A Coconut Deckbuilding Template: 60 Cards, 48 Inkable, 12 Card Advantage

You picked a coconut, you have the named card, and now a pile of playable cards has to become 60. This post gives one number per category for how many cards of each kind go in. The numbers come from simulations and probability math on a 60-card singleton deck. My 18 lists and 89 public community lists are there for comparison. The cases where a number should bend have their own section further down.

Coconut is a social format, and this template is not a recipe for the most competitive deck possible. I do not think decks tuned to the limit come from card-count formulas anyway. The goal is a first version that works well enough for the game to be fun. Commander players have a name for that target, the 75% deck. I do not need the strongest list at the table, and I do not want it. I do need a deck that does not stall on ink and sit there with nothing to play. These numbers are broad averages that tend to work. They are a guide for someone building their first decks, and a fast first draft for someone who has built many.

Each number here is a default, and any of them can change in a specific deck. A deck can run fewer copies of the named card, or none, for whatever reason its builder has. On average I want all 4, and I start cutting copies after testing, or when I have a very specific idea in mind. Deck size works the same way. A deck that puts many of its own cards into its discard, or draws a lot, can want more than 60. Those cases should be the exception.

The rows from card advantage to plan cards add up to about 53, and I left the gap to 60 open on purpose. A card can also fill two rows: Scrooge McDuck — Ebenezer Scrooge makes each opponent lose 1 lore whenever he quests and draws a card for each lore lost, so he fills a lore engine slot and a card advantage slot. Each card that does two jobs frees one more slot for the plan.

60 cards, exactly

The rule says 60 minimum. Frank Karsten’s argument for Core and Infinity, the 1v1 formats, applies here without changes: a 61-card deck is the average of the 61 decks you get by removing one card from it, and an average cannot be better than the best of them. The argument is in his deckbuilding article on the official Lorcana site. All 18 of my lists are at 60. Of the 89 community lists, 72 are at 60.

4 copies of the named card

81 of the 89 community lists run all four. With a hard mulligan for it, where I put back every card that is not the named card, four copies give me the named card in the opening hand 66.5% of the time, and by turn 4 about 75% of the time. So the named card is likely, not guaranteed. A plan that needs it on turn 3 also needs tutors, the cards that search the deck for it, or a second plan.

48 inkable, which means 12 non-inkable

I find it easier to count the other side: a budget of 12 non-inkable cards. I simulated 200,000 games where the deck inks one card and plays one card each turn, and counted the games that missed at least one ink drop.

Up to turn 6, cutting non-inkable cards fixes the problem. By turn 8 the first column is bad in all three rows, because the hand is empty, and extra draw is what fixes that. This is why the next category is as large as 12.

A card I refuse to ink because the deck needs it on the table counts inside the same 12. Going up to 15 is fine when the deck really draws 1.25 cards per turn or more. For comparison, Karsten recommends 11 to 13 for 1v1.

12 card advantage

A card is card advantage when it leaves me with more cards than I had before I played it. An action leaves my hand when I play it, so an action has to draw 2 or more to count. A character, item or location stays in play, so one that draws a card when played, or draws again and again, counts.

• Counts: Friends on the Other Side, “Draw 2 cards”. Genie — Wish Fulfilled, a character with “When you play this character, draw a card”. Repeatable draw on a character, item or location. Characters and items that return a card from my discard to my hand.
• Does not count: Inner Strength says “Chosen character gets +1 {S} this turn. Draw a card”. I spent a card and drew a card. Let the Storm Rage On is the same, and it goes in the removal row. Develop Your Brain looks at the top 2 cards of my deck and puts one into my hand, which is card selection. Ransack says “Draw 2 cards, then choose and discard 2 cards”, which is called loot. All four are good at finding ink and the right card. They do not leave me with more cards.
• Does not count either: draw that everyone gets.

A Whole New World is an example of a card that draws but does not count as card advantage. It says “Each player discards their hand and draws 7 cards”. At a table of four it gives me 7 cards and gives my opponents 21. Diablo — Devoted Herald is the opposite case. His text is “During each opponent’s turn, whenever they draw a card while this character is exerted, you may draw a card”, so the extra card is mine.

The number comes from the game. Going first, by the end of turn 8 I have seen 14 cards: 7 in the opening hand and 7 draws. Inking on all 8 turns takes 8 of them, which leaves about 6 cards to play across 8 turns. A drawn card helps on both sides, because it can be played or inked. But each draw card in the deck takes the slot of a plan card, so at some point more draw means less on the table.

To find that point I simulated the template deck with 0 to 20 card advantage cards in it. Each row of the table is one deck. Cards seen and missed ink drops are counted at turn 8. The columns after that count the plan cards I have on the table at the end of turn 10: once in a game where nobody touches my table, and once in a game where an opponent plays a board wipe, a card that banishes every character on my table, at the end of turn 6.

