Dominion Strategy Forum

Please login or register.

Login with username, password and session length

Show Posts

This section allows you to view all posts made by this member. Note that you can only see posts made in areas you currently have access to.

Messages - rrenaud

Filter to certain boards:

Pages: 1 ... 37 38 [39]
951
Council Room Feedback / Re: Badges
« on: June 16, 2011, 12:03:03 pm »
You see all of your badges.  This page shows them all by name, and also ranks people by who is the most goal getting according to some weird formula I came up with that I never tested but seemed reasonable 3 months ago.

http://councilroom.com/goals

But i mostly gave up on it.  Cute isn't really my thing.

952
Dominion General Discussion / Re: First player advantage
« on: June 16, 2011, 11:53:28 am »
Also, I agree that averages are prettty dumb with respect to dominion scores. But median score differences on the other hand are pretty reasonable/robust.

953
Dominion General Discussion / Re: First player advantage
« on: June 16, 2011, 09:25:33 am »
Don't trust me! Get the data and do your own analysis, http://councilroom.com/margin.txt.gz
Or find bugs in the code! https://github.com/rrenaud/dominionstats/blob/master/oneoff/process_margin_advantage.py

The second column is the marginal difference between the first player and the second player in the median game when sorted by the score difference across all games on isotropic.
The third column is the marginal difference between first and second player with the first player skill bias removed (eg, pair up two players, only take the same number of games from when player A was first as when player B was first, if they only played one game, it will be thrown out, if A was first once, and B was first 10 times, take the 1 game from A, and 1 game at random from B).

The first player loses a half point for having more turns than player two.

You can view the third column as a reasonable upper bound on what a player should willing to bid.  The games are played on iso with no bidding, so if there was bidding, the second player could use his point advantage to close games that he actually lost and turn them into wins (they whole point).  Thankfully, the third column is quite a bit higher than my heuristic estimate of 1.5 points ;).


