This guide separates three questions that are often collapsed into one colored square: Which hand classes are possible? How many exact card combinations remain? How likely is the player to take this action with each one? PokerStars' hand-reading lesson explains the combo distinction; the original worksheet below makes every removal and weighting step visible.
One Exact Holding Versus a Hand Class
A♠K♥ is one exact two-card holding. AKo means ace-king offsuit, a hand class with 12 exact suit combinations in a fresh 52-card deck. AKs has four suited combinations. AA has six ways to choose two aces. Thus saying “AK or AA” without suit or weights hides that there are 16 AK combinations but only six AA combinations before any cards are known.
In no-limit Hold'em, the usual **13×13 starting-hand matrix** has 169 hand classes: 13 pocket-pair classes along the diagonal and, for each of the 78 distinct two-rank pairings, one suited and one offsuit class. A full deck has 13 × 6 = 78 pair combinations, 78 × 4 = 312 suited unpaired combinations, and 78 × 12 = 936 offsuit unpaired combinations. Add them and you get 1,326 exact two-card holdings, matching 52 choose 2 = 1,326.
| Hand class | Meaning | Exact combos before card removal |
|---|---|---|
| AA | Any two aces | 6 |
| AKs | Ace and king of the same suit | 4 |
| AKo | Ace and king of different suits | 12 |
| AK, unspecified | Both suited and offsuit AK | 16 |
The matrix is a compact index, not a frequency table by itself. A cell colored “raise” may contain four, six or 12 exact combos depending on its type, and a solver may mix actions within a cell. Counting colored cells as if each represented the same chance is a basic range error.
Build a Starting Model From Position and Action
Before examining a particular board, note the table format, effective stacks, position and previous action. A button open after folds is not interchangeable with an early-position raise, a call of that raise or a big-blind defense. The later the seat, the more information the player has about who already folded; that can change a plausible opening range. Our poker positions guide explains the seat names. This page focuses on opponent-range updating, not a recommended preflop chart.
For a fully auditable teaching example, suppose a six-max cash-game button raises and we choose this deliberately tiny hypothetical subset of their possible hands: AA, KK, AKs, AKo, QJs, 76s. It is not a realistic complete button-opening strategy or a recommendation to open only these hands. We use it because six hand classes are enough to demonstrate card removal without burying the arithmetic under hundreds of cells. Before we know any cards, their counts are 6 + 6 + 4 + 12 + 4 + 4 = 36 exact combinations.
The statement “button raises all 36” is an assumption for this worksheet, not a measured tendency. In a real study, assign an inclusion or action frequency to each class from a reviewed source or a documented observation set. Avoid reverse-engineering certainty from one opponent action.
Checkpoint 1: Hero Cards Remove Possible Combinations
You hold A♣Q♦ in the big blind. The opponent cannot simultaneously hold either of those exact cards. Remove every combo containing A♣ or Q♦ from the hypothetical starting set. The counts become:
| Class | Fresh-deck combos | After hero A♣Q♦ | Why it changed |
|---|---|---|---|
| AA | 6 | 3 | Only three aces remain; choose two. |
| KK | 6 | 6 | Hero holds no king. |
| AKs | 4 | 3 | A♣K♣ is impossible. |
| AKo | 12 | 9 | The three A♣–non-club-K combos are impossible. |
| QJs | 4 | 3 | Q♦J♦ is impossible. |
| 76s | 4 | 4 | Hero holds neither rank. |
| Total | 36 | 28 | Eight exact combos removed. |
This is card removal, often called a blocker effect. It is not proof that the opponent is “less likely to have aces” in every strategic sense. The relative probability of each hand class also depends on the range construction—which hands the player would raise and how often—not only the number of legal card combinations. But if a combo requires A♣, it is impossible once that card is in your hand. The PokerStars combo lesson gives comparable ace-blocker logic; our six-class counts are independently enumerated.
Checkpoint 2: The Board Removes More Cards
The flop is K♣7♦2♠. Those three cards are also unavailable to the opponent. Apply the same mechanical removal to the 28 combos that survived hero's cards:
| Class | After hero cards | After flop K♣7♦2♠ | Specific effect of board |
|---|---|---|---|
| AA | 3 | 3 | No ace appears. |
| KK | 6 | 3 | K♣ removes one of four kings. |
| AKs | 3 | 3 | A♣K♣ was already blocked by hero; other suited AK remains. |
| AKo | 9 | 6 | Three remaining-ace/K♣ offsuit combos disappear. |
| QJs | 3 | 3 | Neither queen nor jack appears. |
| 76s | 4 | 3 | 7♦6♦ disappears. |
| Total | 28 | 21 | Seven more exact combos removed. |
The order of excluding known cards does not affect the final legal set. We show hero first and flop second so a reader can follow why some possible AKs combos are not removed twice. In particular, A♣K♣ was already impossible before K♣ appeared on the board. A careless “minus one for each blocker” can double-subtract it.
The flop also changes hand strength. In this tiny range, KK now has a set, AK has top pair, and QJs may have a backdoor or overcard-related path depending on suits and future cards. Counting surviving combos does not establish which ones bet. That requires a separate behavioral assumption.
