Dominion Strategy Forum

Please login or register.

Login with username, password and session length
Pages: 1 ... 9 10 [11] 12  All

Author Topic: Updating the Top 5 lists  (Read 87831 times)

0 Members and 1 Guest are viewing this topic.

chwhite

  • Saboteur
  • *****
  • Offline Offline
  • Posts: 1065
  • Respect: +442
    • View Profile
Re: Updating the Top 5 lists
« Reply #250 on: December 17, 2011, 05:20:32 pm »
0

From simulations:

Lighthouse/Minions (+Loan) crushes BMU 8 to 2, but if you let the big money player spam Horse Traders Minions get crushed 3 to 7.

I'd conclude Horse Traders is a very strong counter against Minions.

Your explanatory note admits that the simulator doesn't play Minion very well, and the Minion deck would be further improved if you let it buy one or two Horse Traders as well.  As long as there's some other support for Minions (trashing being the most critical), the presence of Horse Traders is not a reason to avoid Minion; rather it's a reason to get both.

Also I bet you could boost the Minion win percentage by making it avoid Duchies for longer?
Logged
To discard or not to discard?  That is the question.

Anon79

  • Apprentice
  • *****
  • Offline Offline
  • Posts: 264
  • Respect: +39
    • View Profile
Re: Updating the Top 5 lists
« Reply #251 on: December 22, 2011, 03:15:47 am »
0

Man, people keep saying that Horse Traders is this incredible counter to Minion, its mere presence makes Minion less valuable, not worth going for on HT boards, and I just gotta say that's not true in the least.  Minions are still that strong, even with HT out.
While you may have a point, this particular game certainly isn't definitive:

Quote
--- chwhite's turn 6 ---
(...)
chwhite plays a Sea Hag.
... yaron draws and discards a Sea Hag.
... yaron gains a Curse on top of the deck.

(...)
--- chwhite's turn 8 ---
chwhite plays a Sea Hag.
... yaron draws and discards a Gold.
... yaron gains a Curse on top of the deck.

   --- yaron's turn 8 ---
   yaron plays a Sea Hag.
   ... (chwhite reshuffles.)
   ... chwhite draws and discards a Curse.
   ... chwhite gains a Curse on top of the deck.
Logged

Tejayes

  • Golem
  • ****
  • Offline Offline
  • Posts: 176
  • Respect: +132
    • View Profile
Re: Updating the Top 5 lists
« Reply #252 on: December 22, 2011, 05:22:43 pm »
+3

With the $3 lists on their way, hopefully, here's a compilation of Win Rate With/Without data from CouncilRoom.com, along with a more subjective list of overrated and underrated $3's based on percentage of games where the cards are gained.

WIN RATE WITH

1: Masquerade (1.05)
2: Ambassador (1.04)
3: Menagerie (1.03)
4: Warehouse (1.02)
5: Loan (1.01)
6 (tie): Fishing Village/Swindler (1.00)
8 (tie): Scheme/Steward (0.99)
10 (tie): Great Hall/Shanty Town/Trade Route/Tunnel/Wishing Well (0.98)
15 (tie): Lookout/Oasis/Oracle/Village (0.97)
19: Fortune Teller (0.95)
20 (tie): Black Market/Watchtower (0.94)
22: Develop (0.93)
23: Woodcutter (0.91)
24: Smugglers (0.89)
25: Chancellor (0.87)
26: Workshop (0.85)

WIN RATE WITHOUT

1: Ambassador (0.92)
2: Masquerade (0.93)
3: Menagerie (0.94)
4: Warehouse (0.97)
5: Loan (0.99)
6 (tie): Fishing Village/Swindler (1.00)
8 (tie): Oracle/Scheme/Wishing Well (1.01)
11 (tie): Chancellor/Fortune Teller/Lookout/Steward (1.02)
15: Trade Route (1.03)
16 (tie): Oasis/Tunnel/Watchtower (1.04)
19 (tie): Shanty Town/Woodcutter (1.05)
21 (tie): Develop/Great Hall/Workshop (1.06)
24: Village (1.07)
25: Smugglers (1.12)
26: Black Market (1.13)

