Filiq's blog

Poker - probabilities

· 13 min read

Poker

In the world of poker, one of the key skills is knowing the probabilities. Players who have a solid foundation in mathematical probabilities have an advantage that can move them up the hierarchy of successful players. Knowing the probabilities allows them to better estimate their chances of improving their hand and making a stronger hand or, conversely, the risks of continuing to play. This allows them to play more strategically and more wisely, which is crucial for long-term success at the poker table. Regardless of a player's experience and level, knowing probabilities is a valuable tool for anyone looking to improve their poker skills.

In this article, we will look at the probabilities in the most popular variant of poker, Texas holdem. We will calculate how often a player ends up with each hand at the end of the game. For decisions during the game, players mainly use so-called outs (cards that improve their hand), but that is a topic for another article.

  1. Probability
    1. Royal flush
    2. Straight flush
    3. Four of a kind
    4. Full house
    5. Flush
    6. Straight
    7. Three of a kind
    8. Two pair
    9. Pair
    10. No pair / high card
  2. Table

Probability

The first thing to consider is how many possible combinations of cards there are for a player in poker. In total, we are playing with a deck of 52 cards (4 × 13), and a player always has 7 cards available (2 own + 5 on the table).

That makes 133,784,560 combinations.

(527)=52!7!(52−7)!=133,784,560\begin{aligned} \binom{52}{7} = \frac{52!}{7!(52-7)!} = 133{,}784{,}560 \end{aligned}

All probabilities in this article give the probability that the best five-card hand out of all 7 cards is exactly the given hand (assuming the player plays until the end and sees all 5 cards on the table). A pair therefore means exactly a pair, not "at least a pair", which is why all the probabilities add up to exactly 100%.

Royal flush

A royal flush is the rarest and strongest hand in poker. At the same time, calculating the probability of getting it is relatively simple.

A royal flush can be made in only 4 ways:

  • 10♠-J♠-Q♠-K♠-A♠
  • 10♣-J♣-Q♣-K♣-A♣
  • 10♥-J♥-Q♥-K♥-A♥
  • 10♦-J♦-Q♦-K♦-A♦

If we played with only 5 cards in total, the calculation would be easier, because we would know that there are only 4 possible combinations out of all card combinations.

However, we are playing with a total of 7 cards, so we must also include the remaining 2 cards in our calculation.

4⋅(472)=4⋅1,081=4,324\begin{aligned} 4 \cdot \binom{47}{2} = 4 \cdot 1{,}081 = 4{,}324 \end{aligned}

The binomial coefficient 47 choose 2, because 47 (52 - 5, the remaining cards in the deck) and 2 (the last 2 cards that can be placed on the table).

Out of all 133,784,560 combinations, a royal flush occurs in only 4,324.

4,324:133,784,560≈1:30,940≈0.00323%\begin{aligned} 4{,}324 : 133{,}784{,}560 ≈ 1 : 30{,}940 ≈ 0.00323\% \end{aligned}

The probability of a royal flush is approximately 0.00323%.

Straight flush

Unlike the royal flush, there are many more ways here, because we are not limited to the 10 to A sequence. While this increases the probability, it also makes our calculation more difficult.

How many ways are there?

  • A-2-3-4-5
  • 2-3-4-5-6
  • 3-4-5-6-7
  • 4-5-6-7-8
  • 5-6-7-8-9
  • 6-7-8-9-10
  • 7-8-9-10-J
  • 8-9-10-J-Q
  • 9-10-J-Q-K
  • 10-J-Q-K-A

If we don't count the royal flush, we have 36 (9 · 4) ways.

36⋅(462)=36⋅1,035=37,260\begin{aligned} 36 \cdot \binom{46}{2} = 36 \cdot 1{,}035 = 37{,}260 \end{aligned}

This time the binomial coefficient uses 46 cards instead of 47. The remaining 2 cards must not include the card one rank higher in the same suit, otherwise they would make a higher straight flush and we would count it twice. For example, with the straight flush 5♠-6♠-7♠-8♠-9♠ the 10♠ must not come, with 9♠ to K♠ the A♠ must not come (that would be a royal flush).

Out of all 133,784,560 combinations, a straight flush occurs in only 37,260 (41,584 including the royal flush).

37,260:133,784,560≈1:3,590.57≈0.02785%\begin{aligned} 37{,}260 : 133{,}784{,}560 ≈ 1 : 3{,}590.57 ≈ 0.02785\% \end{aligned}

The probability of a straight flush is approximately 0.02785% (including the royal flush approximately 0.03108%).

Four of a kind

This time we pick four cards of one of the 13 ranks, which gives us 13 ways. The remaining 3 cards can be any of the other 48 cards.

