Tuesday, April 12, 2016

Two-Player Draft Formats

I've been doing a good bit of two-player Limited over lunch the last few weeks, and I thought I'd share the formats we've done and considered, and a new one I'm eager to try. All but the last two work with 3 or 4p too.

Solomon Draft is very satisfying. Gather three booster packs per player and shuffle them all together. Take turns revealing seven cards (excluding the five basic land types) and splitting them into two* piles. Then your opponent chooses a pile and you get the remaining pile.
* More with more players.

Trying to balance 1v4 or 2v3 cards with power levels as varied as Magic cards rigorously tests your evaluation skills. Choosing between two piles of very closer power level—while considering what colors and synergies you've got so far, and that your opponent has so far—is a tough and very fun challenge.

The downside is that it's very easy to have both players end with half of each color and build 3+ color decks with insufficient fixing. Also, both players have seen every card, which hurts tricks and bluffing.

Winston Draft is for smart gamblers. Gather three booster packs per player and shuffle them all together. There will be three draft piles next to the deck. On your turn, add a card to the first draft pile, look at the pile, and decide whether you want to keep the entire pile or not. If so, the turn passes. Otherwise, do the same thing with the second pile. If you put that pile back too, do the same thing with the third pile. If you put the third pile back, take the top card of the deck and the turn passes. Always look at the piles in the same order.

It's really neat when a pile builds up with weak cards until the point someone decides quantity is better than quality. Every time you put a pile down, you're hoping that you'll see a better card to choose next. But there's also a chance you're putting something great in that pile. There's still a good bit of skill and card evaluation here, but also vastly more luck than Solomon Draft, which is good and bad. You see some of your opponent's cards but many you don't, so that knowledge affects game play more like a traditional draft.

Winchester Draft I'd forgotten about until writing this article. You shuffle your own three packs and each turn, every player puts two cards from their stack into two draft piles in front of them and the active player chooses a draft pile to take. This will play a lot like Winston Draft but entirely public. Because you see all possible piles, you don't have to worry about the possibility of passing up a decent pile in hopes of a better pile but getting stuck with dross.

Again, it's really cool how the worst pile will get better and better thanks to the sheer quantity of cards. Again, luck will often put the best new card into that pile to negate that coolness. When I try this, I'd like to try letting each player choose which of the two piles in front of them they want to put the two cards they reveal, to help alleviate that luck. That will increase draft time, but may be worth it.

Small World Draft is our first bidding format. Shuffle up your packs and deal five cards face up in a row next to the deck. Each player starts with 5 or so tokens and takes turns selecting a single card to take from the row. The furthest card is free. The second cards costs 1 and you pay by putting one of your tokens on the first card. When someone takes a card, they also get any tokens on it. The third card costs 2, which you pay by putting a token on the second card as well as the first. And so on. Whenever you buy a card, slide the cards behind it outward and put a new card from the deck into the last slot.

This takes a lot of time but is very cool. Do you bid the maximum to take a card when it first appears, or wait until your next turn to get it for 2 less? When do you take a weak card just to refill your token supply? Is that second card better than the first enough for you to put a token on the first, knowing your opponent will get the card and the token?

(I named this after Small World because it's better known and still available, but it's a remake of Vinci where I believe this bidding mechanic debuted.)

Auction Magic is a bidding format that happens while you play, not before. Invented by Reuben Covington and Dan Felder of Remaking Magic, I love quite a bit about this (but my lunch Magic partner doesn't). You shuffle up your packs into a single deck, and on each player's draw step, reveal one. You and your opponent bid on it, and whoever wins plays it immediately for free (or saves it to play for free later).

Shortcutting the mana economy of Magic changes the worth of cards drastically, as does the fact that you're in the middle of a game and so might be willing to pay more to get a blocker just to stem the blood flow from your opponent's turn one Moss Kami.