DIFFERENCE (WRW - WRWO)

1 (tie): Ambassador/Masquerade (+0.12)
3: Menagerie (+0.09)
4: Warehouse (+0.05)
5: Loan (+0.02)
6 (tie): Fishing Village/Swindler (+0.00)
8: Scheme (-0.02)
9 (tie): Steward/Wishing Well (-0.03)
11: Oracle (-0.04)
12 (tie): Lookout/Trade Route (-0.05)
14: Tunnel (-0.06)
15 (tie): Fortune Teller/Oasis/Shanty Town (-0.07)
18: Great Hall (-0.08)
19 (tie): Village/Watchtower (-0.10)
21: Develop (-0.13)
22: Woodcutter (-0.14)
23: Chancellor (-0.15)
24: Black Market (-0.19)
25: Workshop (-0.21)
26: Smugglers (-0.23)

OVERRATED (High gain rates, low WRW-WRWO)

1: Black Market (67% gain rate, -0.19 WRW-WRWO)
2: Great Hall (79.7% gain rate, -0.08 WRW-WRWO)
3: Shanty Town (75.7% gain rate, -0.07 WRW-WRWO)

UNDERRATED (Low gain rates, relatively high WRW-WRWO)

1: Loan (41.3% gain rate, +0.02 WRW-WRWO)
2: Oracle (30.2% gain rate, -0.04 WRW-WRWO)
3: Wishing Well (34.1% gain rate, -0.03 WRW-WRWO)
Logged

theory

  • Administrator
  • *****
  • Offline Offline
  • Posts: 3603
  • Respect: +6121
    • View Profile
    • Dominion Strategy
Re: Updating the Top 5 lists
« Reply #253 on: December 22, 2011, 05:34:33 pm »
0

Excellent work.  This stuff really should be displayed on CR.com directly ...

Of course the data is a little misleading, since obviously FV is "better" than Loan.  I wonder if there is some kind of data that can capture the strength of a card that is frequently a must-buy and sometimes a never-buy.
Logged

Tejayes

  • Golem
  • ****
  • Offline Offline
  • Posts: 176
  • Respect: +132
    • View Profile
Re: Updating the Top 5 lists
« Reply #254 on: December 22, 2011, 05:57:27 pm »
0

Excellent work.  This stuff really should be displayed on CR.com directly ...

Of course the data is a little misleading, since obviously FV is "better" than Loan.  I wonder if there is some kind of data that can capture the strength of a card that is frequently a must-buy and sometimes a never-buy.

My theory (no pun intended, theory) on why FV is said to be "evenly balanced" is because it's bought so freaking often that it cancels itself out. That would also explain why Chancellor is not at rock bottom -- it's bought so infrequently that WRW vs. WRWO is a bit skewed.
Logged

theory

  • Administrator
  • *****
  • Offline Offline
  • Posts: 3603
  • Respect: +6121
    • View Profile
    • Dominion Strategy
Re: Updating the Top 5 lists
« Reply #255 on: December 22, 2011, 06:01:16 pm »
0

The easy solution to this -- look at only games where one person buys it and the other doesn't -- does not work, since there are many boards where FV is a trap.
Logged

Jimmmmm

  • Torturer
  • *****
  • Offline Offline
  • Posts: 1762
  • Shuffle iT Username: Jimmmmm
  • Respect: +2017
    • View Profile
Re: Updating the Top 5 lists
« Reply #256 on: December 22, 2011, 06:04:59 pm »
0

I guess Win Rate With etc has more to do with under- or over-rated a card is than how good it actually is.
Logged

Epoch

  • Tactician
  • *****
  • Offline Offline
  • Posts: 421
  • Respect: +38
    • View Profile
Re: Updating the Top 5 lists
« Reply #257 on: December 22, 2011, 06:16:31 pm »
+1

My theory (no pun intended, theory) on why FV is said to be "evenly balanced" is because it's bought so freaking often that it cancels itself out. That would also explain why Chancellor is not at rock bottom -- it's bought so infrequently that WRW vs. WRWO is a bit skewed.

