Why counting comes first
Many game probabilities reduce to one fraction: the number of outcomes you care about divided by the number of outcomes that are possible. This works when the outcomes are equally likely, as with a fair die or a well-shuffled deck. The hard part is rarely the division. It is counting both numbers correctly without listing them one by one.
Combinatorics is the branch of mathematics that supplies those counting shortcuts. A handful of rules cover most of what you will meet in dice, card and board games, and they are worth learning in order.
The product rule
If one choice can be made in a ways and a second, independent choice in b ways, the two together can be made in a × b ways. This is the most useful counting rule there is.
Roll two six-sided dice, one red and one blue. The red die has 6 results and the blue die has 6, so there are 6 × 6 = 36 ordered outcomes. Three dice give 6 × 6 × 6 = 216. A four-digit code on a lock, where each digit runs from 0 to 9, gives 10 × 10 × 10 × 10 = 10,000 codes.
The rule still works when the number of options changes from step to step, as long as you count the options available at each step. Dealing a first card from a standard 52-card deck gives 52 options. The second card, with the first removed, gives 51. Two cards dealt in order therefore have 52 × 51 = 2,652 possible outcomes.
The sum rule
If a task can be done in one way or another, and the two ways cannot happen together, add the counts. A single card drawn from a deck is a heart (13 cards) or a spade (13 cards), so there are 13 + 13 = 26 ways to draw a heart or a spade, because no card is both.
When the categories overlap, you must subtract the overlap or you will count it twice. The number of cards that are hearts or kings is 13 + 4 − 1 = 16, because the king of hearts belongs to both groups. A common beginner error is to skip that subtraction.
Ordered arrangements
When you pick items one at a time from a pool and the sequence matters, you are counting permutations. Arranging n distinct items in a line gives n × (n − 1) × … × 1 possibilities, written n! and read as “n factorial”. Three distinct cards can be laid out in 3! = 6 orders. Five can be laid out in 5! = 120 orders.
Choosing only some of the items in order uses a shortened product. Picking and ordering 5 cards from 52 gives 52 × 51 × 50 × 49 × 48 = 311,875,200 sequences. Notice how quickly the numbers grow; this is the reason shuffled decks are, for practical purposes, always in an order that has never occurred before.
Unordered selections
In most card games, the order in which you received your cards is irrelevant. You hold a hand, not a sequence. To count hands, take the ordered count and divide by the number of ways to reorder the same cards.
Each five-card hand can arrive in 5! = 120 orders, so the number of distinct hands is 311,875,200 ÷ 120 = 2,598,960. That figure is the denominator behind almost every poker hand frequency you will see quoted, and it is the reason poker odds can be stated exactly. The general formula for choosing k items from n is n! ÷ (k! × (n − k)!), usually shown as “n choose k”.
A smaller example: a two-card starting hand from 52 cards has 52 × 51 ÷ 2 = 1,326 combinations. Those collapse into 169 strategically different starting hands once you treat suits as interchangeable, which is why tables of starting hands list 169 entries rather than 1,326.
The distinction between these two ideas trips up many beginners, so check it every time you set up a count.
Putting the rules together
Real questions usually need more than one rule. Consider the chance that a two-dice roll totals 7. List the outcomes by the product rule: 36 in all. Count the pairs that give 7: (1,6), (2,5), (3,4), (4,3), (5,2), (6,1), which is 6. The probability is 6/36, or 1/6. The same pattern, applied to every total, produces the familiar craps odds table in which 7 is the most common sum.
Here is a slightly longer example. A deck-builder wants to know how many ways exist to choose 2 of the 4 aces and 3 of the 48 other cards in a five-card hand. Choosing the aces: 4 × 3 ÷ 2 = 6 ways. Choosing the other three cards: 48 × 47 × 46 ÷ 6 = 17,296 ways. By the product rule, the total is 6 × 17,296 = 103,776 hands with exactly two aces. Dividing by 2,598,960 gives about 0.0399, or just under 4 percent. Splitting a problem into “choose this group, then choose that group” is how a hypergeometric distribution is built, and it is the heart of reasoning about deck probability.
Counting with repeated trials
Counting also handles repeated attempts. Flip a coin four times and there are 2 × 2 × 2 × 2 = 16 sequences. How many contain exactly two heads? You choose which 2 of the 4 flips are heads: 4 × 3 ÷ 2 = 6. So the probability is 6/16, or 0.375. The same logic scales to any number of trials, though the arithmetic gets tedious by hand.
Common mistakes
- Treating non-equal outcomes as equal. The totals 2 through 12 on two dice are eleven results, but they are not equally likely. Count the 36 underlying pairs instead.
- Forgetting that dice are distinguishable. (2,5) and (5,2) are different outcomes even if both dice look alike.
- Mixing up ordered and unordered counts. If your numerator is unordered, your denominator must be too.
- Double-counting overlaps. Subtract anything counted by two categories at once.
- Assuming a count always means a probability. The ratio only equals the probability when each outcome is equally likely.
When counting gets too large
Some problems are too big or too tangled to count by hand, such as the chance of a certain pattern appearing in a sequence of many dice. In those cases people turn to a computer, and a Monte Carlo method is a practical approximation. It does not replace the counting rules, though. It is far easier to check a simulation against a small case you can count yourself than to trust it blindly.
For lotteries, the same tools give the odds in a single step: choose k numbers from n and the total number of tickets is “n choose k”. The sheer size of that number is what makes the quoted lottery odds so long and jackpot wins so rare. As always, these calculations describe long-run frequencies, and any individual outcome remains uncertain.