You're paying the tokens you bid to your opponent, so giving them even just 1 more than you means they can guarantee access to a huge bomb when it appears. Of course, if they do pay everything for that Blightsteel Colossus, you'll only have to bid 1 for that potential Angelic Edict, and you'll still have a lot more tokens than them. It's pretty sweet.

Epic is a Magic-inspired game that plays similarly to Auction Magic, minus the bidding. It's fast and smashy. It's not Magic, but it scratches the Limited Magic itch well and with zero set-up. Worth checking out.

Two-Headed In-Fighting is a Sealed format for two players. We improvised this format out of the frustration that the drafts we were doing weren't doing justice to the card pools. By collaborating to build two decks as if you were a two-headed giant team (from 8-12 packs), the decks that result will be as good as you can make with your shared pool.

The twist is that you want to make them as comparable in power level as possible because you won't be playing them next to each other but against. You could bid starting hand-size / life total for who gets which, but we just took turns with each deck. (If you do that, don't give the loser first pick, because they've got the winning deck.)

We did this once, and the decks were vastly to superior to anything we'd drafted against each other. The bad news is, we failed to balance them and the blue-black deck won every game. The good news is, the games were all really good and pretty close. The white-green deck put up a good fight.

Color Draft replaces Arena Draft (check the comments to see why). It's deadly fast and offers the highest deck-quality-to-boosters-opened ratio after Two-Headed In-Fighting. Open six boosters. Sort them into five piles by color identity (Narnam Cobra goes in the green pile) and leave the colorless and gold cards unsorted. Randomly choose a first player. That player chooses an entire color pile or any single card (usually, one of the colorless cards). The other player chooses two. The first player chooses two, and so on, until all cards are taken. As soon as a player takes all the piles associated with a gold card's color identity, they get that card too. Now build and play.

Color Draft is efficient and satisfying. I recommend it highly.


  1. Great stuff here. I like In-Fighting, it sounds quite nice, though as you said, difficult to ensure decks are balanced. Could you not do it Netrunner style, where each players gets N attempts with a deck?

    Arena's colorless bug is a big deal. I'd say to distribute the colorless cards among the piles, or have it be a smaller snake-draft or Solomon Draft, as you suggested for handling ties. It'd be best if it could be intergrated into the rest of the draft, though, which is why I think distribution might be nice.

    1. Each player plays N games with each deck. Is that different from Netrunner?

      My partner had suggested putting colorless cards into the colored piles as seems appropriate, and that can actually work for those that produce or consume a single color of mana, but where then do you put dual lands or truly colorless things? If players choose where they go, those choices are tainted by their existing bids and that seems problematic. I quite like the plan to just always draft the colorless cards.

      For most sets, the gold cards can be treated the same way as colorless. But things would fall apart for a gold-focused set. One possible solution is that a player who wins the white pile and the blue pile will also get any white-blue cards, but if those piles go to different players, no one gets the WU cards. But that would be sad in a non-gold focused set where the gold cards are always great.

    2. Ah, that's what you meant by 'turns with each deck'. Got it!

      I was thinking random distribution of colorless cards, and for gold cards, they randomly get distributed to the piles including their colors. Could that work?

      I think drafting colorless and gold separately is fine though.

  2. Have you tried Microsealed? http://www.channelfireball.com/articles/cube-design-microsealed/

    It's by far my favorite two-player limited format. Games are super skill-intensive, and the small decks mean you can get very creative with your deckbuilding.

    1. I've not, but it's perfect for this discussion. Thanks, Benny.

    2. This seems super cool, I'll have to try it!

  3. Having tried a bajillion two player draft methods, I have never found anything that was within three football fields of being as good as Winchester. It is the only draft format I have felt feels like drafting, and I think it addresses a clear issue with almost every other format: once people commit to colors, which tends to happen early, the only reason you would ever take an off color card is to hate draft. This makes formats based on auctioning individual cards incredibly dull in my experience (though you could certainly auction blocks of cards of mixed colors).