Well, also, it has to do with there being plenty of cards that are much worse than Chancellor, which is, at the very least, not harmful if you aren't for whatever reason worried about a terminal clash (only action or a deck that has plenty of +actions).  It's not like Mandarin or Develop or Saboteur where -- aside from all the same terminal conflicts that Chancellor also is prey to -- can be actively detrimental to your deck if bought in many or most situations.
Logged

olneyce

  • 2011 DS Champion
  • *
  • Offline Offline
  • Posts: 245
  • Respect: +210
    • View Profile
Re: Updating the Top 5 lists
« Reply #258 on: December 22, 2011, 08:26:21 pm »
0

I think if you combine these numbers with the "%+" number, it produces a nice approximation of 'real' card strength.  Fishing Village, for example, is bought or gained 92% of the time, while the 3s that are at the top are all below 70%.  Put those two things together and you might have a decent card quality list.

Here's the simple multiplication of the numbers (setting the +/- on a scale of 100, and multiplying by the % gained or bought):

Fishing Village (92)
Ambassador (75.8 )
Menagerie (74.7)
Great Hall (73.3)
Steward (71.7)
Shanty Town (70.4)
Village (68.3)
Tunnel (67.8 )
Scheme (65.3)
Masquerade (64.7)
Swindler (63.8 )
Warehouse (62.3)
Oasis (54.9)
Trade Route (54.7)
Black Market (54.3)
Lookout (45.4)
Smugglers (42.4)
Loan (42.1)
Develop (40.5)
Watchtower (37.2
Wishing Well (33.1)
Woodcutter (30.4)
Oracle (29)
Fortune Teller (28.8 )
Workshop (22.7)
Chancellor (11.3)

There's a few strange results - Great Hall as a spammable VP is purchased fairly often but often more as an afterthought or act of desperation, and Masquerade is so underrated that its buy % overwhelms it's high win rate - but it's pretty close.  In effect, this uses the % gained or bought as a crowd-sourced baseline, and adds in a modest modification according to the actual results.  Basically: it corrects for 'overrating' and 'underrating.'

This method seems likely to help out cards with +action or +buy, since those are often necessities but are rarely the standout parts of a deck.  But that might just be a matter of perspective - if those cards are the fuel of many good engines, then maybe they are 'better' cards than we normally give them credit for. 
Logged

Elyv

  • Moneylender
  • ****
  • Offline Offline
  • Posts: 163
  • Respect: +15
    • View Profile
Re: Updating the Top 5 lists
« Reply #259 on: December 22, 2011, 08:49:07 pm »
0

Watchtower is ridiculously low on that list, imo. Warehouse too, probably. Masquerade is insane, of course, and way too low.
Logged

HiveMindEmulator

  • Mountebank
  • *****
  • Offline Offline
  • Posts: 2222
  • Respect: +2118
    • View Profile
Re: Updating the Top 5 lists
« Reply #260 on: December 22, 2011, 08:58:53 pm »
0

The problem with multiplying the numbers like that is that there is very little % variation in the win rate numbers, while there is much greater % variation in the gain numbers.
Logged

olneyce

  • 2011 DS Champion
  • *
  • Offline Offline
  • Posts: 245
  • Respect: +210
    • View Profile
Re: Updating the Top 5 lists
« Reply #261 on: December 22, 2011, 09:18:14 pm »
0

The problem with multiplying the numbers like that is that there is very little % variation in the win rate numbers, while there is much greater % variation in the gain numbers.
Yeah, like I said, this is really just a modified version of the crowd-sourced "percentage obtained" number.  Clearly the game is too complex for such simplistic numbers to really capture what makes a card 'good.' 

