Most advice about making a card game assumes you're designing one deck that everyone shares. A trading card game doesn't work that way. You design a pool, and players cut their own decks out of it, so you're designing the space a few hundred different games get played in.
Which is why the order below is rules, then structure, then cards. Design a card first and six weeks later you have 80 of them and no idea whether they make a game.
This guide covers the parts specific to TCGs: the rules skeleton, resources, card types, factions, deckbuilding restrictions, set size, rarity, and the spreadsheet that holds it together. Prototyping, artwork, balance and printing get a short linked section at the end. If you've never built a physical prototype, read how to make your own card game first.
Start With the Core Loop
Write the turn on one page before you design anything. Four questions:
- What happens in a turn, in order? Draw, resource, main, combat, end is a common shape, and fewer phases is usually better.
- How does someone win? Damage, objectives, or decking an opponent. Pick one; alternate win conditions are a later problem.
- How big is a hand, and when do you draw? This sets the pace more than any card will.
- What happens when two things clash? Combat maths, priority, resolution order. Miserable to retrofit.
Test it by playing a turn out loud with imaginary cards. If you invent a rule halfway through, you're not done yet.
Choose Your Resource System
This decision has the longest shadow. It sets your game's variance, how many deck slots go to things that aren't your interesting cards, and how fast it escalates.
Dedicated resource cards. Players include them in the deck and play one per turn. The most familiar model and the least forgiving, because every resource card is a dead slot and players routinely draw too many or too few. That variance works in a large game where you can build for consistency. In a 150-card pool you can't.
Any card can be a resource. Players put a card face down each turn as generic fuel. Nobody floods or gets stuck, and each turn asks whether a card is worth more played or buried. For a first set, this is what I'd recommend.
Automatic growth. Everyone gains one resource per turn. No variance, no deck slots, easy to teach, but every player's curve is identical and an axis of deckbuilding disappears.
Exhaust what you control. Your cards are the resource, so playing things makes you stronger. It snowballs, and whoever falls behind stays behind.
No resources at all. Hand size and draw are the only constraint. Fast to learn, and all the pressure lands on your costs being right.
Change this later and you'll rewrite every cost on every card.
Decide Your Card Types
Three types is enough for a first set:
- Units. They stay on the table, have stats, do the fighting.
- Actions. One effect, then gone.
- Persistent non-units. Equipment, locations and ongoing effects that stick around but don't fight.
Add a resource type only if your system needs one. Every type costs you rules text, a layout variant, and a slot in every deck, so before adding a fourth, try deleting it. A subtype like Beast or Vehicle is usually cheaper and carries most of the same flavour.
Types do need to be recognisable at a glance, which is a layout job. Our card graphic design rules cover keeping them distinct while the layout stays consistent.
Build Factions That Are Bad at Something
Three to five factions is the sweet spot. Fewer and decks blur together, more and your set doubles in size.
Each one needs a colour, a mechanical bias, and a hole. The hole is the part people skip. If every faction can remove any threat, refill its hand and field big units, your factions are paint rather than design.
Build a grid to check. Capabilities down the side (removal, card draw, big units, speed, recursion, defence), factions across the top, each cell marked "good at", "bad at" or "can't". Every row needs one "can't" and every faction two weaknesses. If you can't bring yourself to write "can't" anywhere, that's why your factions feel samey.
Keep neutral cards lean, ten to twenty. They appear in every deck by definition, so a large neutral pool flattens the differences you just built.
Write the Deckbuilding Restrictions Early
These are rules, and they're also the calculator for the next section.
- Minimum deck size. 40 or 60. A 40-card minimum makes a small pool viable and lets players see their best cards more often. Start at 40.
- Maximum copies. Three or four. Four is familiar and makes decks consistent. Three pushes variety and means fewer copies to print.
- Faction limits. Mono-faction is strongest for a first set, because identities stay legible and the best cards can't gather in one deck.
- Singleton. One copy of anything, ever. A lovely format that demands a much larger pool.
- Sideboards. Skip them for now.
Work Out Your Set Size From the Rules
With a 40-card minimum and three copies maximum, a player needs 14 distinct cards to fill a deck legally. That's the floor, and a deck built at the floor is closer to an inventory, because no choices went into it.
For a faction to feel like a choice you want two to three times that inside it, so 30 to 40 playable cards each. Four factions at 35 is 140, plus 15 or so neutral, landing around 155.
That's why most first sets come in between 120 and 180 unique cards. Under 100, two decks in the same faction end up nearly identical and players notice within an evening. Over 200, you're balancing cards nobody has played. Design more than you print, though: writing 180 and shipping 150 is normal, and the 30 you cut are the ones playtesting told you were boring.
Rarity Is Not a Power Dial
Rarity does two jobs, and neither is "this card is stronger".
The first is print economics, since scarcity is what makes a booster worth opening. The second is complexity pacing, and it's the one that improves your game. Commons are your grammar, clean repeatable cards players learn from. Rares are where the weird, build-around cards live, because a card that warps deckbuilding is a delight occasionally and exhausting in every hand.
If your rares are simply better commons, you've sold power instead of designing it. Our balancing guide covers why "rare shouldn't mean unbeatable".