    Since Winchester has cards mixed together, a pile likely has things both you and your opponent value. Also, since you naturally accumulate cards of many "off" colors, you have the option to shift colors even relatively late into the draft, and I will often court two (or maybe even three) potential second colors.

    The other thing Winchester has that most two player drafting formats do not really have is serious consideration about wheeling. If you think a pile will still be there on your next turn, why not leave it? Of course, unlike in regular draft, you don't have perfect information, as a card will be added to the pile blind, but since there is so much other information bouncing around in Winchester draft, I don't mind that. (If you want, you can just reveal all the cards that will be added to piles before your turn)

    Speaking of information, the fact that you know literally every card that can possibly be in your opponent's deck is huge, and allows you to build a pre-sideboarded deck against your opponent (provided you can nail down what they are going to play). Sure, maindeck Deathmark and Naturalize. Better pick those walls highly if you see them piling up Borderland Marauders. Defining your role in the matchup (as the beatdown or control) happens naturally over the draft, and I've never seen this dynamic anywhere else.

    So yeah, if you're trying to decide what draft format to use for two, my advice is to just use Winchester. Nothing else comes anywhere close in my mind. Do note, it is best practice to use 10 or 12 packs in Winchester, not 6 as some originally suggested. That makes your decks feel pretty similar to limited decks from an 8 player draft. (I like doing 10 packs but with only 8 rares, but in a low power or high synergy format you might want 12 packs).

    Finally, I'll note another favorite two player limited format that I am a huge proponent of is two player team sealed. Each player gets 12 packs, and builds 3 decks, A, B, C out of them. Lay out the possible matches on a 3x3 grid, and the winner is the first to either win 5 matches, or win all the matches in a row, column, or diagonal. Loser picks the next match of course.

    Of course if you like two player team sealed and Winchester, you can have the ultimate two player limited experience by Winchester drafting 24 packs, and then building team sealed decks out of your pool.

  4. Arena Draft sounds fun but it seems like it is too easy for a player to match whatever their opponent did so that only the last couple of bids matter. This can be solved with another layer of hidden information or randomness. I would make Jack - Ace be worth 2 bids. Each player has a shuffled deck with their 13 cards. They draw 2 (or 3) cards and bids them face down.

    1. I was skeptical about some cards being worth two, but in our test today, the tie breakers never mattered. While they still could, using cards and choosing which to play where (facedown this time) is too much effort for no result. So maybe some of them need to be worth double.

  5. I got a second test of Arena Draft in. We still made better decks than via solomon, but a detrimental trend showed itself: In both our tests, one player got the two best colors while the other got the three worst, and that doesn't make for an ideal match-up. While it's true that players bidding optimally can and should avoid that, for the format to be worth sharing, it needs to push players that way naturally.

    At a very conceptual level, here's my thought: Instead of players bidding on stacks, you're secretly evaluating them. One of the two stacks evaluated the highest between both players will go to one player and the other won't.

    So let's say I rate
    W=4 U=2 B=3 R=2 G=4
    and you rate
    W=5 U=1 B=4 R=1 G=4
    We add those...
    W=9 U=3 B=7 R=3 G=8

    And find that neither of us should get both W and G.
    Since you rated W higher, you get W and I get G.
    Next we find the most even combination of the other three colors. In this case U+R v B. B's 7 beats U+R's 4, so I get that because G's 8 < W's 9.
    The result is BG (15) vs WUR (16).
    (Those numbers happen to be close, but they wouldn't always. That's probably fine?)

    1. This comment has been removed by the author.

    2. This comment has been removed by the author.

    3. For brevity, I removed the above two comments walking through exactly how the rating system can be gamed, whether we favor higher ratings or lower ratings.

    4. My lunch Magic partner came up with a super fast and simple way to determine who gets which color pile:

      Snake draft. Flip for first pick. Other player picks the next two. First player gets the last two.

      This way, the split is always 1+4+5 v 2+3. AKA Best color and two worst colors versus second- and third-best colors. As long as the middle-ranked color isn't nearly as good as the first and the fourth-ranked color isn't completely terrible, this should always be a good split.