13⋅(483)=13⋅17,296=224,848\begin{aligned} 13 \cdot \binom{48}{3} = 13 \cdot 17{,}296 = 224{,}848 \end{aligned}

Out of all 133,784,560 combinations, four of a kind occurs in only 224,848.

224,848:133,784,560≈1:595≈0.16807%\begin{aligned} 224{,}848 : 133{,}784{,}560 ≈ 1 : 595 ≈ 0.16807\% \end{aligned}

The probability of four of a kind is approximately 0.16807%.

Full house

A full house is a combination of 1 three of a kind and 1 pair. In terms of combinatorics, it can be obtained in 3 ways. We will calculate all 3 and add them up at the end.

First way

The first way is 2 three of a kinds and 1 remaining card.

(132)⋅(43)2⋅(441)=54,912\begin{aligned} \binom{13}{2} \cdot \binom{4}{3}^2 \cdot \binom{44}{1} = 54{,}912 \end{aligned}

We pick the two ranks of the three of a kinds out of 13, each three of a kind can be made in 4 suit combinations, and the last card can be any of the 44 cards of the remaining 11 ranks.

Second way

The second way is 1 three of a kind and 2 pairs.

(131)⋅(43)⋅(122)⋅(42)2=123,552\begin{aligned} \binom{13}{1} \cdot \binom{4}{3} \cdot \binom{12}{2} \cdot \binom{4}{2}^2 = 123{,}552 \end{aligned}

The three of a kind can be made of 13 different ranks (A, 2, 3, ..., K), and each of them can be made in 4 suit combinations (♠-♣-♥, ♠-♣-♦, ♠-♥-♦, ♣-♥-♦). The pairs can be made of the remaining 12 ranks, each pair in 6 suit combinations (squared, since we are counting 2 pairs).

Third way

The third and most likely way is 1 three of a kind, 1 pair, and 2 remaining unpaired cards.

(131)⋅(43)⋅(121)⋅(42)⋅(112)⋅(41)2=3,294,720\begin{aligned} \binom{13}{1} \cdot \binom{4}{3} \cdot \binom{12}{1} \cdot \binom{4}{2} \cdot \binom{11}{2} \cdot \binom{4}{1}^2 = 3{,}294{,}720 \end{aligned}

The three of a kind can be made of 13 different ranks, in 4 suit combinations. The pair can be made of the remaining 12 ranks, in 6 suit combinations. The last 2 unpaired cards can be made of 11 ranks, each in 4 suits (squared, since we are counting 2 cards).

Final probability

Let's add up all the ways mentioned above.

54,912+123,552+3,294,720=3,473,184\begin{aligned} 54{,}912 + 123{,}552 + 3{,}294{,}720 = 3{,}473{,}184 \end{aligned}

Out of all 133,784,560 combinations, a full house occurs in only 3,473,184.

3,473,184:133,784,560≈1:38.52≈2.59610%\begin{aligned} 3{,}473{,}184 : 133{,}784{,}560 ≈ 1 : 38.52 ≈ 2.59610\% \end{aligned}

The probability of a full house is approximately 2.59610%.

Flush

We will also calculate the combinations for a flush in 3 ways. Each calculation has one thing in common, and that is the number 4 (suits). In the calculations, we will subtract straight flushes and royal flushes.

The subtracted numbers 217, 71, and 10 are the numbers of ways to pick 7, 6, or 5 different ranks out of 13 so that they contain a straight. If all these cards were of the same suit, it would not be a flush, but a straight flush.

First way

The first case is a hand where all 7 cards are of the same suit.

4⋅[(137)−217]=5,996\begin{aligned} 4 \cdot \Bigg[ \binom{13}{7} - 217 \Bigg] = 5{,}996 \end{aligned}

Second way

The second way is 6 cards of the same suit and 1 card of a different suit.

4⋅[(136)−71]⋅39=256,620\begin{aligned} 4 \cdot \Bigg[ \binom{13}{6} - 71 \Bigg] \cdot 39 = 256{,}620 \end{aligned}

The last card can be any of the 39 cards of the other suits.

Third way

The third way is 5 cards of the same suit and 2 cards of a different suit.

4⋅[(135)−10]⋅(392)=3,785,028\begin{aligned} 4 \cdot \Bigg[ \binom{13}{5} - 10 \Bigg] \cdot \binom{39}{2} = 3{,}785{,}028 \end{aligned}

Final probability

Let's add up all the ways mentioned above.

5,996+256,620+3,785,028=4,047,644\begin{aligned} 5{,}996 + 256{,}620 + 3{,}785{,}028 = 4{,}047{,}644 \end{aligned}

Out of all 133,784,560 combinations, a flush occurs in only 4,047,644.

