in the black market, but you don’t get it early enough (pulling a mountebank on turn 15 is nice… but will you even get to play it again?)

      I think I have a tendency to remember being hosed by black market giving my opponent a turn 5 witch, and forget all the times that the opponent uses it and flips: Expand, Moat, Treasure Map, with 5 coin in hand.

      Black market is a specialized woodcutter!

      • Zaphod says:

        I think Black Market suffers because:

        1. There is a delay in playing any cards gained with Black Market; you will shuffle at least twice before playing any of those cards. There is a similar lag time with most Alchemy cards, and note that Potion falls in the Terrible range on this chart.

        2. Luck plays a tremendous factor with Black Market. If your only options are weak cards and/or cards you can’t afford, it becomes only +2 Coin.

        3. It is almost useless in the endgame, since most kingdom cards are action or treasure cards, and you won’t get to play any cards you gain. If you play Black Market near the end, it’s probably just for the coin.

        4. It’s impossible to work out an opening strategy for the cards you’ll gain with Black Market because you don’t know what they will be. Without a clear strategy, it’s harder to win.

        I typically buy a Black Market if I can’t find a good strategy for the other cards in the set, and I will definitely buy it the first time it shows up in a set with Fairgrounds. Nonetheless, it has its drawbacks.

        • Zaphod says:

          For further evidence of point #1, look where Ironworks, University, Workshop, Horn of Plenty, Smugglers, Transmute, Forge, Mine and Remodel are listed; all are Terrible, or worse. The only “gain a card” actions that show up near the top of this chart are Tournament, which puts the gained card into your hand, and cards you gain by playing the Tournament. When you gain a card is a crucial factor in a game.

          • ksf says:

            I think this is a selection bias manifest from weaker players obsession with getting “cards,” and as many of them as possible. I still remember thinking Workshop was the greatest card ever, because you got a free card. Maybe even another Workshop, so you can get even more cards!

  5. DStu says:

    First thanks to the work, but I’m afraid, concerning the new Cards and esp. Tournament prizes we will have to wait some more weeks to get some good numbers.
    “Unfortunately, it’s watered down because EVERYONE is buying it. When you also include the win rate without, you get a much different picture of the card.
    Just as the WinRateWith is probably biases to the negative (because everybody buys it, even if it doesn’t fit the strategy) the WinRateWithout should be biases to the positve (because no one is not buying it, except some good players who have a good reason).

    Would be intersting to have a history of the WinRateWith/Without/Delta to see how this evolves over the next weeks…

  6. ipofanes says:

    Remember that these charts only list the cards bought, leaving out the Golds gained by Hoard/Explorer/Mint/Mine (and the Silvers gained with Bureaucrat … wait …).

    • captainfrisk says:

      Going to need rrennaud on this one, but I do think that with / without does reflect cards gained (otherwise the prizes wouldn’t show at all)…. so all those silvers gained by your well played Bureaucrat are factored into the rankings of both silver and b-crat.

    • theory says:

      No. The data includes all cards gained on your turn (so including Golds from Hoard, cards from Remodel, and Curses from Embargo, but not cards from an opponent’s Ambassador or Mountebank).

      • vidicate says:

        Are you certain? I’m not super active on isotropic, just some casual games with friends (so it’s easier to manually parse thru my data (Ooh, baby, I love it when you parse my datums like that… ahem, sorry)). As of this writing there are 12 games in my record. You can see for Curse that is shows relevant values like “Gains”, “%+”, and “Win Rate with”. Now head over to my game logs and you’ll see that I only had two games where Curses were ever in my deck–one from a Swindler, and three times from a Sea Hag. I never gained a curse on any of my own turns.

        • theory says:

          Hmm. I just talked to rrenaud and you are right. According to him, “Popular Buys” is not restricted to gains on your turn. The graphs (winrate vs. difference and winrate vs. turn gained) are, but Popular Buys includes cards that you gain at any time.

          So it’s even more expansive than I thought. But I think this only really affects the numbers for Curse, Copper, and Estate, as cards that you are forced to gain on other players’ turns.

          The point I made above, though, still stands. Popular Buys, contrary to its name, includes all cards gained, not just bought.