      Gold cards go to whomever got all their colors. Truly colorless cards are snake drafted individually.

      Will try it soon and let you know how it goes.

    5. I really like the speed and elegance of this version.

      Reading this I wasn't sure whether 1+4+5 vs 2+3 was better than 1+4 vs 2+3+5 -- my first guess would be latter, since distributions typically have higher variance near the extremes, so the difference between 1 and 2 is likely on average to be slightly bigger than between 3 and 4. But I could easily believe that empirically the first format works better.

      Or you could add another level of balancing by having one player choose a format (e.g. 1+4+5 vs 2+3), and the other player choose whether to go first or second. My guess is that the slowdown from this wouldn't be worth it, though.

    6. Aaand we've managed to bring it all the way back to solomon draft. Awesome! One player splits the five color piles into two groups, and the other chooses. Then you reverse roles for splitting the colorless cards.

    7. We snake-drafted the piles, and then the colorless cards*. Worked rather well. In this case, four piles were surprisingly close in value. Will test again, to see how it works in other card pools.

      *Gold cards automatically go to a player who gets all colors they include. Otherwise, they go into the colorless pile.

    8. I replaced Arena Draft with Color Draft entirely in the main section above. Here, for posterity, is what used to be up there:

      Arena Draft is my newest idea. I've tried it once… writing it out is what inspired this article. Get your six boosters and open one each round, sorting the contents face-up between six piles by color and colorless (or maybe 7, separating lands from artifacts). Then both players bid 1 token on two of those piles (yours on your side and mine on my side, to track whose they are). Continue this way until all six boosters are sorted, except place three tokens in the final round. At that point, whoever has the highest bid on a pile takes it. I'll discuss ties in a bit.

      This way, you're going to be getting all of the cards in two or three colors, ensuring you can build a coherent deck. The lack of hate-drafting (present in all of the other formats) means you'll get to play all the best cards in your colors and so the "pool" should be done justice, in the sense that bombs that would be fun to play aren't sitting spitefully in someone's sideboard, or stretching someone's mana base as an awkward splash.

      The simple way to handle a tie is that neither player gets the pile. What we did in our first game was to solomon draft the tied piles. (What actually happened was that white was our only tie and there were only two cards worth fighting over—Oblivion Ring and Day of Judgment, so those were our piles and we flipped a coin to decide who got which. Day's {W}{W} requirement proved too much to splash for.) You could also snake-draft the pile. (We flip for first, I take one, you take two, I take two, etc.)

      What I'm planning to try next is using playing cards as our bidding tokens: Each player takes all 13 cards of a suit and our bidding is done with these cards. At the end, ties (on the number of cards bid) are broken by the highest unmatched card among the cards bid. (So, your King ♥, 2 ♥, 4 ♥ beats my 7 ♣, 9 ♣, 10 ♣.) What I'm still debating is whether we bid face-down or face-up, randomly or not. Some options are more interesting than others, but we need this mechanic to be quick and simple because it's not the focus of the bidding, just a tiebreaker.

      Our first go at this format was extremely promising. We both had two color decks with power, consistency and synergy. One player getting all the colorless fixing felt like a bug, though. I'll definitely update you as we play and learn more. (I named this format after Colossal Arena because it's the only game I can think of with the concept of progressive bidding for an entire class of cards.)

  6. I read and enjoy this blog, and am just de-lurking to share a couple of 2-player draft formats I've developed, as I thought the community here might appreciate them. In particular the one we're calling 'Minidraft' I think is a lot of fun as a smooth-flowing draft format that still feels roughly like a draft.

    Here's the text of an email I sent to share the formats with friends:

    1. I posted that quickly without much context because it was a busy week. I should add that 'Minidraft' is the combination of two different developments which are pretty separable:

      1) The draft process. We originally developed this for normal-sized decks and called it "charcoal draft" (because some cards are burned twice). Just double the number of cards you use if you want to do this (5 boosters per player instead of between you).

