Heads-up poker is not six-max poker with four seats removed. With only two players, the blinds arrive every hand, ranges widen dramatically, position changes on every deal and small strategic leaks are repeated at maximum frequency. A player who waits for conventional “good cards” will surrender too many blinds; a player who attacks every pot without regard to stack depth, rake or opponent response will create a different but equally expensive leak.
The format rewards clear range construction and rapid adaptation. It does not reward guessing what one individual hand “must mean.” The objective is to choose actions that perform well across the opponent’s plausible range while updating that range as reliable evidence accumulates.
Position is built into the blind structure
In heads-up hold’em, the button posts the small blind and acts first before the flop. After the flop, the button acts last. This creates a powerful combination: the button can enter cheaply and then use position on three later streets.
The big blind receives a discount to continue because part of the call is already posted, but must play the rest of the hand out of position. That difference explains why competent button opening ranges are very wide and why the big blind must defend far more hands than in a full-ring game.
| Seat | Preflop role | Postflop role | Main strategic pressure |
|---|---|---|---|
| Button / small blind | Acts first and may limp, raise or fold | Acts last | Use positional value without opening so wide that 3-bets become highly profitable |
| Big blind | Closes the preflop action against a call or raise | Acts first | Defend the blind while avoiding dominated, difficult hands at excessive frequency |
At 100 big blinds, a button may profitably enter with most starting hands against many opponents. At 20 big blinds, opening size, limp strategy and all-in thresholds change. A chart that ignores effective stack depth is not a heads-up strategy.
Starting ranges are wide but not random
Wide ranges still have structure. Strong pairs, high cards, suited aces and connected hands retain robust equity and playability. Weak offsuit hands can be opened because position and fold equity matter, but they realize less equity when called and are vulnerable to aggressive 3-betting.
A practical button strategy can use a mixture of small raises and limps. Raising every hand gives the big blind a simple target and enlarges pots with the weakest part of the range. Limping some medium and weak hands can reduce variance and protect the limp range, provided strong hands are included often enough that a raise over the limp is not automatically profitable.
The big blind should not defend by copying the opening range. It needs separate calling and 3-betting ranges. Calls preserve dominated and medium-strength hands at a lower price; 3-bets generate fold equity and build pots with value or selected bluffs.
Preflop sizing changes the required defence
A min-raise risks 1.5 additional big blinds to win the 1.5 already posted. It needs folds 50% of the time to show an immediate profit before postflop equity is considered. That does not mean the big blind should defend exactly 50%; equity realization, rake and future actions matter. It does show why folding three quarters of hands to a min-raise is usually untenable.
Larger opens create more immediate fold equity but risk more and give the big blind a worse price. Against a player who overfolds, increasing size can be effective. Against an aggressive 3-bettor, a smaller open or protected limp strategy may keep the weakest opens from becoming too expensive.
Rake alters everything at low stakes. A marginal call that would be defensible in a rake-free model can become a fold once a percentage of the pot is removed. Heads-up players should confirm whether the site charges rake preflop, caps it by player count and applies different structures to cash, sit-and-go or jackpot formats.
Board coverage matters more than one pair
Because both players reach the flop with wide ranges, weak pairs, ace-high and draws have more value than they would in a multiway pot. At the same time, neither player can credibly represent every strong hand on every board.
A high-card dry flop such as A-7-2 often favours the preflop raiser because that range contains many strong aces. A low connected board such as 8-7-6 can interact well with the big blind’s calling range. The player with the range advantage can bet more often; the player with the nut advantage can use larger sizes with the strongest hands and appropriate bluffs.
Automatic continuation betting is still exploitable. A defender who check-raises frequently on favourable textures can punish bets made with hands that have little equity and no useful blockers.
Bet size should express the range, not emotion
Small bets work when a large part of the range can bet for thin value or protection and the opponent’s response is constrained. Large bets work when the betting range is polarized between strong value and bluffs.
In heads-up play, overbets appear more often because ranges can be capped. If one player rarely holds the strongest combinations after taking a passive line, the other can apply pressure with large river bets. The bluffing range should contain hands that block calls and unblock folds rather than random missed draws.
A simple river ratio illustrates the logic. A pot-sized bet risks one pot to win one pot and offers the caller 2-to-1. A balanced bettor can include roughly one bluff for every two value bets. An opponent who folds substantially more than one third permits extra bluffs; one who calls too widely should face more value and fewer bluffs.
Equity realization separates playable hands from raw equity
Two hands can have similar all-in equity but perform differently through multiple streets. A suited connected hand can make strong draws and disguised value; a weak offsuit king may be dominated and struggle to call pressure on later streets.
Position improves equity realization because the button sees the opponent’s action before deciding. It can take free cards, control pot size and value-bet more accurately. The big blind compensates through check-raises, leads on changing textures and preflop aggression.
Software equity percentages should therefore not be treated as direct calling instructions. The model must include position, stack-to-pot ratio, rake and the future strategy of both players.
Exploitative adjustments should be evidence-based
Heads-up poker makes tendencies visible quickly, but small samples still mislead. A player who folded the first four big blinds may simply have received poor hands. More reliable patterns include repeated sizing tells, extreme fold-to-3-bet rates over meaningful samples and consistent river behaviour.
- Opponent overfolds preflop: open more hands and consider a larger size.
- Opponent 3-bets too often but folds to 4-bets: add blocker-based 4-bet bluffs and trap selected premiums.
- Opponent calls too widely: reduce low-equity bluffs and value-bet thinner.
- Opponent check-raises excessively: check back medium-strength hands and continue with stronger draws.
- Opponent underbluffs rivers: make disciplined folds rather than paying to “keep them honest.”
The adjustment should target a documented imbalance. Changing strategy after every showdown produces unstable play and makes the player easier to manipulate.
Balanced play protects against counter-adjustment
Against an unknown or strong opponent, a coherent baseline prevents obvious exploitation. Strong hands must appear in checking, calling and limping ranges as well as aggressive ranges. Bluffs require credible value companions.
Balance does not require memorizing a solver output to the decimal. It means avoiding actions that reveal hand strength with certainty. If every limp is weak, every large river bet is value and every check-back is surrender, an observant opponent does not need sophisticated software.
Solver study is most useful when it teaches relationships: how opening size changes defence, how ranges shift with stack depth and which blockers matter. Copying a solved range from a rake-free 100-big-blind model into a 25-big-blind high-rake game can be worse than using a simpler adjusted strategy.
Variance and mental pressure are unusually concentrated
One opponent receives every winning pot and applies every difficult decision. There is no quiet orbit in which other players compete. This can make normal variance feel personal and encourage revenge-based adjustments.
Heads-up win rates can also look large in big blinds per 100 hands, but standard deviation is high and opponent quality changes quickly. A player who dominates one weak opponent may be losing against the pool as a whole.
Practical controls include a stop condition based on decision quality, a separate bankroll for the format and a willingness to leave when the opponent is stronger. Quitting a bad match is table selection, not defeat.
A practical heads-up study process
- Build button opening, limping and response ranges for each common stack depth.
- Build big-blind call and 3-bet ranges against each opening size.
- Review boards by range advantage and nut advantage rather than memorized slogans.
- Track decisions by category: missed value, overfold, loose bluff-catch and sizing error.
- Separate population tendencies from one-opponent reads.
- Recalculate strategy when rake or format changes.
Heads-up poker is a contest of repeated small decisions. The widest range does not automatically win; the player who converts position, price and observed tendencies into disciplined ranges usually does.
Related GambleRoad guides cover online poker rules and stakes, fold equity and poker variance.