rrenaud@kingscourt:~/dominionstats/oneoff$ python process_margin_advantage.py
from 1122324 sampled down to 329378
*all                           1.0     2.5
Adventurer                     0.5     2.5
Alchemist                      1.0     3.0
Ambassador                     1.0     3.5
Apothecary                     1.0     3.0
Apprentice                     1.0     2.5
Bank                           1.0     3.0
Baron                          0.5     2.5
Bazaar                         1.0     2.5
Bishop                         1.0     3.0
Black Market                   1.0     3.0
Bridge                         1.0     2.5
Bureaucrat                     1.0     2.5
Caravan                        1.0     2.5
Cellar                         1.0     2.5
Chancellor                     1.0     2.5
Chapel                         0.5     3.0
City                           0.5     3.0
Colony                         1.0     4.0
Conspirator                    0.5     2.5
Contraband                     1.0     3.0
Coppersmith                    1.0     2.5
Council Room                   1.0     2.5
Counting House                 1.0     3.0
Courtyard                      1.0     2.5
Cutpurse                       0.5     2.5
Duke                           1.0     2.5
Embargo                        1.0     2.5
Envoy                          1.0     2.5
Expand                         1.0     2.5
Explorer                       1.0     2.5
Fairgrounds                    1.0     3.5
Familiar                       1.5     4.0
Farming Village                1.0     3.0
Feast                          1.0     2.5
Festival                       0.5     2.5
Fishing Village                1.0     3.0
Forge                          1.0     3.0
Fortune Teller                 1.0     2.5
Gardens                        1.0     2.5
Ghost Ship                     1.0     2.5
Golem                          1.0     3.0
Goons                          1.5     4.0
Grand Market                   1.0     3.0
Great Hall                     1.0     2.5
Hamlet                         1.0     3.0
Harem                          1.0     2.5
Harvest                        1.0     3.0
Haven                          1.0     2.5
Herbalist                      1.0     3.0
Hoard                          1.0     3.0
Horn of Plenty                 1.0     3.0
Horse Traders                  1.0     2.5
Hunting Party                  1.0     3.0
Ironworks                      0.5     2.5
Island                         1.0     2.5
Jester                         1.0     3.0
King's Court                   1.0     3.5
Laboratory                     1.0     2.5
Library                        1.0     2.5
Lighthouse                     1.0     2.5
Loan                           0.5     3.0
Lookout                        1.0     2.5
Market                         1.0     2.5
Masquerade                     0.5     2.5
Menagerie                      1.0     3.0
Merchant Ship                  1.0     2.5
Militia                        1.0     2.5
Mine                           1.0     2.5
Mining Village                 1.0     2.5
Minion                         1.0     2.5
Mint                           1.0     3.0
Moat                           0.5     2.5
Moneylender                    1.0     2.5
Monument                       1.0     3.0
Mountebank                     1.0     3.5
Native Village                 1.0     2.5
Navigator                      1.0     2.5
Nobles                         1.0     3.0
Outpost                        1.0     2.5
Pawn                           0.5     2.5
Pearl Diver                    1.0     2.5
Peddler                        1.0     3.0
Philosopher's Stone            1.0     2.5
Pirate Ship                    0.5     2.5
Platinum                       1.0     4.0
Possession                     0.5     3.0
Potion                         1.0     3.0
Quarry                         1.0     2.5
Rabble                         1.0     2.5
Remake                         1.0     3.0
Remodel                        0.5     2.5
Royal Seal                     1.0     3.0
Saboteur                       1.0     2.5
Salvager                       1.0     2.5
Scout                          0.5     2.5
Scrying Pool                   1.0     3.0
Sea Hag                        1.0     3.0
Secret Chamber                 1.0     2.5
Shanty Town                    1.0     2.5
Smithy                         1.0     2.5
Smugglers                      0.5     2.5
Spy                            1.0     2.5
Stash                          1.0     2.5
Steward                        1.0     3.0
Swindler                       1.0     3.0
Tactician                      1.0     2.5
Talisman                       1.0     3.0
Thief                          0.5     2.5
Throne Room                    1.0     2.5
Torturer                       1.0     3.0
Tournament                     1.5     4.0
Trade Route                    0.5     2.5
Trading Post                   1.0     2.5
Transmute                      1.0     3.0
Treasure Map                   1.0     3.0
Treasury                       1.0     2.5
Tribute                        0.5     2.5
University                     1.0     2.5
Upgrade                        1.0     2.5
Vault                          1.0     3.0
Venture                        1.0     3.0
Village                        1.0     2.5
Vineyard                       1.0     3.0
Warehouse                      1.0     2.5
Watchtower                     1.0     2.5
Wharf                          1.0     3.0
Wishing Well                   0.5     2.5
Witch                          1.5     3.5
Woodcutter                     1.0     2.5
Worker's Village               1.0     3.0
Workshop                       1.0     2.5
Young Witch                    1.5     3.5

954
Dominion General Discussion / Re: First player advantage
« on: June 16, 2011, 09:06:39 am »
It's like trying to extract useful information from experiments conducted in randomly-varying conditions, with each experiment itself producing random data.

FWIW, I think scientists call the process of learning from experiments with randomly varying conditions with each experiment producing random data "science."

Of course, they try to control for important factors, and then run what they intend as the same experiment over and over to get rid of that noise.

955
Council Room Feedback / Re: Win Rate vs. Turn Card was played
« on: June 16, 2011, 01:42:25 am »
I already have the per turn play data in the database.  I worry that it will be noisy and also not so informative (high mean reversion, for a given card, it's probably going to be played a lot by both sides).  But yeah, it's wouldn't be super hard to do it. 

(#include std blah blah, i'd rather help you help the community than help the community directly.. ).

956
Dominion General Discussion / Re: Is the point counter cheating?
« on: June 15, 2011, 09:53:06 pm »
I refuse to get sucked into this argument again ;).

957
Feedback / Re: Dominion Card Beneath User Name????
« on: June 15, 2011, 09:19:46 pm »
I think I've gotta pick the ambassador.

958
Feedback / Re: Dominion Card Beneath User Name????
« on: June 15, 2011, 05:16:17 pm »
theory's representative card will forever be chapel.