      I designed this because I often get the opportunity for 2-player drafts, but I found existing 2-player formats (Winston, Winchester, Rochester) somewhat unsatisfying: they offer complicated decisions which can give them a feeling of analysis paralysis, and every card that doesn't end up in your pool ends up in your opponent's, which makes them more zero-sum. For casual play, I prefer to feel that I should be focusing on my deck during the draft (as in regular booster draft). Charcoal draft tries to correct for this.

      2) Small decks. I give some analysis of this in the linked googledoc, but basically I think that small decks are a good idea whenever you plan to play with a deck only a handful of times. You could play with small decks with any other drafting (or limited) mechanism, scaling down the number of cards available similarly.

    2. The clarification is appreciated, Owen.

  7. A merging of Winchester/Winston and Vinci/Small World drafts:

    Reveal 4 cards in a row. On your turn, take one of those piles. Reveal and add a new card to each pile you skip. Then, move each pile left to fill in the gap and reveal a new card on the right.

    So, you can take the first pile free (in which case the second pile becomes the free pile for your opponent). Or you can add a card to the first pile to take the second (making the free pile better for your opponent). Or add a card to the first and second piles to take the third. Or to those three to take the fourth.

    Like Winchester, all information is public, but unlike Winchester, only bad piles improve. This is effectively Small World Draft with the currency eliminated. That elegance has me excited to try this one out.

    1. We've also done this where cards you add to a pile are face-down, which creates decisions like "Take this weak but playable card in slot 2, adding a random card to slot 1; or take the unplayable card in slot 1 and the unknown card that comes with it, sliding slot 2 into slot 1 for my opponent?"

      It's not a bad format.

      Our complaint with it is that sometimes the card that gets added to the first slot is a bomb, and that short-circuits the purpose of the system.

    2. Maybe this: There are three new card slots and two free pile slots. Fill each slot with a random card to start (but if any bombs would start in the free pile slots, swap them with the card in the newest premium slot).

      On your turn, take either free pile or any of the cards in premium slots. If you take a free pile, slide the all three cards in premium slots over, so that the oldest refills the pile you took. Then draw a new card (if there's still a deck) and add it to the newest premium slot.

      If you take the oldest premium slot, combine the two free piles into one, and slide the remaining premium cards over two places, one creating the new second free pile, and one filling the oldest premium slot. Refill the other two premium slots from the deck.

      If you take either of the other premium slots, slide the older premium cards into any of the free piles, slide any remaining premium cards over, and refill the now-empty slots.

      (More complicated to type than do, but maybe wonky in general. Will test and update.)

    3. While the process is awkward to explain, this played reasonably well.

  8. Wobbles mentioned Tic Tac draft and it sounds neat.
    Lay out a 3x3 grid of face-up cards. On your turn, choose a row or column of 3 cards, then replace them.

    1. This has been a lot of fun. It doesn't really solve any problems other formats don't, and it has its own—players often take the same row or column that was just taken because there's nothing adding new value to untouched lines—but picking three cards at a time across two-dimensions is choicey and satisfying.

    2. We did a variant where you start with a 3x3 grid, but each time you place one of your 3 new cards—before looking at it—choose any spot in a 4x4 grid to place it. Sometimes you take a row or column with just 2 cards, or with 4 cards, and that's fine.

  9. Peacemaker
    A modified Winchester draft. There are six piles. When you take a pile, replace it with a new card, then look at the top card of the deck and add it to any of the six piles face-down.

    Revealing fewer cards each turn reduces luck and makes choices more challenging/interesting. Choosing where to put the new card adds strategy and bluffing. Placing the new card face-down adds bluffing and memory (nothing's perfect).

  10. Simultaneous Bid
    Shuffle up six packs and deal a face-up card to each of six* piles, marked by a number or die face, 1-6.
    Each player sets 3d6 to any faces they choose and reveal their choices simultaneously. Each pile that you bid more dice on than anyone else, you claim. Add a new card to each pile that no one bid on. Replace the empty piles with one new card each. Repeat.