4,047,644:133,784,560≈1:33.05≈3.02549%\begin{aligned} 4{,}047{,}644 : 133{,}784{,}560 ≈ 1 : 33.05 ≈ 3.02549\% \end{aligned}

The probability of a flush is approximately 3.02549% (including straight flushes and royal flushes approximately 3.05658%).

Straight

Again, we split the calculation into 3 ways, by the number of different ranks. In the calculations, we will subtract all combinations that make a flush (and therefore also straight flushes and royal flushes), because a flush beats a straight.

First way

7 different ranks (for example, 4-5-6-7-8-J-K).

217⋅(47−756−4−84)=3,372,180\begin{aligned} 217 \cdot (4^7 - 756 - 4 - 84) = 3{,}372{,}180 \end{aligned}

217 is the number of ways to pick 7 different ranks that contain a straight. Each of the 7 cards can have any of the 4 suits (474^7), but we subtract the suit distributions that make a flush: 756 distributions with exactly 5 cards of the same suit (4⋅(75)⋅324 \cdot \binom{7}{5} \cdot 3^2), 4 distributions with all 7 cards of the same suit, and 84 distributions with exactly 6 cards of the same suit (4⋅7⋅34 \cdot 7 \cdot 3).

Second way

6 different ranks (for example, 4-5-6-7-8-8-K).

71⋅36⋅990=2,530,440\begin{aligned} 71 \cdot 36 \cdot 990 = 2{,}530{,}440 \end{aligned}

71 is the number of ways to pick 6 different ranks that contain a straight. The pair can be any of the 6 ranks and can have 6 suit combinations (6⋅6=366 \cdot 6 = 36). The remaining 5 cards have 45=1,0244^5 = 1{,}024 suit distributions, 34 of which make a flush, which leaves 990.

Third way

5 different ranks (for example, 4-5-6-7-8-8-8).

10⋅5⋅4⋅(256−3)+10⋅(52)⋅2,268=277,400\begin{aligned} 10 \cdot 5 \cdot 4 \cdot (256 - 3) + 10 \cdot \binom{5}{2} \cdot 2{,}268 = 277{,}400 \end{aligned}

The 5 different ranks must form the straight itself (10 options). The remaining 2 cards make either a three of a kind (first part) or 2 pairs (second part). For the three of a kind, we pick 1 of the 5 ranks and 4 suit combinations, and the remaining 4 cards have 44=2564^4 = 256 suit distributions, 3 of which make a flush. For 2 pairs, we pick 2 of the 5 ranks, the pairs have 6⋅66 \cdot 6 suit combinations and the remaining 3 cards 434^3, of which 36 distributions make a flush, so 36⋅64−36=2,26836 \cdot 64 - 36 = 2{,}268.

Final probability

Let's add up all the ways mentioned above.

3,372,180+2,530,440+277,400=6,180,020\begin{aligned} 3{,}372{,}180 + 2{,}530{,}440 + 277{,}400 = 6{,}180{,}020 \end{aligned}

Out of all 133,784,560 combinations, a straight occurs in only 6,180,020.

6,180,020:133,784,560≈1:21.65≈4.61938%\begin{aligned} 6{,}180{,}020 : 133{,}784{,}560 ≈ 1 : 21.65 ≈ 4.61938\% \end{aligned}

The probability of a straight is approximately 4.61938% (including straight flushes and royal flushes approximately 4.65046%).

Three of a kind

A three of a kind must consist of 3 cards of the same rank and 4 other different ranks (for example, A-A-A-4-5-6-7).

[(135)−10]⋅5⋅4⋅(44−3)=6,461,620\begin{aligned} \Bigg[ \binom{13}{5} - 10 \Bigg] \cdot 5 \cdot 4 \cdot (4^4 - 3) = 6{,}461{,}620 \end{aligned}

The cards must have 5 different ranks in total (to avoid, for example, a full house), and at the same time we must subtract the 10 options that make a straight. The three of a kind can be any of the 5 ranks and can be made in 4 suit combinations.

The remaining 4 cards can have any suits (444^4), but they must not all have the same suit as one of the cards in the three of a kind, otherwise they would make a flush (that's why we subtract 3).

Out of all 133,784,560 combinations, a three of a kind occurs in only 6,461,620.

6,461,620:133,784,560≈1:20.70≈4.82987%\begin{aligned} 6{,}461{,}620 : 133{,}784{,}560 ≈ 1 : 20.70 ≈ 4.82987\% \end{aligned}

The probability of a three of a kind is approximately 4.82987%.

