Drawing Your Cards

Every deckbuilder eventually asks the same question: if I put this card in my deck, when will I actually see it? These tables answer that exactly, and the answers are consistently more pessimistic than intuition suggests.

All odds assume you are on the play, so by turn T you have seen your opening seven plus T−1 draws. Tutors, cantrips, and other selection are not modelled; they all improve your real odds. Every number is an exact hypergeometric probability computed at build time.

Seeing at least one copy: 60 card decks

CopiesTurn 1Turn 2Turn 3Turn 4Turn 5Turn 6Turn 8Turn 10
111.7%13.3%15.0%16.7%18.3%20.0%23.3%26.7%
222.1%25.1%28.0%30.8%33.6%36.3%41.5%46.6%
331.5%35.4%39.1%42.7%46.2%49.5%55.6%61.3%
439.9%44.5%48.8%52.8%56.6%60.1%66.5%72.2%

Two numbers here are worth internalising. A single copy of a card in a 60 card deck appears in your opening hand about 12 percent of the time, and you have still only seen it 20 percent of the time by turn six. A full four copies is in your opening hand about 40 percent of the time, and has appeared by turn six only about 60 percent.

That second figure is the one that reshapes deckbuilding. Four copies is the legal maximum, and it still means your "always" card is missing from four games in ten by turn six. A plan that depends on drawing a specific four-of is not a plan.

How much redundancy a target actually takes

Inverting the question: to have seen an effect by a given turn with a given reliability, how many copies of it would you need in your deck?

ReliabilityBy turn 1By turn 3By turn 5By turn 7
60%7654
70%9765
80%121087
90%1613119

Every useful number in that table is larger than four, which is the entire lesson. You cannot reach reliability with copies of one card, because the rules cap you at four. You reach it by counting effects rather than cards: four removal spells plus four more that do a similar job is eight sources of "kill their creature," and eight is a number the table says something encouraging about.

This is why experienced deckbuilders talk about "eight-ofs" and "virtual copies." When you hear that a deck runs eight one mana interaction spells, the player is not describing eight identical cards. They are buying the reliability this table prices.

Drawing two copies

For cards that need a partner, or combo pieces where one is not enough, the odds fall off a cliff:

CopiesTurn 1Turn 2Turn 3Turn 4Turn 5Turn 6Turn 8Turn 10
33.4%4.4%5.6%6.9%8.4%9.9%13.3%17.1%
46.3%8.2%10.3%12.6%15.0%17.5%23.0%28.7%
614.0%17.8%21.8%25.9%30.2%34.4%43.0%51.2%
823.2%28.8%34.4%40.0%45.4%50.7%60.5%69.1%

This is why combo decks are built the way they are. Assembling two specific pieces by drawing them naturally is unreliable at any legal copy count, so combo decks spend their remaining slots on tutors and card selection rather than on more copies.

Limited: 40 card decks

The smaller deck helps considerably, which is part of why a single bomb matters so much more in Limited than a single powerful card does in Constructed:

CopiesTurn 1Turn 2Turn 3Turn 4Turn 5Turn 6Turn 8Turn 10
117.5%20.0%22.5%25.0%27.5%30.0%35.0%40.0%
232.3%36.4%40.4%44.2%47.9%51.5%58.3%64.6%
344.8%49.8%54.5%58.9%63.0%66.8%73.7%79.5%
455.2%60.7%65.6%70.0%74.0%77.6%83.6%88.4%

How to use these tables

Related: the mana tables, lands by curve, and how to mulligan. Spotted an error? Write us at contact@bopmtg.net.