We'd have to go way deeper into when the cards are gained, how often they win when alone vs. when they have other supplementary options, how useful a card is when by itself vs. when you need more than one.  And frankly, it would be really helpful to divide the overall lists into quadrants of skill level.  If we want to know what the best card is, we really want to know what the best card is in games with good players.  I mean, Village is gained more by n00bs, but it's WORSE for them than it is for better players.  I would imagine some of the discrepancies would narrow for the best players, though it would be interesting to see whether and how.
Logged

DG

  • Governor
  • *****
  • Offline Offline
  • Posts: 4074
  • Respect: +2624
    • View Profile
Re: Updating the Top 5 lists
« Reply #262 on: December 22, 2011, 09:59:37 pm »
0

I don't think there is anything unusual about those win rate graphs except for fishing village and loan. The loan is a reasonable 3 cost card but I suspect it has a high success rate because novices just don't buy it. On the other hand the rest of the cost 3 cards probably suffer badly because novices buy them randomly when they're failing to make a decent deck. It's harder to unnecessarily buy a bank than unnecessarily buy a woodcutter.

The most interesting statistic for me is the "win rate without" fishing villages (1.00). Irrespective of how strong fishing villages seem to be and how frequently they are bought, when people ignore fishing villages in a kingdom they are getting entirely average results. It's not like the chapel which is also frequently bought and players lose more often by ignoring it. I also can't believe that all players in a game ignore fishing villages in unison to make the result redundant. Perhaps too many players overload their decks with terminal actions and can't beat simpler alternatives, even with the help of fishing villages.
Logged

timchen

  • Minion
  • *****
  • Offline Offline
  • Posts: 704
  • Shuffle iT Username: allfail
  • Respect: +234
    • View Profile
Re: Updating the Top 5 lists
« Reply #263 on: December 23, 2011, 01:58:57 am »
0

I think it is the opposite: people think FVs are so good and just load them up, without terminals to really take advantage of it. Without effective terminals, FV is just a Lighthouse.

Or to be more specific, in a game without good drawers, an excess of FV may actually end up not doing much good. I've won quite some games by not getting *as many* FVs as my opponent. Probably only in very BM-ish games can one win without a single FV, though.

Logged

theory

  • Administrator
  • *****
  • Offline Offline
  • Posts: 3603
  • Respect: +6121
    • View Profile
    • Dominion Strategy
Re: Updating the Top 5 lists
« Reply #264 on: January 06, 2012, 10:50:58 am »
0

Here's a question I'm grappling with as I write the Worst $4's list:

Are these rankings based on how bad the cards are when you consider adding them to any random deck you're building?  Or are they based on how helpful they are when you do buy them?  Or both?

Clearly, throwing Coppersmith into a random deck is pretty awful.  But when you do buy Coppersmith and build for it, it's glorious.  Spy, on the other hand, is basically always a mediocre addition to your deck.

(Maybe multiply how good it is by how often it is likely to be good?)

My sense is that, especially since these are sort of instructional, that I should go with the former (i.e., Coppersmith is worse than Spy), and explain why they are so rarely bought in the first place.  This is also, I believe, the approach the list has mostly taken thus far.

Maybe it's time to bite the bullet and work harder at making these official.  Qvist's topic is quite good.
Logged

Thisisnotasmile

  • Saboteur
  • *****
  • Offline Offline
  • Posts: 1493
  • Respect: +676
    • View Profile
Re: Updating the Top 5 lists
« Reply #265 on: January 06, 2012, 10:54:56 am »
0

You should rank the cards based on how good they are compared to Warehouse.

Edit: But seriously, I think how often you would buy a card is just as important as how useful it can potentially be in the ideal deck, so basically, yeah, multiply them together.

