TL;DR
Run the maximum legal number of copies when a card is central to your plan, useful early, and still good in multiples. Start with two or three when you want to draw the card eventually but dislike duplicates. Use one copy when the card is searchable, replaceable, narrow, expensive to deploy, or mainly intended for a specific matchup.
The practical rule is simple: count backward from when the card must appear. If your deck needs it in the opening turns, maximize access through literal copies, interchangeable cards, draw, or search effects. If it only matters late, you can run fewer copies because you will see more of your deck before needing it.
Always check the rules for your game and format first. A legal maximum is a boundary, not a recommended ratio. The correct number is the smallest count that produces the access you need without making duplicate draws actively worse.
A practical starting point for one, two, three, or four copies
| Copies | Use this count when | Main risk |
|---|---|---|
| 4 | The card is essential, needed early, strong in multiples, or one of the best cards in the deck | Drawing too many if the card becomes redundant |
| 3 | You want it in most games, but the first copy matters more than later copies | Missing it more often in opening hands |
| 2 | It is useful later, situational, searchable, or awkward in multiples | The card may not appear in shorter games |
| 1 | It is a narrow answer, late-game finisher, search target, restricted card, or redundant fifth copy of an effect | You cannot rely on drawing it naturally |
These numbers describe roles rather than rigid laws. A four-copy card can still fail to appear in a given game, while a one-copy card may be functionally accessible through several search effects. Judge the complete access package, not just the number printed beside one deck-list entry.
Four copies: cards your plan needs consistently
Use four copies when the deck is substantially worse without the card and you want to see it as early as possible. Typical candidates include efficient starters, low-cost interaction, resource enablers, and engine pieces that remain useful when you draw a second copy.
The phrase “four copies” is shorthand for the maximum ordinary count in games that use that limit. The actual rule depends on the game. For example, the official Pokémon TCG rulebook specifies a 60-card deck and generally permits up to four cards with the same name, apart from Basic Energy. It also explains that running four copies of a desired Pokémon makes drawing it more likely. The official Pokémon TCG rulebook provides the complete construction rules.
Digimon uses a 50-card main deck and normally permits up to four cards with the same card number, although a card-specific rule can change that count. One Piece also uses a 50-card deck, alongside a Leader and separate DON!! deck, and normally limits cards with the same card number to four. Digimon’s Comprehensive Rules and the One Piece Card Game Comprehensive Rules show why legality must be checked game by game.
Do not automatically maximize every powerful card. Four copies are justified by access and usability, not merely by card quality. A costly finisher can be excellent when cast and still be a poor four-of if the first copy sits unusable in your hand for several turns.
Three copies: important, but not ideal in multiples
Three copies fit cards you want to see regularly but do not need in every opening hand. This count is especially useful when the first copy enables your plan and the second has sharply lower value.
Suppose an ongoing effect does not stack, or a unique permanent makes later copies temporarily unplayable. Four copies improve the chance of finding the first, but they also increase the chance that one of your later draws does nothing useful. Moving from four to three accepts a consistency loss in exchange for fewer redundant hands.
Three is also a reasonable testing count when four feels crowded but two causes the card to disappear from too many games. Do not leave it at three because the number feels balanced. Track whether losses come from missing the card or drawing extra copies, then move the ratio in the direction indicated by the games.
Two copies: cards you want eventually
Two copies suit cards that matter after the opening turns, answer only some threats, or compete with several similar options. Because you have more opportunities to draw before a late-game deadline, you do not need the same opening-hand density as you would for a turn-one play.
This is also a useful count for expensive cards. The deck can still find one during a longer game without repeatedly opening hands containing cards it cannot yet use. If the card is part of a searchable package, two copies may provide both access and some protection against drawing or losing the only target.
Be careful with two-copy combo pieces. If the deck cannot function until it finds one, two may create the appearance of a consistent strategy without the numbers to support it. Search cards, interchangeable pieces, and heavy card selection can improve practical access, but each must be evaluated according to its restrictions and cost.
One copy: a specialist, not a plan you hope to draw naturally
A singleton is appropriate when it serves a narrow purpose, is poor in multiples, can be searched for reliably, or acts as an additional functional copy of another card. One copy can also be correct when deck-construction rules restrict the card or the format itself requires singleton construction.
The crucial distinction is between having access to a one-of and hoping to draw it. In a 60-card deck, one natural copy is unlikely to show up on a strict early deadline. If your strategy depends on it, you need more literal copies, legal substitutes, or reliable ways to find it.
One-of answers become more attractive when the deck can search by a relevant property, such as card type, cost, or another defined characteristic. They are less convincing when the search card is slow, conditional, or needed for a different part of the deck. A search effect only counts as access when you can realistically cast it and retrieve the target before the target matters.