This model plays every card it can pay for, so it empties the hand faster than the ink table above: the same deck with no draw misses an ink drop in 49% of games here and in 40% there. Read each table by its own rows.

With no board wipe, the best count is 10. After the board wipe, the best count is 16. The deck with more draw has more cards in hand right after its table is cleared, 3.5 with 16 against 2.0 with 10, and those cards are what rebuilds the table. With 12, I am within 2% of the best count when nobody clears my table, 9.3 against 9.4. After the board wipe, 12 has 4.2 plan cards on the table against 3.9 with 10.

The board wipe in the model takes my whole table, which is the hardest case. Grab Your Sword leaves a character with more than 2 willpower standing, and Be King Undisputed takes one character per opponent.

My first version of this model played alone, and it said 10. Then I let the opponents clear my table, and that moved the number to 12.

My 18 lists run a median of 9.5 and the community lists average 9.8. So the lists run about 10, and the template asks for 2 more because of the rebuild after a board wipe. With 12 in the deck, 80.9% of opening hands have at least one before any mulligan.

4 hard removal, 3 board-wide cards

Hard removal answers one opposing character for real: banish, 2 or more damage, or return to hand. One damage, exert effects like Elsa — Snow Queen, and “can’t quest” effects are useful, and they do not fill this row. Narrow removal does not fill it either. The Sword of Hercules banishes an opposing Deity character. I would play it when Deity characters show up at my table.

A board-wide card hits every opposing board at once. Grab Your Sword deals 2 damage to each opposing character. Be Prepared banishes all characters, mine included, and it is the real board wipe of the group. Be King Undisputed makes each opponent choose and banish one of their characters, so it removes their worst one. A board-wide card keeps its full value against three boards, which single-target removal cannot do.

Both numbers come from the same question: how often do I want to have drawn one by turn 8? This is the chance of at least one, with no mulligan. The middle column is a deck with no extra draw, which sees 14 cards. The right column is the template deck, which sees about 19 with its 12 card advantage cards.

Each extra copy adds less than the one before. In the 14-card column, from 2 to 3 adds 14 points and from 5 to 6 adds 6. For single-target removal I want an answer in two games out of three before I count on any draw, and that is 4. My own 18 lists run a median of 2, which means no answer by turn 8 in more than half of my games when the draw does not show up. I like spending those slots on my own plan. I am not going to call that the default for anyone else.

For board-wide cards I accept one game in two, which is 3. At least one of the 3 has to handle a big board by itself, which means 2 damage or more to each opposing character. A card that deals 1 damage to each opposing character fills a slot in this row and does not do that job. Grab Your Sword costs 5 ink and Be Prepared costs 7, and both need full opposing boards to pay for themselves. The same table explains why I build around them even when I run few. If each player runs 3, at least one of my three opponents has drawn one by turn 8 in 91.3% of games.

4 lore engines

This category comes from the win condition. The race is to 25, and questing alone gets there late. A lore engine is a repeatable source of lore, or of lore loss for opponents, on top of the lore a character quests for: Scrooge McDuck — Ebenezer Scrooge, Jasmine — Rebellious Princess. One-time lore cards are burst and I count them with the plan.

An engine needs time in play to pay. I use the same goal as for removal, two games out of three. In the table above that is 4 copies: 66.5% by turn 8 with no extra draw, 79.2% with it. My 18 lists run a median of 4 as well. The range is 0 to 13, and both zeros, Snow White and Winnie the Pooh, are decks that fill the board with characters and win by questing.

The curve peaks at 3

The averages of my 18 lists by printed cost are 6.6, 12.3, 13.3, 11.8, 8.3 and 7.7. The community lists also peak at 3, and so does Karsten’s 1v1 skeleton. The difference from 1v1 is at cost 6 or more: his skeleton has 4 cards for turn 6 and later, my lists average 7.7 at cost 6 or more, and the community averages 10.3.

Printed cost can mislead. Karsten counts a card by the turn he plans to play it, and that idea matters even more here. My Mickey Mouse list has 7 cards printed at cost 7 or more. 6 of them are Mickey Mouse characters, and the coconut gives Mickey Mouse character cards Shift 2, so I play them for 2 ink on top of another Mickey Mouse. My Mufasa list has 26 cards at cost 6 or more on top of 14 ramp cards, which put extra cards into the inkwell.

8 similar cards make a theme

Singleton leaves one 4-of, the named card, so a theme is built from different cards that do the same job. With the same hard mulligan:

Karsten’s Rule of Eight for 1v1 holds in singleton: 8 similar cards are there at the start of 9 games in 10. I use 6 as the floor for a secondary theme. Below that, the cards are bonuses and the deck cannot depend on them.