    * Worked well with 5. Might work with 4. You could add a third drafter by upgrading to d8s and using 6 or 7 piles.

    This was fun, predicting what the other drafters would choose. Sometimes each getting a 2 or 3 cards, sometimes one player getting one bomb and the other getting 4+ filler cards. Interesting decisions like bidding 3 dice for a bomb when your opponent might bid 2 or 0. It was a little slow, on account of the mind games, but I would do this again for sure.

  11. "A draft format in which the first seven picks form your starting hand, and each following pick happens when you would draw a card mid-game. There are basic lands in the pool."

    Live Booster Draft 2+ players
    Each player opens a pack and adds one of each basic land to it. Draft one and pass the pack to the left. Your first seven picks form the starting hand of your first game. Each time you would draw a card during that game, instead draft a card from the next pack.
    If you draw extra cards, take a pack from the right for each. In this way, packs will skip players.
    If you would interact with a library, shuffle the next pack and interact with that instead. (You can ignore scry effects, and library-ordering effects.)
    When all packs are empty, open a single new pack.

    Live Access Draft 2 players
    Each player opens three packs and splays them face-up without changing their order. Gather three of each basic land. Take turns drafting: Either take a card you have access to in any pack, or take a basic land and increase your access to a pack. You can always access the top-most card of each pack. When you take a land, choose a pack and mark a card three deeper than your current access point.
    Your first seven picks form the starting hand of your first game. Each time you would draw a card during that game, instead draft a card.
    If you would interact with a library, choose a pack and treat that as your library for that effect. If a player had 3-deep access to a pack and its order is rearranged, they still have 3-deep access after, etc.

    1. I just did Live Booster Draft twice (with Eternal Masters) and it was So Much Fun. Even with just two players!

      The number of basic lands in each pack was perfect and having to draft them along with spells was such a novel experience and really interesting: If there are spells that cost double of a single color, you have to get two sources of that color to cast them at all, and if the packs don't include extra basics, dual lands, or other mana sources, you have to draft both of the two basic lands of that type added to your packs (in 2p).

      Things we discovered:
      • One booster pack per player is enough for most games. That makes this a potential replacement for Mini-Master and based on the two games we played, a huge upgrade too.
      • When you search a pack as if it's your library, does it move ahead (as it does when you draft from it each time you would draw) or does it stay in place? We went with the latter in the moment, and it matters a lot in case-by-case scenarios. Maybe it's more intuitive to pass it along as well?
      • You could replay the deck after game 1, but you'd have to add land. Maybe you can add as many basics of you like, or maybe you can only duplicate the basics you drafted? Haven't done this yet.
      • If you live draft twice, you could then combine your two decks into a single deck for round 3. (Again, adding land. Potentially allowing cards to be removed from your pool, with a 30- or 40-card minimum deck size.)

    2. Speaking of group sizes:
      2p: Duel
      3p: Three-Headed In-Fighting
      4p: Two-Headed Giant or two duels
      5p: Star or Conspiracy
      6p: Three duels or Conspiracy or Emperor
      7p: Conspiracy
      8p: Two 2HG games or four duels or Conspiracy

    3. We combined our two live booster draft pools from yesterday into a 30-card deck today (adding basic lands as we like), and it was good.

      We drafted once more—and oh man, trying to draft so that you'll win that individual game while looking forward to combining those cards into your larger pool… that is hard and fun.

      Next we'll combine all three pools and build 40-card decks.

      I've never gotten so much 2p play out of six booster packs.

      With 5+ players, you could each play a live booster draft against an opponent, again against another, play 30-card decks against a third, live draft again vs #4, and play 40-card decks against a fourth opponent.

  12. We did the Access Draft mentioned above (not live). While novel, it wasn't nearly worth the added time and fiddlyness.