What the draw probabilities actually look like
Drawing from a shuffled deck without replacement can be modeled with a hypergeometric distribution. For a deck containing N cards, K desired copies, and n cards seen, the chance of finding at least one is 1 − C(N−K,n) / C(N,n). The model changes when you add mulligans, extra draws, search effects, or other deck manipulation. OpenStax’s explanation of the hypergeometric distribution covers the underlying probability model.
| Copies in a 60-card deck | At least one in an opening seven | At least one after 11 cards seen |
|---|---|---|
| 1 | 11.7% | 18.3% |
| 2 | 22.1% | 33.6% |
| 3 | 31.5% | 46.2% |
| 4 | 40.0% | 56.6% |
These figures assume a shuffled 60-card deck with no mulligan, search, selection, or other deck change. They illustrate why “I run four” does not mean “I always have it.” Even four copies appear in only about 40% of seven-card opening hands under those assumptions. After 11 cards have been seen, four copies raise the chance to about 56.6%.
Deck size and opening-hand size also matter. In a 40-card deck with a five-card opening hand, the chance of seeing at least one copy is approximately 12.5% with one copy, 23.7% with two, 33.8% with three, and 42.7% with four. Smaller decks are generally more concentrated, but a smaller opening hand can offset part of that advantage.
Most importantly, drawing a card is not the same as using it on time. You may find an early play without having the right resource, supporting piece, or board state. Copy-count analysis should therefore be paired with resource and curve analysis. For a Magic-specific example, the process for choosing how many lands to put in a 60-card deck applies the same deadline-based thinking to mana access.
Count functional copies, but do not pretend they are identical
A deck may contain four copies of its preferred card plus several cards that perform a similar job. These are functional copies: different cards that satisfy the same immediate need. Eight low-cost attackers may give an aggressive deck eight acceptable opening plays even though no individual card appears more than four times.
Functional copies are strongest when the differences rarely matter. If two cards have different costs, targeting limits, colors, or timing restrictions, they should not be counted as perfect substitutes. Start by defining the job precisely. “Interaction” is too broad; “an answer I can use on turn two against an opposing unit” is a testable role.
Search, card draw, selection, recursion, and modal cards can also improve access. Their value depends on timing. A search card that costs three resources cannot make a required turn-one play consistent. A card-selection effect that sees several cards may improve access, but only if the deck can afford to spend time and resources using it.
Choose the count by answering five questions
- When must I see this card? Name a turn or game stage rather than saying “early.”
- How bad is it if I never draw the card? Essential engine pieces deserve more access than optional upgrades.
- How good is the second copy? If duplicates remain useful, the cost of maximizing the card is lower.
- How many functional copies or search effects already reach this role? Include only options that work before the deadline.
- What gets cut for another copy? Every added card displaces interaction, resources, another threat, or a different matchup tool.
These questions turn copy counts into tradeoffs. If a card must appear by turn two and duplicate copies are fine, begin near the legal maximum. If it matters only in long games and the second copy is nearly blank, begin with one or two. Then test the assumption rather than defending the initial number.
Diagnose the ratio with controlled testing
Change one ratio at a time when possible. If you replace six cards after a few losses, you will not know which change affected the deck. Start by identifying a repeated failure: missing an enabler, flooding on expensive cards, drawing narrow answers in the wrong matchup, or finding the right card one turn too late.
- Write down the card’s job and the turn by which it should be available.
- Record whether each loss involved missing it, drawing it too late, or drawing an unwanted duplicate.
- Increase the count when absence is the repeated failure and the displaced card is less important.
- Decrease the count when duplicates repeatedly clog hands or the card lacks valid targets.
- Retest against the same kinds of matchups before making another change.
Opening-hand samples are useful, but full games reveal costs that sample hands miss. A four-of can look excellent in opening tests and still create weak late draws. Conversely, a two-of may appear rarely in opening hands yet show up often enough by the turn when it becomes useful.
For an expensive deck, test the ratio before buying additional copies. A structured proxy-deck testing process can help separate cards that are essential from cards that merely look essential on paper.
The right count is tied to a deadline
If you are unsure how many copies to run, start with this recommendation: four for essential early cards that tolerate duplicates, three for important cards with diminishing returns, two for situational or later cards, and one for searchable specialists or narrow answers. Then adjust for the rules and deck size of your game.
Do not ask only, “How good is this card?” Ask, “How often must this deck access this effect, by what turn, and what happens when I draw another copy?” That is the difference between filling a deck with strong cards and building a deck that performs its plan consistently.