959
Council Room Feedback / Re: Win Type
« on: June 15, 2011, 03:53:28 pm »
This is the kind of thing I'd really love other people to contribute.  It's a nice way to ease into the project, it wouldn't really be all that hard, and it could be useful.  I already record the game end reason from the logs and import it into the database (https://github.com/rrenaud/dominionstats/blob/master/parse_game.py#L282), and while the code that renders the player pages is kind of a mess, it's not really doing anything complicated (https://github.com/rrenaud/dominionstats/blob/master/frontend.py#L195)

960
Other Games / Re: What other games do you enjoy playing?
« on: June 15, 2011, 02:37:17 pm »
FWIW, "national" is a bit of a stretch.  WBC mostly attracts people from the mid atlantic region.  OTOH, there was a tournament where the game designer was present near Portland, Oregon, and me and two of my friends flew from NYC and took 1,2, and 3, so that was pretty cool.

961
Dominion Articles / Re: Hunting Party
« on: June 15, 2011, 02:22:12 pm »
I am happy to have links to the cr.com versions, but I don't think Doug will follow through and expire his game links.  He just didn't want people to go and download all million game pages one at a time, instead he'd rather have you download ~200 compressed tarballs.

If you want to download the all the tarballs, I wrote a script that will first try to download the tarballs from cr.com, and then from isotropic if my server doesn't have them, which minimizes load on Doug's server.  https://github.com/rrenaud/dominionstats/blob/master/scrape.py

962
Council Room Feedback / Re: Bugs
« on: June 15, 2011, 01:59:32 pm »
There is a parsing process that converts the raw game logs into a useable format (it's by far the nastiest part of all the code) that gets imported into the database, but I only run the new parsing code on new data.  Every once in awhile I update the parser to extract more info from logs or fix a bug, and then I re-run the whole thing.

963
Council Room Feedback / Re: Bugs
« on: June 15, 2011, 12:58:48 pm »
Thanks, I'll probably take a look at it tonight.

964
Dominion General Discussion / Re: First player advantage
« on: June 14, 2011, 11:24:34 pm »
Maybe it's worth adding a straight win/loss record to keep a sanity check?

965
Even worse, we aren't British ;P.  From brief perusal of the internet, I think degree level is what Americans call undergraduate college education.

966
Dominion Articles / Re: Hunting Party
« on: June 14, 2011, 12:39:13 pm »
And yet it's still a level+1 opening, while Laboratory/- is level 0! I take this as a strong sign of how good Hunting Party is: it's better than Laboratory even when it is almost guaranteed to fail the first time.

The first HP will fail less than you think.  If you draw it on turn 4 or 5 rather than turn 3 (which is more than half the time), and you also buy something on turn 3 like a Silver, it will actually still draw 2 cards and find that Silver.

967
Dominion General Discussion / Re: First player advantage
« on: June 14, 2011, 02:45:57 am »
I definitely think the average value of first player (.5 turn) should be more than 1 vp on average.  Consider a hypothetical end game turn, say turn 16.  You can probably pull a duchy.  So heuristically, you'd expect a turn to be worth at least 3, and first player advantage to be worth half a turn, so that would put it at 1.5 points.

Alternatively, in a 16 turn game, you expect to have at least 27 points (4 provs + 3 estates), which means the average value of your turns is 27 / 16 ~= 1.6.  Halving this would only give .8, but then the early turns are buildup, and the later turns are scoring, you the marginal value of an extra 'last' turn should almost certainly be greater than average of 1.6 points.

968
Dominion Articles / Re: Hunting Party
« on: June 13, 2011, 09:29:34 pm »
I think it's basically better than a Lab, except in rare situations.  Certainly a super trashed deck might be one of those situations, but I still would like opening HP + chapel, since the HP is going to hunt down and find that chapel so I can get to play it more early, even if the HP eventually becomes a pearl diver in my super trim deck.

http://councilroom.com/win_weighted_accum_turn.html?cards=Laboratory%2CHunting%20Party

Certainly, you can more easily overdose on HPs than labs.  You want to be careful about not having so many of them that you play until you cycle your deck (turning the remaining draw into mostly copper/estate garbage) and then draw further, turning the garbage from discard (fine) into garbage in your deck and killing your next turn.  And you also want to be careful that you are actually almost always getting +2 cards with the HPs, so you can't have all that many HPs in a small deck.