  7. susie says:

    Chancellor scored higher than bridge. HA! Maybe bridge isn’t always used effectively (and I don’t think chancellor is snatched up often) but that one made me laugh.

  8. ipofanes says:

    What will be the result if you calculate the odds ratio instead of the rate difference?

  9. Matthew Ryan says:

    What jumps out at me on this list is Masquerade. It’s seen as a mediocre card, yet it’s up there rubbing shoulders with big name stars like Wharf and Nobles. Is there strategy related to this card that’s pushing up its win rate deltas?

    Maybe what’s been underestimated is the power of “trash a card (from a larger selection) while still leaving enough cards to do something significant with the turn”?

    • rrenaud says:

      Also, it is a solid (second best in the game, next to ambassador?) counter to many of the other cards on top (Witch/Mountebank/Sea Hag/Familiar).

    • Last Footnote says:

      Also, it helps against some common strategies, like Chapel. Many players who are learning the strategies of the game will always jump on Chapel when it’s there. Once all their Coppers and Estates are gone, you can harvest very nice cards from them using Masquerade.

    • DG says:

      Who sees it as a mediocre card?

      • Zaphod says:

        I did, before I actually played it. It looked silly until I saw how much it could change a game, especially if curse-giving cards are available.

    • chwhite says:

      Masquerade would be a decent card even if there was no passing to other players. +2 Cards and the option to trash is great in the early game as a way to trim and cycle your deck at the same time.

      • captainfrisk says:

        Ding! The ability to possibly mess with your opponents hand is just icing on the cake. Maybe every once in a while you’ll sneak an estate or two into their deck in exchange for a copper, but most of the time this is: Draw 2 Cards, Maybe trash 1, leaving you with a hand of 5 cards, hopefully good ones. This usually leaves you in a place where you can get a 5 cost card (a 6 if you’re really lucky), which is more than you can say if you were playing Hag or Ambassador.

      • tlloyd says:

        Compare Lookout, which for the same price provides comparable cycling but doesn’t increase the size of your hand, requires rather than permits trashing a card, and has no effect on your opponent. The only advantages for Lookout are it being non-terminal and that it can set up your next draw better.

        • ipofanes says:

          +1 action is nothing to sneeze at, although it depends on the cicumstances. Scout without +action would rest in the shelves for $2. Lookout sans action would rather be placed at $1.

    • chesskidnate says:

      it also had that KC+Goons+masq combo that may have helped its winrate

    • BelZoness says:

      Even without the “killer combo” masquerade has great interactions with other cards

      In a village-laden game, masquerade along with any hand size reducer (militia, goons, ghost ship–but not torturer–) is a nasty interaction that will often cause your opponent to give you one of his best cards–that you can often immediately use! Conversely, in games with heavy trashing, draw combos, and masquerade I often make sure to keep a “crap” card to protect the rest of my hand.

  10. Dennis says:

    This statistic is definitely flawed as a way of judging goodness/badness of cards. Gold, for example, should be a pretty good card — most of the time, trying to win without gold is very tough. However, since gold is such an obvious buy, the only reason not to gain it is a strategy that works better without it or doesn’t need it, so overall gold has a slightly better win rate without than win rate with.

    This would take a little coding, but the numbers I’d like to see are these statistics calculated only from two player games where not every player gained the card in question and the histogram of (both ignored, bought and won, bought and lost, both bought). That would help us separate the cards that are so awesome everybody always buys them absent an extremely good reason and the situational or hidden gem type cards.

    • captainfrisk says:

      Agreed on all points. Rrennaud, are you listening?

      • rrenaud says:

        Hey jerk, I told you to consider %gained in your data, and you ignored me :P. It encodes basically the same thing.

        Really, just a graph of win rate delta on the y with % gained on the x axis would be interesting.

