Cube draft

[BrotherD]

Draft is my second favorite format of magic, the first is of course Quest , and the draft pool in Forge is rather limited. I tried to replace them with all the cards availble in Forge, but u know, drafting 45 cards from a 6k+ card pool with all kinds of grizzly bears may not be very fun. So I made my own cube, which has a relatively small pool and almost all of the most powerful cards that forge offers, to draft.

The cube consists of 480 cards: 50 for each of the mono color, 55 artifacts, 65 lands, 10 for each bicolored and 10 tri-colored spell.

If you are interested you can download cube.zip, unzip and replace common.txt, uncommon.txt and rare.txt in \res\draft. All the 3 files are actually identical, since it's a cube draft I don't think the original rarity applies here.

The cuberoster.zip in cube.zip is a list of cards in every color, so you can easily see which card is in and replace the cards u don't like.

And last but not least, another big thank you to everyone contributed to this great program, I had lots of fun playing it.

Oh, and i have a small feature request, it would be really great if i could start a quest with the deck i draft. But currently decks exported from a draft cannot be imported to quests, i can only make a similar deck in deck editor and then import them.

[Corwin72]

I will try this out! Thanks!

[juzamjedi]

I haven't played Forge in a while - I have been playing Manalink. I actually did the same thing because drafting is much more fun with a constructed cube instead of random cards from all of Magic. My list is a bit outdated (I am sure there are a lot of cooler / better cards now) but for reference / comparison I'm posting it now.

I included a stats tab on my spreadsheet too, so I could keep track of distribution by card types. Green and White should have the most creatures and Blue should have the most instants. I didn't want too many planeswalkers.

After a lot of trial and error I decided to really cut down on multi-colored cards. Hybrid is good (e.g. Kitchen Finks), but straight gold cards like Terminate were bad because AI would frequently put them in a deck that didn't have the mana to cast that gold card. I did keep Wall of Denial though to keep an even distribution of cards for each color pair.

The worst part about my list is that it's too small. 266 cards means you get a lot of repeats. But maybe you will find it useful anyway.

[bane221]

Did you guys have any issues using Cube drafts in Forge?

[tchiseen]

Have you considered real-working your cube, now that cube drafting is officially supported in 5/17?

I would love to try drafting a different cube, as i've done about 20 with the same one, and it's starting to get predictable.

[friarsol]

I'd be interested in seeing a Cube with only cards that the AI is known to be able to handle, this might limit some "fun" Cube cards out for the Human, but could be a great test for drafting and deck building ability of the AI.

[Chris H.]

This would also make a good cube set to distribute with the forge project.

[juzamjedi]

Hi all,

Cube drafting is my primary use for Forge. And you all are correct, there are a lot of cards in this cube that AI won't play properly and/or draft properly. I have been having a lot of fun with drafting this cube, but there are only so many times you can draft all the Blue + power 9 before it gets old.

If dev's want to include my cube with the official distro that would be cool.

Edit: Friarsol, I read your message as you would want power in the cube as long as it's fully playable by AI. The problem is that I know AI is not prioritizing artifacts / colorless cards as highly as it should. Rereading your message now I think you want power in there anyway, correct?

[friarsol]

Edit: Friarsol, I read your message as you would want power in the cube as long as it's fully playable by AI. The problem is that I know AI is not prioritizing artifacts / colorless cards as highly as it should. Rereading your message now I think you want power in there anyway, correct?

I want the cube to be cards the AI can in full-confidence use.

Having a fairly sized cube where the AI can use every single card can aid the drafting AI as we can be focus our testing with the knowledge that the AI will not pass on cards it can't use, even if they are extremely strong (eg Black Lotus)

[juzamjedi]

OK, I can do that. Looking at my cube list now, I see some cards have AI = ?

Should those cards be excluded? I think most of hte cards actually get played by the AI. I have seen AI play Rishadan Port pretty well. On the other hand AI will play Pernicious Deed, but plays it badly.

[Chris H.]

OK, I can do that. Looking at my cube list now, I see some cards have AI = ?

Should those cards be excluded? I think most of hte cards actually get played by the AI. I have seen AI play Rishadan Port pretty well. On the other hand AI will play Pernicious Deed, but plays it badly.

`
Cards that have a "no" in the AI column are either unusable by the AI or will be used poorly.

Cards that have a "?" in the AI column are for the most part, combo cards. The AI may have the code needed to play the card but the the cards that are needed to make the combo work may be missing from the pool.

There are likely to be some cards that are miss-marked. We try our best but we are not perfect.

[tchiseen]

...there are only so many times you can draft all the Blue + power 9 before it gets old...

I was playing with your cube, and I was finding the same thing. Sometimes I'd feel like splashing black, but for the most part it's hard to pass up last-pick fact or fictions.. lol

[juzamjedi]

OK. I startd by taking my cube and removing all the cards that have AI = No. There were a LOT of Blue cards and a pretty good number of Black cards I have to replace now. Some of them make sense (I mean AI could waste Force of Will on literally anything) and some are head scratchers (what is difficult about Sol Ring?).

This ended up being a radical change, so I won't be posting tonight. I've had to add entire new elements to the cube that will take at least a little playtesting. Blue and Black will get some "aggro" cards to fill most of the missing parts.

[Chris H.]

The mana cost paying code for the AI is separate from the mana paying code for the hyuman. This has allowed some improvements for the huamn player.

The AI on the other hand ends up being far more complicated. Currently, the AI can only use mana sources that produce just one mana at a time.

The Sol Ring produces two and as such is incompatible currently with the AI mana paying code. At some point the computer mana system will be re-written and brought up to date.

Sol feels that it is best to work on other areas of the project first and leasve this part for a latter time. It could take a long time to write all of this code and then to de-bug and add in missing AI code to help clue the computer in how to use these new for the AI mana sources.

[juzamjedi]

Thanks for the explanation Chris

I playtested some drafts with the cube last night and so far I like the direction that Black is taking. I was able to push the Swamps Matters enough to make it worthwhile and Phyrexian Obliterator was a house in that deck. Blue still needs major rework and testing.