Collectible Card Game Headquarters

*Edit: based on feedback since I started this thread most people seem interested in the good limited set, minus broken cards. To that end I will periodically update this thread with whatever "best" limited set I am playing at the moment.

9-22-10: Large update. Fetchlands are gone with Shocklands / Terramorphic Expanse replacing them. Sol Ring is gone; it was easily the strongest card in the set and the AI never played it. Goal for this update was to include a better mix of "aggro" and "control" cards for every color. Blue has "aggro" cards like Daze, Welkin Tern, Rishadan Airship, etc. Green has "control" cards like Cultivate, Plummet, and Harmonize.

I suppose my reply in the bug section illustrates the current state of the drafting code in Forge quite well: it has been neglected.

Ayway, I've never really drafted at all, but what you're saying makes sense to me: drafts should use a limited cardpool, otherwise there's just no balance.
So for a start, I'll change some of the code so that drafting will use a different set of rarities files (later on, it could even be possible to choose between different "cubes" / sets of rarities files).

As for improving the AI's drafting strategies, this is much more difficult. Maybe we could have some rating system somehow?

Gando at one time created a set of rarity files that limited the cards found in a draft/quest pool of cards. Gando's cube set can be found here if anyone is interested in examining the list of cards included. Please note that a large number of cards have been added to forge and that some of the broken/unfinished cards that Gando ignored are now at least worthy of consideration for a cube rarity set.

Re: Helper Files - tokens, cards pictures, set editor


alternate rarity sets 05-28-2009.zip

Yes I can appreciate that a rating system would be easier to implement for the AI's drafting skills. If you want the AI to pick the "best card in it's colors" then it will most likely work out the best to have a rating included as part of the cube list somehow. The rating file should mix all rarities together because some commons like Lightning Bolt are strong first picks.

I would be willing to do the card rankings for the cube too. I'm not the world's best drafter but I'm way better than the computer.

Chris, the list from Gando sounds like a great place to start. I downloaded it and will try it out!

juzamjedi wrote:Chris, the list from Gando sounds like a great place to start. I downloaded it and will try it out!

`
There are two different rarity file editors in the topic for people to download. These editors allow you to determine exactly which cards are available for draft and quest.

The all rarity files that I release with each beta are designed to include the entire card set. The archive that you downloaded has four different sets and each set has it's own theme.

I've finished the code that uses a separate set of rarity files (I tested it with a "cube" only containing cards that start with A, B or C - not exactly the most balanced set ). So if anyone can make the rarity files containing some balanced subset I would be happy to include it with future releases. We could even have multiple sets of rarity files, to allow for different types of drafts.

I do love the drafting, but as you say, the bigger the card pool, the sillier the draft with regard to strategy.

Wouldn't the simplest thing to implement also be one of the most useful? That is, limiting cards by block or set. I don't know if card set is a field you keep, but I bet it would ratchet up the popularity of draft in a hurry!

I never had a better time in my MtG career than when I was drafting IPA block for about 6 months.

DennisBergkamp wrote:I've finished the code that uses a separate set of rarity files

`
I just updated my local copy to r632 and when I build the project I get these errors:

Code: Select all

Description   Resource   Path   Location   Type
ReadDraftBoosterPack cannot be resolved to a type   BoosterDraft.java   /ForgeSVN/src/forge   line 71   Java Problem
ReadDraftBoosterPack cannot be resolved to a type   BoosterDraft.java   /ForgeSVN/src/forge   line 71   Java Problem
ReadDraftBoosterPack cannot be resolved to a type   BoosterDraft.java   /ForgeSVN/src/forge   line 113   Java Problem
ReadDraftBoosterPack cannot be resolved to a type   BoosterDraft.java   /ForgeSVN/src/forge   line 113   Java Problem
ReadDraftBoosterPack cannot be resolved to a type   BoosterDraftAI.java   /ForgeSVN/src/forge   line 49   Java Problem
ReadDraftBoosterPack cannot be resolved to a type   Gui_BoosterDraft.java   /ForgeSVN/src/forge   line 427   Java Problem
ReadDraftBoosterPack cannot be resolved to a type   Gui_BoosterDraft.java   /ForgeSVN/src/forge   line 427   Java Problem
ReadDraftBoosterPack cannot be resolved to a type   Gui_BoosterDraft.java   /ForgeSVN/src/forge   line 501   Java Problem
ReadDraftBoosterPack cannot be resolved to a type   Gui_BoosterDraft.java   /ForgeSVN/src/forge   line 501   Java Problem
ReadDraftBoosterPack cannot be resolved to a type   BoosterDraftAI.java   /ForgeSVN/src/forge   line 49   Java Problem


Oops, should be fixed now

DennisBergkamp wrote:Oops, should be fixed now

`
Thank you.

Professor wrote:I never had a better time in my MtG career than when I was drafting IPA block for about 6 months.

Unfortunately I don't think any of the blocks are implemented at a high enough percentage that we could do it that way. 80% + of a block would probably be "close enough" to get the feel for it...

I put together an initial draft list and have been playtesting it. My draft file definitely has too many rares. My first pass allows a number of decks that are "good for limited". It has been hard to draft the "powerful effects" decks because they take up a lot of rare slots i.e. you don't always get the Darksteel Colossus to go with your Tinkers and moxen, but it's possible to get there. If some people could download this set and give me some feedback that would be great.

Excellent, I'll include this with the next release, thanks Juzamjedi!
Whatever small balance issues your version might have, I'm sure it's much better than using the entire card pool.

Dennis, do you have any idea how the AI currently values draft picks? I thought I put a good number of strong aggro creatures in the draft, yet I frequently feel that my aggro decks are lacking in the low end / efficient creature department. I frequently see Path to Exile, Repulse, and Lightning Bolt wheel around though (?)

I wanted the aggro decks in this cube to have synergies together. However, after more drafts I feel like the Mercenary theme in Black is pretty bad. For the next revision I think I am just going to pick the best black creatures at common and throw them in there. Any suggestions?

Sounds good, I'm not sure how the current AI picking strategy (or lack thereof) works.
As for certain cards showing up more often than others in drafts, I don't know either... but I'm pretty sure it's not 100% random at all (it favors a certain distribution of creatures/spells)... it probably also takes colors into account.
For both cases, I'll look into the code.

How does the AI pick cards? One of the only things that I know. Each of the 7 AI players pick a two color combination. First they pick 16 creatures in those colors then they choose spells. It sounds simple but getting the AI to actually pick cards and create a deck on the fly is pretty hard.

The simplest improvement would be to create a stable set of card for draft and then create separate files that lists the pick order for the AI. So the AI would be able to know to pick Lightning Bolt instead of Shock. (The AI code is in BoosterDraftAI.java in the forge package.)

The idea is that Forge gives an illusion of drafting, it won't ever compare to drafting in real life.

p.s.
I've been offline for around 15 days, I think its a new world record for me, ha.