Deckbuilders and Probability: Drawing the Perfect Hand

You flip your opening hand. Heart quick. Do you keep? Do you mulligan? This moment is not luck alone. It is math you can steer.

The good news: you do not need deep math to get real gains. You need three numbers and one rule. We will use clear steps, small words, and live cases from real games. You will see how to plan a “perfect” hand on purpose.

What are we really counting here?

In deckbuilders, you pull cards without putting them back until the deck runs out. This is “no replacement.” That one fact points to one key tool: the hypergeometric distribution. It tells us the chance that you see at least one copy of a target card in your hand size, from a deck of a set size, with a set number of copies. For a formal course view, see the MIT OpenCourseWare probability course: MIT OpenCourseWare probability course.

If you want a simple, friendly intro to the idea, this short lesson helps: hypergeometric distribution explained.

The three questions that decide your odds

Ask this any time you plan a deck:

• How many copies of the key card are in my deck? (k)
• How big is my deck? (N)
• How many cards do I draw in the opener? (n)

Then the core idea is this: Chance to see at least one copy = 1 − C(N−k, n) / C(N, n). You can try your own numbers with this tool: hypergeometric calculator. For a strict definition and symbols, see formal definition of the hypergeometric distribution.

Quick Odds to “See It in the Opener”

Use this table as a cheat sheet. It shows the chance to see at least one copy of a target card in your first hand. We assume draws without replacement, no scry/tutor effects, and no mulligan smoothing.

Note: Mulligans, scry, tutors, cantrips, and other effects change these odds. We cover that below.

At least two copies example (N=30, k=4, n=5): P(≥2) = 1 − P(0) − P(1) = 1 − C(26,5)/C(30,5) − [C(4,1)·C(26,4)/C(30,5)] ≈ 11.9%.

Case 1: Dominion — hit the Chapel, win the race

Dominion has a famous card: Chapel. If you see it early, you trash junk, thin fast, and snowball. What are the odds? In a 10-card start, with a 5-card opening hand, one Chapel in the deck gives 50% to see it in your first two hands. Two copies push you much higher. See the card details here: Dominion Chapel.

Players have tested this for years. A good write-up on the logic of odds for board gamers lives on BGG: probability primer for board gamers. The short lesson: add more “outs” or shrink the deck. Both raise your chance to draw the key tool on time.

Case 2: Slay the Spire — cycle, scry, and dead hands

In Slay the Spire you draw a small hand each turn and then cycle. Scry lets you look ahead and skip weak cards. That is like soft control over the top of your deck. It changes the draw pool and lifts the chance to hit a power turn. A neat design talk is here: GDC talk on building a deckbuilder. If you want the scry rules, see Scry mechanic in Slay the Spire.

Tip you can use now: remove strikes and defends when safe to do so. Every low card you cut raises the share of good cards. Your odds to open with a key power go up the next fight.

Case 3: MTG and the London Mulligan — a big shift

Magic’s London Mulligan lets you draw a full hand and then put cards on the bottom. That smooths your opener a lot. It lifts the chance to see a one-lander or a key piece by quite a bit. Read the rule here: London Mulligan rule.

Want data on real opening hands? The 17Lands team posts stats from draft games. It is not the same as all of Magic, but it shows strong signals on hand quality and win rate: 17Lands data on opening hands.

Design levers that move your odds

How do you raise your chance to draw “the card” on turn one?

• Redundancy: add more copies or cards that act like it. More outs = higher P(≥1).
• Deck-thinning: remove weak cards. Smaller N with same k = higher odds.
• Cantrips: draw a card after play. They speed the cycle and reduce dead turns.
• Tutors: fetch what you need. They turn one draw into a search.
• Top control: scry, surveil, scrubs the top and keeps good cards.

Small experiment: test the math with a sim

You can verify the table with a tiny Monte Carlo sim. Run this idea in Python or any language. Change N, k, n and the number of trials. Compare the hit rate to the formula 1 − C(N−k, n) / C(N, n). For a gentle note set on counting and combos, see Stanford notes on counting.

In practice, your result will sit very close to the table (within a few tenths of a percent) once you run enough trials.

Rules of thumb you can use today

• One copy is weak. In a 30-card deck with a 5-card opener, one copy is ~17% to show. Add copies.
• Three copies in 30 with a 5-card opener is ~43%. Four copies break 50%.
• In a 60-card deck with a 7-card opener, four outs is ~40%. Eight outs is ~65%.
• Thin if you can. Every bad card you cut makes each draw a bit better.
• Mulligans change the math. With a good mulligan rule (like London), you can keep more two-land or key-piece hands.

Watch out for thinking traps

Do not fall for the gambler’s fallacy: past draws do not force future draws. The deck does not “owe you one.” A clear note on this is here: gambler’s fallacy explained.

Over-adding copies can also hurt. Redundant dead draws can clog late turns. Balance the plan: early hits vs. late bricks.

Case notes, quick math

Dominion Chapel start (two turns to see it): with one Chapel in 10, you have about a coin flip to see it in your first two 5-card hands, since you see all 10 cards unless you gain/shuffle mid-turn. Two Chapels make it very likely one shows by turn two. That drives an early trash plan.

Slay the Spire scry: if you scry 3, you look at three cards and bottom the duds. This is like removing bad paths before you draw. Over a fight, that small edge stacks up.

MTG London mull: drawing seven then putting cards back lets you shape hands with two lands plus a one-drop more often. If your deck has eight one-drops and 22 lands, your chance to find a smooth curve hand rises a lot compared to old rules. For data on how small changes hit win rate, scan the posts on 17Lands and compare trends across sets.

Try it yourself in 90 seconds

1. Pick your deck size, hand size, and target copies.
2. Look up the row in the table above.
3. Then change just one knob. Add one copy, or cut two weak cards, or add a cantrip.
4. Check the new odds. Keep the change if it helps the early plan without harming the late game.

Deepen your base

If you want to keep it light and interactive, this short page is a nice way to review: combinations and permutations refresher. If you like a full course, the MIT OCW probability course is free and has strong sets of notes and problems.

A few closing lines

The “perfect hand” is not a magic trick. It is a plan on paper before you shuffle. More outs, fewer blanks, and a good mulligan rule make the math work for you. Use the table. Test with a sim. Tune your list. Then enjoy more keeps and fewer sad redraws.

Transparency and method

• All odds use the hypergeometric model: draws without replacement.
• Formula: P(≥1) = 1 − C(N−k, n) / C(N, n). Example for P(≥2) shown above.
• Numbers were spot-checked with the StatTrek hypergeometric calculator.
• Game rules like mulligans, scry, hand smoothing, or tutors change the base. Check your game’s rules first.

Author note

About the author: I build and test deck math for card games and roguelikes. I log many hours in Dominion, Slay the Spire, and MTG. I keep a small lab of scripts to check odds and edge cases. This page is kept up to date as rules and tools change.

Last updated: 2026-08-18