Alternatively, wait for Qvist to overtake you and then copy his list...
« Last Edit: January 06, 2012, 11:03:46 am by Thisisnotasmile »
Logged

rinkworks

  • Saboteur
  • *****
  • Offline Offline
  • Posts: 1316
  • Respect: +937
    • View Profile
    • RinkWorks
Re: Updating the Top 5 lists
« Reply #266 on: January 06, 2012, 11:04:52 am »
0

Here's a question I'm grappling with as I write the Worst $4's list:

Are these rankings based on how bad the cards are when you consider adding them to any random deck you're building?  Or are they based on how helpful they are when you do buy them?  Or both?

I think there's no way to definitively resolve this dichotomy.  No matter how you balance it out, someone is going to come along and lodge a perfectly sound objection from some other point along that continuum.

Personally, I prefer a balance, but (like Pearl Diver vs. Herbalist) Coppersmith vs. Spy features such extremes that it's tough to balance them out.  I think in that case I'd rank Coppersmith worse than Spy, but that may be because I recently tried a Coppersmith strategy with a great board for it (Tactician) and still got trounced.  Then again, I'd rank Herbalist worse than Pearl Diver, too -- so maybe that confirms that my own general feeling is to ever-so-slightly favor general purpose usefulness over niche power.  But even that is going to be a case-by-case judgment call.

Bottom line:  You're doing a great job; keep doing what you're doing.
Logged

mischiefmaker

  • Golem
  • ****
  • Offline Offline
  • Posts: 185
  • Respect: +108
    • View Profile
Re: Updating the Top 5 lists
« Reply #267 on: January 06, 2012, 12:58:43 pm »
0

I think you're doing a great job and I would recommend you not worry too terribly much about whether the lists are objective/scientific/precise -- I think the whole point of lists is to spark discussion. I mean, if there were a universally acknowledged metric and a monotonic ordering, the list wouldn't be very interesting, right? Like, nobody makes a list of "The 5 best baseball cards"*.

I also think that the answer to your question depends a little bit on your audience. If dominionstrategy.com is aimed at a small community of high-level players who are trying to refine their games, then I think it makes sense to more heavily favor "when this card is good, how good is it?"; i.e. Coppersmith > Spy. But, if it were my blog, I would be aiming at the audience of all people who play Dominion, in which case I think the metric should be something like:

(how often is this card useful) x (how good is it when it is useful) - (how often/badly does it actively harm your deck)

...in which case I would put Spy > Coppersmith since Coppersmith tends to be useful a whole lot less than Spy and can be a shiny trap card for newer players.

But like I said, you're doing a great job, and if people disagree (even vehemently), that's probably a good thing.

* Well, ok, that's not strictly true.
« Last Edit: January 06, 2012, 01:01:07 pm by mischiefmaker »
Logged

tlloyd

  • Tactician
  • *****
  • Offline Offline
  • Posts: 404
  • Respect: +84
    • View Profile
Re: Updating the Top 5 lists
« Reply #268 on: January 06, 2012, 01:39:51 pm »
0

I agree that you've been doing a good job and shouldn't stress too much about this. That said, here's my two cents:

 I prefer comparisons among cards of similar function (i.e., trash-for-benefit cards) instead of cards of equal price, but if we're doing the latter we should certainly consider the frequency with which any given card's upside outweighs its downside and by how much.  To throw in another wrinkle, I think it's worth considering how many of a card you need in order for the card to serve its purpose. You rarely want more than one or two Tacticians, whereas one or two labs are generally helpful but not nearly so much. That should nudge Lab down and Tactician up. Same for Spy (you need lots to do much good with them) and Coppersmith (one can be enough to have an impact).
Logged

HiveMindEmulator

  • Mountebank
  • *****
  • Offline Offline
  • Posts: 2222
  • Respect: +2118
    • View Profile
Re: Updating the Top 5 lists
« Reply #269 on: January 06, 2012, 01:52:00 pm »
+3

Here's a question I'm grappling with as I write the Worst $4's list:

Are these rankings based on how bad the cards are when you consider adding them to any random deck you're building?  Or are they based on how helpful they are when you do buy them?  Or both?