A workable distribution, by share of unique cards:
| Rarity | Share of set | In a 120-card set | In a 150-card set |
|---|---|---|---|
| Common | 60% | 72 | 90 |
| Uncommon | 25% | 30 | 37 |
| Rare | 12% | 14 | 18 |
| Mythic | 3% | 4 | 5 |
Three tiers is fine, and mythic is worth dropping under about 120 cards, where the top tier is only three or four cards.
The Percentage That Trips People Up
Two numbers get conflated here constantly. Share of the set is how many cards exist at each rarity. Share of what players open is how often each turns up in a pack. In a 12-card booster of 7 commons, 3 uncommons, 1 rare-or-mythic and 1 token, rares are 12% of your set but 1 of 12 slots, about 8% of what a player opens.
How many rares exist affects variety. How often they appear affects how long it takes to build the deck someone wants. So keep decks buildable from commons and uncommons and let rares be upgrades, because if a rare is mandatory then a lot of players never play that deck.
If you're doing print-and-play or selling fixed decks, skip rarity completely. It's a distribution mechanism before it's a design one.
Put the Whole Set in a Spreadsheet
One row per card, in one sheet. A folder of 150 image files isn't a set, as you'll discover the first time you need to ask a question about the whole thing, because every such question turns out to be a sorting question. Cost curve lumpy? Sort by cost. Enough two-drops in Faction C? Filter and count. Which cards are untested? Filter by status.
The Columns That Earn Their Place
| Column | Type | Why it's there |
|---|---|---|
| Name | Text | Unique, and what you'll search by |
| Collector number | Text | A stable ID, like EMB-042, that survives renaming |
| Type | Select | Drives which layout the card uses |
| Faction | Select | Colour-coded so the grid is readable at a glance |
| Cost | Number | Sort by this constantly |
| Attack / Health | Number | Units only, leave blank elsewhere |
| Rules text | Text | The printed effect, and nothing that belongs in a number column |
| Keywords | List | Reusable ability words, kept consistent |
| Flavour text | Text | Optional, and the first thing cut when space runs out |
| Rarity | Select | Complexity tier, not a power tier |
| Art brief | Text | What the illustration needs to show |
| Art | Image | The final illustration |
| Status | Select | Draft, Testing, Locked |
One rule above all others: if a number matters for balance, it gets its own column. The moment a cost or a stat lives only inside a sentence of rules text, you can't sort, filter or check it, and that's when balance problems start hiding.
(In ShuffleKit, a browser-based card game designer, importing the sheet turns every column into a typed property that layouts read directly, so changing a cost in the sheet changes the card. CSV to card game walks through it.)
Starter Decks and Testing Archetypes
You don't need a finished set to play. With a 40-card minimum and three copies maximum, two starter decks need roughly 28 distinct cards between them, so pick two factions, write about 30 cards, and play. What you learn changes the other 120.
Aim each deck at a different archetype, so you're testing the shape of the game rather than card values:
- Aggro. Cheap, fast, wants it over by turn six.
- Midrange. Efficient cards, wins on quality.
- Control. Answers everything, wins late.
- Combo. Assembles a specific interaction.
Every faction should support one of these clearly and one badly. A faction that plays all four equally well doesn't have an identity yet.
Then play every pairing, not just the fun ones, because the matchup you're avoiding is usually the broken one. If a single deck beats everything, look at what it does that nothing else answers rather than at the card that killed you.
Keep starter decks as saved lists with quantities, so "3x Ember Scout" is one entry you change in one place (decks in ShuffleKit work this way, and layouts can vary by deck membership). For getting games in, our playtesting comparison covers the options honestly, free ones included. Our own browser table is two players, enough for the testing above but not a four player game.
The Rest of the Pipeline, Briefly
These matter, they're just not TCG-specific:
- Prototyping. Print, sleeve, play, repeat. See how to make your own card game.
- Card layout. Readability at arm's length, hierarchy, icons. See game card graphic design and the template editor.
- Balance. Cost curves, answers, win rates. See balancing a trading card game.
- Printing. Bleed, safe zones, stock, files a manufacturer accepts. See print and ship and card game printing.
For all of it in one narrative, the card game design guide is the pillar version.
Your First Set Checklist
- The turn is written down and playable out loud
- A resource system chosen, with its variance understood
- Three to five factions, each with two documented weaknesses
- Deck minimum, copy limit and faction rules set before card one
- Set size calculated from those rules, likely 120 to 180
- Every card in one sheet, balance numbers in their own columns
- Two starter decks played against each other, and every matchup after that
Getting the First Set Out of Your Head
The gap between a folder of card ideas and a set someone can play is mostly structure, and structure is unglamorous work: numbering, sorting, keeping 150 cards consistent when one rule changes. None of it is the fun part, and all of it decides whether the game exists.
So do the boring parts early. Write the turn, pick the restrictions, and let the set size fall out of the arithmetic. Cards come easier once the frame around them holds still, and two rough starter decks will teach you more in one evening than another month of designing alone.
If the slow part turns out to be keeping the sheet and the cards in sync, that's the job ShuffleKit does. It's free to start, and I'd love to hear how your first set goes. Build your first set.