Two pair

Two pair can be made either of 2 pairs and 3 other cards (for example, K-K-J-J-4-5-6), or of 3 pairs and 1 other card (only the 2 highest pairs count). Paradoxically, the probability of two pair is higher than the probability of no pair. We will split the calculation into 2 ways.

First way

2 pairs and 3 different cards, so 5 different ranks in total.

[(135)−10]⋅10⋅(6⋅62+24⋅63+6⋅64)=28,962,360\begin{aligned} \Bigg[ \binom{13}{5} - 10 \Bigg] \cdot 10 \cdot (6 \cdot 62 + 24 \cdot 63 + 6 \cdot 64) = 28{,}962{,}360 \end{aligned}

We pick 5 different ranks that don't make a straight, and 2 of them for the pairs ((52)=10\binom{5}{2} = 10). The suits of both pairs can be chosen in 6⋅6=366 \cdot 6 = 36 ways, and the remaining 3 cards must not complete a flush. It depends on how many suits the pairs share: 6 combinations of pairs with the same suits (62 distributions of the remaining cards), 24 combinations with one shared suit (63), and 6 combinations with no shared suit (64).

Second way

3 pairs and 1 different card.

(133)⋅(42)3⋅(401)=2,471,040\begin{aligned} \binom{13}{3} \cdot \binom{4}{2}^3 \cdot \binom{40}{1} = 2{,}471{,}040 \end{aligned}

We pick 3 ranks for the pairs, each pair has 6 suit combinations, and the last card can be any of the 40 cards of the remaining 10 ranks. Neither a straight nor a flush can occur here.

Final probability

Let's add up all the ways mentioned above.

28,962,360+2,471,040=31,433,400\begin{aligned} 28{,}962{,}360 + 2{,}471{,}040 = 31{,}433{,}400 \end{aligned}

Out of all 133,784,560 combinations, two pair occurs in 31,433,400.

31,433,400:133,784,560≈1:4.26≈23.49554%\begin{aligned} 31{,}433{,}400 : 133{,}784{,}560 ≈ 1 : 4.26 ≈ 23.49554\% \end{aligned}

The probability of two pair is approximately 23.49554%.

Pair

A pair consists of 6 different ranks (for example, J-J-2-3-6-7-8).

[(136)−71]⋅6⋅6⋅990=58,627,800\begin{aligned} \Bigg[ \binom{13}{6} - 71 \Bigg] \cdot 6 \cdot 6 \cdot 990 = 58{,}627{,}800 \end{aligned}

We pick 6 different ranks that don't make a straight. The pair can be any of the 6 ranks and can have 6 suit combinations. Just like with the straight, 990 suit distributions of the remaining 5 cards don't make a flush.

Out of all 133,784,560 combinations, a pair occurs in 58,627,800.

58,627,800:133,784,560≈1:2.28≈43.82255%\begin{aligned} 58{,}627{,}800 : 133{,}784{,}560 ≈ 1 : 2.28 ≈ 43.82255\% \end{aligned}

The probability of a pair is approximately 43.82255%.

No pair / high card

The last possibility in poker is not getting any hand at all. In this case, you can only rely on the rank of your highest card.

As mentioned, paradoxically, there is a higher chance of getting one or even two pairs than getting no pair.

1,499⋅(47−756−4−84)=23,294,460\begin{aligned} 1{,}499 \cdot (4^7 - 756 - 4 - 84) = 23{,}294{,}460 \end{aligned}

1,499 is the number of ways to pick 7 different ranks that don't contain a straight ((137)−217\binom{13}{7} - 217). The suit distributions are the same as in the first way of the straight.

Out of all 133,784,560 combinations, no pair occurs in 23,294,460.

23,294,460:133,784,560≈1:5.74≈17.41192%\begin{aligned} 23{,}294{,}460 : 133{,}784{,}560 ≈ 1 : 5.74 ≈ 17.41192\% \end{aligned}

The probability of no pair is approximately 17.41192%.

Table

The "1 in N" column shows how many hands it takes on average for the hand to occur once. The "Odds against" column shows the ratio of hands where it doesn't occur to hands where it does.

HandProbability1 in NOdds against
Royal flush0.00323%30,94030,939 : 1
Straight flush0.02785%3,590.573,589.57 : 1
Four of a kind0.16807%595594 : 1
Full house2.59610%38.5237.52 : 1
Flush3.02549%33.0532.05 : 1
Straight4.61938%21.6520.65 : 1
Three of a kind4.82987%20.7019.70 : 1
Two pair23.49554%4.263.26 : 1
Pair43.82255%2.281.28 : 1
No pair / high card17.41192%5.744.74 : 1
Total100%