Clearly, throwing Coppersmith into a random deck is pretty awful.  But when you do buy Coppersmith and build for it, it's glorious.  Spy, on the other hand, is basically always a mediocre addition to your deck.

(Maybe multiply how good it is by how often it is likely to be good?)
I think I've mentioned this before, but considering adding them to "any random deck you're building" is a completely useless idea. No one reading the blog just buys random cards. "How often is it good" should mean how often do you actually actively want to add the card to your deck. You can then "multiply" (quotes because you're not really doing anything mathematical) this by how good/important it is when you do buy it. Considering randomly adding cards is going to naturally overrate cantrips and not really going to be useful.

A good heuristic I think is to ask yourself "if card X had a one-way embargo on it that only affected me and not my opponent, how much would I rage?". Rank them in order of decreasing rage.
Logged

Fabian

  • 2012 Swedish Champion
  • *
  • Offline Offline
  • Posts: 666
  • Respect: +542
    • View Profile
Re: Updating the Top 5 lists
« Reply #270 on: January 06, 2012, 02:13:13 pm »
0

HME has it right imo.
Logged

tlloyd

  • Tactician
  • *****
  • Offline Offline
  • Posts: 404
  • Respect: +84
    • View Profile
Re: Updating the Top 5 lists
« Reply #271 on: January 06, 2012, 02:53:51 pm »
0

The flipside of the question "how many of this card do I want/need" is the fact that sometimes you buy cards not because they actively contribute to your strategy, but in order to deny your opponent. I personally hate Fool's Gold, but if you let your opponent buy them all they are pretty strong. Should that affect FG's ranking?
Logged

HiveMindEmulator

  • Mountebank
  • *****
  • Offline Offline
  • Posts: 2222
  • Respect: +2118
    • View Profile
Re: Updating the Top 5 lists
« Reply #272 on: January 06, 2012, 02:59:58 pm »
0

The flipside of the question "how many of this card do I want/need" is the fact that sometimes you buy cards not because they actively contribute to your strategy, but in order to deny your opponent. I personally hate Fool's Gold, but if you let your opponent buy them all they are pretty strong. Should that affect FG's ranking?
I think this (how bad is it to ignore your opponent going for something) fits nicely into the one-way embargo heuristic.
Logged

Fabian

  • 2012 Swedish Champion
  • *
  • Offline Offline
  • Posts: 666
  • Respect: +542
    • View Profile
Re: Updating the Top 5 lists
« Reply #273 on: January 06, 2012, 03:01:22 pm »
0

I would say the part that "affects FG's ranking" is that "if you let your opponent buy them all they are pretty strong". Sounds like a strong card to me -> rank appropriately.

I guess "I have to buy 4 FG to deny my opponent all 10, because they're really strong if you get all 10" is sort of just another way of saying this though? Certainly cards like FG, Gardens, Duchy (in Duke games), maybe Minion, etc, are cards you often can't ignore if the opponent goes for that strategy (given that the kingdom is favorable of course), and that in itself speaks volumes about the card's strength I think.
Logged

tlloyd

  • Tactician
  • *****
  • Offline Offline
  • Posts: 404
  • Respect: +84
    • View Profile
Re: Updating the Top 5 lists
« Reply #274 on: January 06, 2012, 05:20:47 pm »
0

Right, but the interesting thing is how this effect differs card to card. I almost always buy a tactician if it's available, and two any time double-tactician is plausible (and often when it isn't!). But I think even in a six-player game it would be rare to see the Tactician pile emptied. Minion I also buy frequently, but a single Tactician is way better. But if you don't buy any and let your opponent get ten, that's just asking to lose. So how do we compare the cards?  Does the embargo heuristic work? Because a single curse wouldn't stop me from buying a Tactician, but it sure would stop a minion strategy in its tracks. So based on "relative rage" Minion seems stronger, when I think the opposite is true.
Logged
Pages: 1 ... 9 10 [11] 12  All
 

Page created in 0.101 seconds with 21 queries.