An Observed Bet Filters a Range Only Through a Model
Suppose the button continuation-bets this flop. You may be tempted to cross out every unpaired hand and call the bet “obviously value.” That is not an observation; it is an assumption about the opponent's betting frequencies. To make the inference auditable, assign explicit illustrative frequencies to the six remaining classes:
| Class | Legal flop combos | Hypothetical bet frequency | Weighted betting-combo equivalents |
|---|---|---|---|
| KK | 3 | 100% | 3 |
| AKs | 3 | 100% | 3 |
| AKo | 6 | 100% | 6 |
| AA | 3 | 0% | 0 |
| QJs | 3 | 33⅓% | 1 |
| 76s | 3 | 33⅓% | 1 |
| Total | 21 | Mixed | 14 |
The 14 is a sum of weighted combo equivalents, not 14 guaranteed exact hands the opponent always bets. For QJs, three legal suit combinations each get a one-third betting weight; their aggregate contribution is one. Another opponent might check all QJs, bet AA, or use different sizes. If you change those frequencies, the updated betting range changes. The worksheet demonstrates conditional weighting, not a claim that this is a GTO strategy or an observed player profile.
After the illustrative bet, the model's weighted betting mass is 14. A reader could compute the modeled share of KK + AKs + AKo as 12/14≈85.71%. But that number is only as credible as the stated initial range and invented bet frequencies. It is not a live opponent's measured top-pair-or-better rate. The correct hand-reading habit is to expose the assumptions, test them against evidence, and revise them when later action conflicts.
Bet Size, Turn Card and Later Action
A small continuation bet and a large one need not come from the same range mix. A turn card can remove new combinations, complete draws or make a previously plausible hand less likely to take another aggressive action. At each street, update in two separate passes: first remove impossible cards, then adjust action weights. You can use a range grid or a spreadsheet, but preserve the input range and frequencies so someone else can reproduce the result.
Suppose a turn card is 4♥ in our example. It does not remove any additional listed pairs, AK, QJs or 76s combos through rank overlap, but it can change whether a given hand class wants to bet again. A second bet is not permission to delete all checks or bluffs by fiat. You would need a new action-frequency column for that size and situation. We stop at the flop because no opponent-specific frequencies have been observed; pretending to know the turn strategy would add false precision.
This separation also protects against a common misuse of solvers. A solver output belongs to its configured ranges, bet sizes, stacks and board; copy its colorful result into a different game and the assumptions no longer match. Our GTO guide introduces the strategic framework, but no framework rescues an incorrectly specified range. Document the model before trusting the percentages.
Equity Is Measured Against a Range, Not One Imagined Hand
If you hold A♣Q♦ on K♣7♦2♠, your prospects differ against the holdings in our toy range: a set of kings, top pair with ace-king, pocket aces, a pair of sevens or unpaired queen-jack. Comparing your hand with exactly K♥K♦ answers a narrower question than comparing it with a weighted set of possible opponent holdings. A range-equity calculation needs both the remaining card combinations and their action weights, then accounts for turn and river outcomes. Our 14 weighted betting-combo equivalents are not an equity figure.
Nor does “my hand wins against six of 14 weighted combos” automatically equal its showdown equity: hands can improve on future cards or split pots, and later betting can change a decision. Pot odds compare the current call cost with the pot it can win; they are another input. Keeping range composition, equity versus that range and price to continue distinct prevents a neat-looking chart from smuggling in a conclusion it never calculated.
A Reproducible Worksheet You Can Audit
For each hand class in a model, record: class label, fresh-deck combos, known hero cards, visible board cards, surviving exact combos, inclusion weight before action and action weight after each bet or check. Sum weighted combos only after removing impossible exact cards. If you cannot reproduce a highlighted cell from those inputs, the chart is an illustration, not an auditable range.
Our six-class worksheet uses a normal 52-card Hold'em deck, hero A♣Q♦, flop K♣7♦2♠ and an artificial button-opening set. Its mechanical totals are 36 → 28 → 21. Its behavioral example is 21 → 14 weighted betting-combo equivalents only under the frequencies printed in the table. Those numbers should stay attached to the diagrams and translated copies so a reader never mistakes them for a universal button strategy.
Avoiding False Certainty
The most useful phrase in a hand review may be “under this model.” That is not indecision; it names the boundary of the evidence. A prior hand sample can justify a range estimate, but it is still a sample. A single bet is evidence of an action, not proof of one exact holding. A board card makes some exact combinations impossible; it does not dictate how often an opponent bluffs. An attractive heat map can hide all three distinctions.
The hand-review guide shows how to preserve the original decision and compare plausible ranges without letting the showdown rewrite the evidence.
When the model changes, state what changed: a known card, a position assumption, an opponent tendency, a bet size or a stack constraint. If a result changes dramatically when one unsupported betting frequency moves from 33% to 0%, tell the reader. The worksheet's purpose is to make that sensitivity visible, not to make the answer sound certain.
Quick Answers
How many starting-hand classes are there in Hold'em? A 13×13 matrix holds 169 classes. They represent 1,326 exact two-card combinations before card removal.
How many combinations are in a pair, suited hand or offsuit hand class? Normally six, four and 12, respectively, before known cards are removed.
Does holding an ace mean my opponent cannot have pocket aces? No. If you hold one ace and no other ace is known, three aces remain and pocket aces still has three exact combinations.
Can a bet tell me the opponent's exact hand? No. A bet can update a range only through an explicit model of which legal hands would bet and how often. Record that assumption separately from card removal.
About the Editorial Team
The Poker-Eye Editorial Team researches, checks, and maintains product and educational content. Sources and product boundaries are reviewed before publication.