When to deviate

My test for leaving a default is whether I can say the reason in one sentence. These are the reasons in my own lists:

• Card advantage 10 or 14, by how often my table gets cleared. At a table that rarely plays board-wide cards I would run 10. At a table that plays them often I would run 14.
• Card advantage 22 in Nick Wilde and 17 in Dumbo. In Dumbo the named card exerts and pays 1 ink to “Draw a card and gain 1 lore”, and it gives that ability to my other characters with Evasive. In Nick Wilde the coconut banishes 4 of my items for 4 lore, and the deck refills them with cards that return item cards from the discard to my hand, so the category and the plan are the same cards.
• Loot 13 in Snow White. Loot sits outside card advantage in my count, and that deck needs it to assemble its combo.
• Ramp and card selection are Sapphire territory. The median across my lists is 0 ramp cards, and the four lists with 6 to 14 all run Sapphire. In those lists ramp takes its slots from the plan and the deck can run more cards at cost 6 or more. Without Sapphire I skip the category.
• Hard removal 2 or 3. If each player runs 4, at least one of my three opponents has drawn one by turn 8 in 96.3% of games. My lists run 2 because I count on a threat to me also being a threat to another player. That takes table talk, so I would not start there.
• Non-inkable up to 15, when the deck has a repeatable draw source it can count on. A board-wide card from an opponent removes that source when it is a character, so I keep 12 as the default.

What the community lists do differently

The 89 lists are 89 of the 90 public Coconut decks on duels.ink, captured on August 26, 2026. Published lists are not the same thing as played decks, so I read these as things to investigate. The community numbers count distinct cards per deck, and 17 of the 89 lists have more than 60 cards.

In my column every row is a median except cost 6 or more, which is an average. In the community column, non-inkable is a median and the other rows are averages per list.

The community runs 3 more non-inkable cards and 2 fewer card advantage cards than the template, so by the ink table those lists miss ink drops more often. It runs twice my removal, and on that row the math above agrees with the community. It plays more cards at cost 6 or more and fewer lore engines.

If you come from Commander

The format of this post is borrowed from The Command Zone’s deckbuilding template episode: one number per category and the math shown next to it. The test in the deviation section comes from their follow-up episode on the problems with templates.

Some things transfer and some do not:

• Their template asks for 12 card advantage in 100 cards, and a card that only replaces itself is outside their count too. Mine asks for 12 in 60 by the same count, a bigger share of the deck, because in Lorcana a drawn card is also ink.
• Their rule that cost 2 should be the biggest slot does not transfer. Lorcana decks peak at 3 in 1v1 and in Coconut.
• The 8x8 idea (8 categories of 8 cards) transfers as a method and the number 8 does not mean the same thing. Without a mulligan, 8 cards in 60 show up in 65.4% of opening hands and 8 cards in 99 show up in 45.6%. It takes about 14 in 99 to match.
• Their template asks for 38 lands, which leaves about 60 cards with an effect. A Coconut deck has 60 cards with an effect, and 48 of them can also be ink. When someone asks for a 100-card Coconut, this is my answer.

Methodology

The counts and percentages came out of scripts. Each model uses a 60-card deck, a hand of 7, and going first with no draw on turn 1. The mulligan is free for any number of cards.

• Named card and theme tables. 200,000 simulated games with the hard mulligan, which is the best case. The model reproduces Karsten’s published figures for 1v1: about 40% and 67% for a 4-of, 65% and 90% for 8 similar cards.
• Ink table. 200,000 simulated games. The mulligan puts back non-inkable cards. The deck plays one random card per turn and never skips a play, so the differences between rows matter more than the absolute values.
• Table for removal, board-wide cards and lore engines. An exact hypergeometric calculation with no mulligan. The deck with 12 card advantage cards sees 19.5 cards by turn 8, and I rounded that down to 19.
• Card advantage model. The template curve and 12 non-inkable cards. The mix is measured on my 18 lists: 56% of the draw cards are repeatable sources, 36% of the ones that draw when played draw 2 cards, and the rest draw 1. A repeatable source draws 1 card per turn and survives a round of turns 75% of the time. The deck inks every turn through turn 8 and plays every card it can pay for, draw cards first. A draw card never counts as a plan card, even when it is a character. There is one board wipe per game, and it also takes the repeatable draw sources. The deck does not hold cards back to wait for it. The turn 8 columns use 50,000 games per row and the table columns use 30,000.
• List counts. My 18 published lists, 60 cards each, were matched card by card against the full card pool. A script sorted the cards into categories by card text, and I ruled on the borderline cases. My lists count copies per deck. The community lists count distinct cards per deck.

What would change these numbers

• A change to the free mulligan. The named card and theme tables depend on it.
• Cheap draw printed in several inks would move the non-inkable budget up.
• A table where boards stay small. The 3 drops to 2.
• How often tables get cleared. A table that almost never clears boards moves the 12 to 10. A table that always does moves it to 14.
• The card advantage model assumes a repeatable draw source survives a round of turns 75% of the time. At 50%, in the game with no board wipe, no count plays more plan cards than the deck with no draw, and the draw cards only lower the share of games that miss an ink drop.

What did I miss?