One-card and two-card draw probabilities

Nine outs can describe a 19.15% chance or a 34.97% chance. The difference is how many cards you will see. Before using either number, name the unseen cards, the target cards and the event you want to count.

This lesson uses a standard 52-card deck and a fixed target set. Every unseen card is equally likely under the model, and a drawn card stays out of the sample. The result measures target hits. Winning the hand requires more information about opponents, redraws and ties.

Ace: “I used nine targets for both trays.”

Amina: “Keep that count. One tray gets one card; the other gets two draws without replacement.”

Define the unseen cards and targets

Let U be the number of unseen cards and o the number of target cards among them. Both are whole numbers; o can be zero but cannot exceed U. For one card, the target probability is o/U. Multiply that fraction by 100 to express it as a percentage.

On a Hold’em flop, your two hole cards and three board cards reveal five identities, leaving 47 unseen. If four hearts are among those five known cards, nine hearts remain unseen. For the event that the next card is a heart, the probability is 9/47, about 19.15%.

Here 47 is an information count, not a count of the physical cards left in the dealer’s hand. Opponents’ hidden cards and unknown burn cards remain among the identities you have not seen. Under the uniform model, do not subtract guessed cards simply because other players were dealt hands.

An additional exposed card changes the information set. If it is one of your nine target hearts, remove it from both counts: eight targets among 46 unseen cards. If it is a non-heart, you have nine targets among 46. Those observations lead to different next-card probabilities.

Count each physical card once. When a card completes both a straight and a flush, it belongs to the target set once, even though it helps in two ways. A nine-card flush set and an eight-card straight set with two shared cards contain 9 + 8 – 2 = 15 distinct targets.

Information about an opponent’s likely holding can also change the model. This desk does not assign weights to ranges or hidden cards. Its uniform answer is exact for the stated card-count experiment, while the target list and assumptions still need your review.

Follow the first card into the second draw

After a non-target turn in the 47-card example, one miss has disappeared. There are 46 unseen cards and the same nine targets, so the river target chance becomes 9/46, about 19.57%. That is the one-card answer after observing this particular turn.

After a target turn, only eight of the original nine targets remain among 46 cards. The river chance of another original target is 8/46, about 17.39%. The first target has already appeared; it cannot appear again. The two branches differ because they leave different cards unseen.

For at least one target across two draws, count the opposite event first: both cards miss. The first miss has probability (U – o)/U. After that miss, the second miss has probability (U – o – 1)/(U – 1). Multiply these conditional probabilities, then subtract from one.

For U of at least 2, P(at least one target in two cards) = 1 – ((U – o)(U – o – 1))/(U(U – 1)). In the 47-card, nine-target example, that is 1 – (38 x 37)/(47 x 46), about 34.97%.

Using 38/47 for both misses would put the first card back before the second draw. That describes sampling with replacement. In this lesson, the first card stays on the table, so the second denominator is 46 and the second miss count is 37.

You can count the same result as unordered pairs. There are 47 x 46 / 2 = 1,081 possible pairs. The 38 misses make 38 x 37 / 2 = 703 all-miss pairs. The remaining 378 pairs contain at least one target, giving 378/1,081 = 34.97%.

Ace: “I will leave the first card beside the shoe.”

Amina: “After a miss, one fewer card is unseen and all nine targets remain. Use 9 out of 46 for the next draw.”

Separate exactly one from at least one

The 378 successful pairs include two different outcomes. Exactly one target means choosing one of nine targets and one of 38 misses: 9 x 38 = 342 pairs. Exactly two targets means choosing two of the nine: 9 x 8 / 2 = 36 pairs.

For general inputs, exactly one has probability 2o(U – o)/(U(U – 1)). Exactly two has probability o(o – 1)/(U(U – 1)). Adding these gives the same at-least-one answer as subtracting the two-miss probability from one.

The factor of two in exactly one accounts for both orders: target then miss, or miss then target. Either order qualifies. For exactly two, both draws must be targets, so its ordered calculation is 9/47 x 8/46, giving the same 3.33%.

Doubling 9/47 gives 18/47, about 38.30%. It counts target appearances across the two draws, including two appearances whenever both cards are targets. The question ‘Did at least one appear?’ counts each such pair once. Subtract the 3.33% two-target overlap to recover 34.97%.

Say which event you mean when discussing two chances. Seeing at least one heart on turn or river and seeing a heart on both streets are different events. A result labeled exactly one also excludes the two-heart runouts, even though those runouts qualify for at least one.

Check the rule of two and four

Multiplying outs by two gives a rough percentage for one card in familiar Hold’em situations; multiplying by four approximates two cards. With nine fixed targets among 47 unseen cards, the shortcuts give 18% and 36%. The exact answers are 19.15% and 34.97%.

