40  Short-Handed & Heads-Up Cash

This chapter will leave you able to widen your ranges, sharpen your positional play, and ramp up aggression as the table shrinks from six-max to heads-up, then read and counter a single opponent over a long session.

Most players learn poker at full ring (nine-handed) or standard six-max tables, and they carry full-ring instincts into smaller games where those instincts quietly bleed money. Short-handed and heads-up cash are not just the same game with fewer seats. They are faster, more aggressive, more positional, and more variance-heavy. They are also where the strongest players make the most money per hour, because the edges compound. Every street, every decision, every range is wider and more contestable.

This chapter covers the progression from six-max down to four-max, three-max, and finally heads-up, the purest form of the game, where every hand is blind-versus-blind and the button decides everything. We build the strategic logic from the top down: why ranges widen, why position becomes decisive, why aggression is rewarded, and how to read and counter-adjust to a single opponent you will play for hours.

40.1 Why fewer players changes everything

Three structural facts drive every short-handed adjustment. Understand these and most of the tactical advice becomes obvious rather than memorized.

1. You pay the blinds more often. At a full-ring table you post the big blind one hand in nine. At six-max it is one in six; at heads-up, you post a blind every single hand and the big blind every other hand. The blinds are a fixed tax on doing nothing. The more often you pay them, the more you must fight to win them back, which means playing more hands and playing them more aggressively.

2. Fewer players means fewer hands can beat you. A hand’s strength is relative to how many opponents could be holding something better. With eight opponents, a hand like K♥T♠ is mediocre, since someone probably has a better king, a pair, or a dominating ace. With one opponent, K♥T♠ is a monster in relative-strength terms: its raw equity against a single random hand is strong, on the order of 60%. That does not make it a hand you 3-bet for value against a wide button (more on that below), but every hand’s equity rises as players are removed, because the chance that somebody woke up with aces collapses.

3. Position is worth more when the pot is contested more often. Position (acting last) is valuable because information is valuable. In short-handed play you reach contested, blind-vs-blind pots constantly, so the positional edge is exercised on a huge fraction of hands rather than occasionally. The button in a heads-up match has position on 100% of hands. That is the single most important number in this chapter.

TipKey idea

Short-handed poker is a game of contested pots. At full ring you can fold-fold-fold and wait for premiums. Short-handed, that “tight rock” strategy gets blinded to death and exploited by anyone willing to attack your folds. The correct baseline posture is aggressive selectivity: wide opening and 3-betting ranges, applied with positional discipline and a willingness to barrel (keep betting across streets).

40.2 Six-max cash: the modern baseline

Six-max is the most popular cash format online and the natural starting point. Four positions open before the blinds: the lojack/under-the-gun (UTG), the first seat in six-max; the hijack (HJ); the cutoff (CO); and the button (BTN), plus the two blinds. Each of these four seats has a distinct opening range, shown in the table below. Because two full-ring positions are removed, every range shifts looser.

Typical opening ranges (raise-first-in, or RFI: you are the first player into an unopened pot with a raise), expressed as a percentage of hands, look roughly like this. Treat them as approximate solver-informed baselines, not gospel. Exact frequencies depend on rake, stack depth, and your opponents.

Position	Approx. opening range	Character
UTG (first seat 6-max)	~15-19%	Pairs, broadways, suited aces, strong suited connectors
Hijack	~21-25%	Adds more suited gappers, weaker broadways
Cutoff	~26-30%	Most suited hands, all pairs, many offsuit broadways
Button	~42-50%	Any ace, any suited hand worth a raise, most connectors, wide offsuit
Small blind (RFI)	~36-44% raised	Linear/condensed: raise the strongest hands, fold the rest, plus a real limp/complete range (see below)

The button opening close to half of all hands is the defining feature of six-max. With only two players left to act and position guaranteed, the button profits simply by raising and taking the blinds uncontested a large fraction of the time. The flip side is that the big blind must defend wide, roughly ~40-50% (call plus 3-bet combined) against a typical 2.5x button open, to avoid being run over. Two qualifiers on that figure. First, the “defend ~50%” heuristic is tied to smaller opens: a 2x/min-raise gives the BB a better price and pulls the defend frequency up, while against a 2.5x it shades down. Second, the BB is not the only defender; the small blind can also cold-call or 3-bet, so the BB does not shoulder the whole minimum-defense burden alone. BB defense is mostly calling, since you close the action and get a price, with a 3-betting range layered on top.

Blind-vs-blind: the hidden battleground

When everyone folds to the small blind, you have a blind-vs-blind (BvB) confrontation. These pots are frequent in six-max and dominant in shorter formats, yet they are where weak players leak the most because the dynamics are unfamiliar.

The SB acts first preflop and is out of position for the whole hand, acting first on every postflop street with the BB closing behind it. This spot is worth comparing to the heads-up button covered later, because the two look alike preflop. Once it folds to the small blind in six-max, you are in a pure SB-versus-BB confrontation with the same 0.5/1 blinds, stacks, and rake as a heads-up button-versus-big-blind pot, and in both you act first preflop and choose to open, limp (just call the big blind), or fold. The decisive difference is postflop position. In a true heads-up match the small blind is the button, so it acts last postflop and is in position. In six-max (and four-/three-max) blind-vs-blind, the small blind acts first postflop and is out of position; the big blind holds the closing, button-like seat after the flop. That is why the BvB SB opens a much tighter range (~36-44%, below) and sizes up, while the heads-up button opens far wider (~80-90%) and sizes down. They share the preflop logic, but the BvB SB should never borrow the heads-up button’s width or small sizing.