  11. Lost Alpaca says:

    This list is being used as a buying guide by some newer players. The following exchange occured after I opened Militia/Silver on this board: [Hoard, Horn of Plenty, •Cutpurse, •Militia, •Spy, Throne Room, Tournament, Masquerade, Embargo, Pearl Diver]

    14:54 Kyula: hi
    14:54 Kyula: gl
    14:54 Kyula: hf
    14:54 Lost Alpaca: hi u2
    14:54 Kyula: militia has only -0.15
    14:54 Kyula: wouldn’*t buy it
    14:55 Lost Alpaca: ??
    14:55 Kyula: win rate delta
    14:56 Lost Alpaca: ok, don’t buy it then
    14:59 Lost Alpaca: it certainly sucks to be on the receiving end of one
    14:59 Kyula: no
    14:59 Kyula: it don’t hurt that much
    15:02 Kyula: grats
    15:02 Kyula: you win
    15:03 Lost Alpaca: thanks
    15:03 Kyula: gg
    15:03 Kyula has returned to the lobby.
    15:03 Lost Alpaca: gg

  12. Alex M says:

    Sea hag would likely be there because you are having sea hags used on you as well. Therefore, more curses, slower game, less likely to blitz for provinces, and weakened followers.

  13. yukonhorror says:

    I kind of skipped to the end. I am curious how the “lose” numbers compare. How many of those cards “lose” by buying them and by opting out. That would be a nifty analysis.

    have a lose rate delta
    a delta defined as win/with-lose/without (must buy delta)
    a delta defined as win/without-lost/with (never buy delta)

    • mischiefmaker says:

      Isn’t lose-without just (2.00 – win-without) and lose-with just (2.00 – win-with)?

      If that’s correct, lose-rate delta is just -(win-rate delta), and the other two metrics are always 0.

  14. Jeremy says:

    As I’m sure it’s been noted, something interesting about a lot of the cards labeled “decent” is they are cards that work better with more of said card. I’m sure there’s a connection there that would indicate they aren’t as bad as their rating suggests.

    These numbers, in no way, indicate losing percentages, do they? Like the Caravan, for example. Yes, both players in all likelihood buy this card, and even though one of the two players has to lose, I don’t think that’s what is getting represented here. I’d wager that a “lose with/lose without” chart would look completely different than this (and no, I don’t mean in reverse order). And I’d bet Caravan is ranked way better.

    A couple other inquiries/observations:
    -I’m totally shocked that Throne Room is labeled “Terrible”.
    -I’m surprised Festival is as low as it is.
    -Does anyone think Tribute’s value will go up in Cornucopia games (with the amount of different cards there will be)?
    -I’ve always thought City was a somewhat overrated card.

  15. Seamus says:

    Actually this chart is spot on for showing two things. How situational a card is e.g. Witch ranks high because you have to acknowledge it’s presents. Also how much skill it takes to use a card. Some cards that rank high are very self synergistic, where as low ranked cards need support from very select cards. Knowing complex interactions is the mark of a skilled player, and the “oh shiny” factor of some cards leads most players to play it badly. It would be nice to see a side by side comparison of the sample size for each card.

  16. Can we get this list updated for the more recent cards?

    • captainfrisk says:

      Request received. I can’t promise it will get done before the end of the year, but I’ll see what I can do. While we’re making requests: theory can you post an article on how to not get wrecked by allfail in tournaments?

  17. no potions says:

    This is an example of how pure data can be hugely missleading. The preface at the beginning is good though, but those who don’t pay it enough head may be mislead. I pose the following.

    1) play as a robot using this list exclusively to determine bought cards. It may influence cards trashed or play order, but that should probably still be human intervention. I bet it does well at lv1-2, and horribly at lv5-50.

    2) re-calculate your data using only games where each player was ranked a certain skill level. The higher you go, the more accurate it theoretically should be, but also the more susceptible to meta-gaming and group think as you reduce number of players. But then you are also looking at cards that are strong when you know how to use them. So a strong card for a pro may suck for a novice. But it would still suggest cards that are good for novices to figure out. So ~lv20-50 would probably be a good range.

    There are many examples to state how the data isn’t spot on. But the replies are so numerous, that I won’t add to that level of fine detail.

    • rrenaud says:

      I did some latter work on this kind of thing here:

      You can sort by quality or introduce the dependence on some cards being available, and see how quality changes. I think it mostly does a decent job of ranking cards and finding synergies, but of course, I am pretty biased ;P.

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s