It's fine to have multiple copies of non-terminals with the HP, just play the non-terminal before the next HP so you can draw other non-terminals of the same kind. 

It's great with Goons or other nasty attacks because it so quickly turns them from purchased in discard to in hand. And it's fine to a have a pretty terminal heavy deck (as long as you have corresponding village types).  Indeed, for the mixed terminal and +action deck, HPs are probably better than labs.  With a terminal heavy deck, if you were free to order your deck in anyway that you wanted, you would put infrastructure (village, throne rooms, etc) cards first and then terminals last.  In fact, HP pushes your deck in this direction.  Start with a hand of 4 junk cards and a Lab or HP.  If the next sequence of cards on your deck was Goons, Goons, Village, Village, and you had one Lab in hand, you'd draw into two Goons, never hit the Villages, and be sad.  With the HP, you skip the second Goons and hit the juicy Village chain, and you can possibly draw through the deck and find the Goons that was discarded earlier in your turn.  You can keep one of the terminal in your hand for most of your turn, and use the HPs to hunt down +action cards, and then unleash the nasty terminals after you've got enough actions (because you used it to find your infrastructure) to play the nasty terminals and still keep your turn going.

969
Dominion General Discussion / Re: Tournament + Black Market
« on: June 13, 2011, 08:50:12 pm »
Is it that much better than getting a witch or mountebank in a game with no trashing?  Or pulling out a trasher early in a game with cursing but no trashing?  The tournament still has a lot of lag to catch up with the other kind of advantages you can get from the black market.

970
Game Reports / Re: Fast Gardens Game
« on: June 13, 2011, 06:34:56 pm »
There is a fair amount of evidence that suggests simple bishop decks smack gardens decks.

http://boardgamegeek.com/thread/620889/bishop-vs-gardens

971
Other Games / Re: What other games do you enjoy playing?
« on: June 13, 2011, 05:14:31 pm »
I played a little bit of seven wonders and wasn't all that impressed.  I think theory has a somewhat more favorable opinion of it though.

After reading the answers to my post, I think the game just isn't that deep.

http://boardgamegeek.com/thread/599990/is-there-much-strategy-and-planning-in-seven-wonde

972
Other Games / Re: What other games do you enjoy playing?
« on: June 13, 2011, 03:49:52 pm »
I used to play starcraft a lot.  It was basically my religion in high school.

Before I switched to Dominion, I was pretty good at Race for the Galaxy.  I definitely recommend RftG if you have a competitive friend and Dominion is feeling stale.

973
Dominion General Discussion / Re: Best/Worst Openings discussion
« on: June 13, 2011, 12:07:48 pm »
On the technical side, I have some trouble understanding what the ratings mean and how they are calculated. For one thing, if I open the same thing with say, LV 5 with my opponent, but I am at LV 40 and he is LV 0, now we become LV 45 and LV 5, which persumably would have a different win rate I think. (Or does the win rate depend only on the difference of the level of the players? If so it seems a rather strong assumption for the levels to follow a normal distribution.) Even for asymmetric openings, it seems very impossible for every kind of player match up to shift for the same amount for the different openings. It would be easier for me to understand if there is just an explanation of how it is calculated.

TrueSkill makes the normal distribution assumption.  So holding variance constant, if you add a fixed level to both you and your opponents skills, the model says there is no difference.  The hard part is to understand TrueSkill.  http://www.moserware.com/2010/03/computing-your-skill.html

The openings data also makes the assumption that you and your opening form a team and that you add your skills together with equal weight on the player skill and opening skill.  I'd guess the player matters much more than the opening, but optimizing this for the right criteria for multiplayer games is hard, http://stats.stackexchange.com/questions/10067/maximum-of-two-or-more-gaussian-distributions-with-known-and-possibly-different).  Likewise the opponents team does the same thing, then you model a little noise from the outcome of the game, and then your combined normal distributions fight.

974
Dominion General Discussion / Re: Duke Analysis?
« on: June 13, 2011, 11:53:28 am »
This page does the math right on the balance between Dukes and Duchies.

http://blog.gameroom.io/

Pages: 1 ... 37 38 [39]

Page created in 0.114 seconds with 18 queries.