With 15 targets, the two-card shortcut overshoots by about 5.88 percentage points. That difference is large enough to matter when comparing close estimates. Use the exact fraction while studying so you can see where an approximation starts to drift.

The shortcut also assumes a familiar unseen-card count. In a small teaching sample with two targets among ten cards, the next-card probability is 20%, not 4%. The exact method still works because it uses the actual denominator.

Match the probability to the cards you will see

A turn-and-river calculation assumes both cards are included in the experiment. A flop call may be followed by another bet on the turn. It does not automatically purchase two cards at the price of one decision.

Use the one-card result when asking only about the next card. If everyone involved is already all-in and both remaining board cards will be dealt, the two-card horizon can be appropriate. Even then, hitting a target is only the event counted; it need not equal winning at showdown.

To evaluate a decision with future betting, you would need possible turn actions, later prices and what happens after different cards. The practice desk has no pot or call input. Its two-card percentage contains no allowance for those costs.

Check whether the target stays useful

A live target is still available to appear among the relevant unknown cards. A known exposed card cannot be drawn again. A clean out is useful under the opponent and runout assumptions you are making; a dirty out may improve your hand while also producing a stronger hand for an opponent.

For example, a flush-completing card can pair the board and give an opponent a full house. Completing your draw and winning the pot then come apart. Conversely, some misses may still leave you with the best hand. Counting only draw completions does not capture either possibility.

A clean turn improvement may also be overtaken on the river. A model that treats every target as a permanent win ignores those redraws. If opponent cards are known, remove their identities from the unseen count before evaluating complete runouts; do not keep the familiar 47 automatically.

The desk keeps the original target set fixed. It can answer whether at least one of nine specified hearts appears. It cannot follow a straight draw whose useful ranks change after the turn, or decide which hearts remain winning cards against a particular range.

A backdoor flush example asks for two targets. Suppose your two hole cards are hearts and exactly one flop card is a heart. Ten hearts are unseen among 47 cards, and making a five-heart combination requires a heart on both remaining streets. The probability is 10/47 x 9/46 = 45/1,081, about 4.16%.

That 4.16% measures the two-heart event. It does not rule out winning another way or losing after completing the flush. For a changing draw or actual showdown equity, list the branches and evaluate full hands rather than replacing them with a larger fixed out count.

Use the desk and test its limits

Enter 47 unseen cards and 9 fixed targets, then choose Count the outcomes. Read the one-card fraction, followed by the two-card pair counts. Check that 703 + 342 + 36 = 1,081 and that exactly one plus exactly two equals at least one.

Change the unseen count to 46 while leaving nine targets. The next-card answer becomes 19.57%. That reproduces the after-miss river question. The desk also displays a hypothetical two-draw result for those inputs; do not mistake it for a Hold’em river calculation when only one board card remains.

Inputs must be whole cards: unseen count from 1 through 52, targets from zero through the unseen count. Blank means missing information. Fractions, negative counts and scientific notation are rejected. Editing either field clears the previous receipt; Reset restores 47 unseen cards and 9 targets.

Zero targets give zero chance of a hit. If every unseen card is a target, the next card must hit, and any available two-card draw contains exactly two hits. If only one miss remains, missing twice is impossible, so at least one target in two cards is certain.

With one unseen card, the one-card result is valid but a two-card draw is unavailable. There is no second card to draw. The tool reports that limit instead of dividing by zero or treating a repeated card as a new draw.

Ace: “The target appeared, so I marked the hand as won.”

Amina: “Record a hit first. We still need the opponent’s hand and the final board to decide the winner.”

Check a final-card price in a limited model

To connect probability with a price, define a deliberately narrow toy: one final card, no later betting, no fees, no ties, every target wins and every miss loses. Only under those assumptions can the target probability stand in for the win probability.

Let P be the pot already available before your call, including the opponent’s bet, and C the call amount. A win gains P net from this decision; a loss costs C. Expected value is pP – (1 – p)C. Setting it to zero gives a break-even probability of C/(P + C).

For P = 40 and C = 10, the price is 4:1 and the break-even probability is 10/50 = 20%. Working backward, a 20% win probability has 80% misses, or 4:1 odds against. Both forms describe the same toy boundary.

With nine winning targets among 46 cards, p = 9/46 and the odds against are 37:9, about 4.11:1. At that 40-to-10 price, the toy’s expected value is (9 x 40 – 37 x 10)/46, about -0.22 chips per call. This follows from the stipulated win-or-lose model, not from nine outs alone.

For a real hand, first check whether those win, loss and cost assumptions hold. The Info & Tools lessons provide more worked exercises. Keep your review note specific: “Nine fixed targets among 47 unseen cards, two draws without replacement, at least one target.” That is the question the 34.97% answer actually resolves.

↑ Back to top