At equilibrium the SB is not purely raise-or-fold. Solvers complete (limp) a meaningful chunk of the SB range, often around ~15-30% of hands in low-rake, deeper-stacked games, alongside a raising range. Limping lets marginal holdings see a cheap flop, realize equity, and keeps the SB’s calling range from being absent, which would otherwise let the BB attack a raise-or-fold SB freely. The limp range shrinks as rake rises and stacks get shallower, and widens as the game gets deeper and cheaper.

Many players adopt a defensible compromise: simplify to raise-or-fold. Raise around 36-44%, fold the rest, and skip the limp. The justification is rake and simplicity rather than any claim that limping is theoretically bad. A single raise-or-fold tree is far easier to execute, and in a meaningfully raked pool the thinnest limps lose value first, so folding them costs little. The same reasoning applies to the heads-up button later, which also carries an equilibrium limp range you can choose to drop (though the in-position button limps a wider set than this out-of-position SB). When you do raise the SB, size a touch larger (around 3x) than at a fuller table. Because you will be out of position postflop, a bigger raise denies the BB’s wide calling range its cheap price and helps compensate for the equity you will struggle to realize without position.

The BB, closing the action with position, defends against an SB raise very wide, frequently 50%+ against a min-raise and less against a larger SB open, mixing calls with a robust 3-betting range. Whoever fights harder in BvB tends to win the format, because these pots recur all night.

40.3 Four-max and three-max: the squeeze toward heads-up

As the table shrinks to four- and three-handed, every range slides further open and the button’s edge grows.

Four-max has four seats: the cutoff (first to act, effectively the “UTG” of a four-handed table), the button, the small blind, and the big blind. With only one player to act before the blinds, the first-in seat opens much wider than a six-max UTG, roughly 25-32%, while the button opens around 45-50% with position guaranteed and only the two blinds behind, essentially the same wide steal you would use three-handed on the button. The blinds adjust accordingly: the SB is again the out-of-position spot that acts first (limp/raise/fold, simplifiable to raise-or-fold), and the BB defends wide and 3-bets to punish a wide opener.

Three-max has only three positions: button, small blind, big blind. The button opens ~45-50% of hands. Folding the bottom half of your button range is correct, but folding too tight, surrendering blinds you should be attacking, is the cardinal three-handed sin. The blinds must counter by 3-betting more, because flat-calling out of position against a wide, aggressive button is a losing proposition.

A few transferable principles for short-handed (4/3-max) play:

• 3-bet more, flat less, especially out of position. Cold-calling out of position lets the in-position player realize equity and control the pot. 3-betting takes the initiative and the dead money. As the table shrinks, the ratio of 3-bets to flats should rise.
• Widen your value thresholds. Top pair with a decent kicker is a strong hand three-handed; it is often a premium heads-up. Adjust what you are willing to stack off with.
• Attack capped ranges. A range is capped when it can no longer hold the strongest hands. When an opponent just calls preflop and then checks, their range is rarely strong, and short-handed this happens constantly. Sustained, sized-up barreling against capped ranges drives short-handed win rates.

WarningCommon mistake

Treating four-max and three-max like a slightly looser six-max. Players widen their opening ranges correctly but fail to widen their 3-betting and barreling ranges to match. The result: they open A♦9♠ from the button, get flatted by the BB, c-bet (continuation bet) the flop, then shut down on the turn because “I only have ace-high.” Against a capped, over-folding range, that turn barrel is often a clearly +EV continuation, though whether to fire depends on the turn card, the board texture, and how much the opponent actually folds. Short-handed aggression has to be sustained across streets, well past the preflop raise.

40.4 Heads-up cash: the purest game

Heads-up (HU) is poker distilled. Two players, two cards each, position alternating every hand. There is no hiding, no waiting for a better spot, no folding into the money. You will play thousands of hands against the same opponent in a session, which makes it simultaneously the most reads-driven and the most adaptive format in poker. It is widely regarded as the highest-skill, highest-variance form of NLHE cash, and it is where the strongest players test each other directly.

The two seats

Heads-up has exactly two positions, and they have confusingly overlapping names, so let us be precise:

• The button (BTN) posts the small blind, acts first preflop but last on every postflop street, and is therefore in position postflop. This is the strong seat.
• The big blind (BB) posts the big blind, acts last preflop but first on every postflop street, and is out of position postflop. This is the weak seat.

Note the reversal from multiway play: heads-up, the small blind is the button and the aggressor. Every hand is blind-vs-blind by definition. There are no other players to provide dead money or to fear, just you, your opponent, and the rake.

TipKey idea

In heads-up, the button has position on 100% of hands. Position is the single largest structural edge in poker, and heads-up hands it to one player on every deal. A competent heads-up player wins money from the button and tries to lose as little as possible from the big blind. If you are not winning from the button, you do not have a heads-up game yet.

Button strategy: raise relentlessly

Because the button has position guaranteed, the baseline is to raise a wide range, frequently 80-90% of all hands at typical stacks and at the extreme nearly every hand. The raise takes down the blinds uncontested when the BB folds and guarantees position when called. Raising this wide is a sound, popular simplification, but it is still a simplification. At 100bb deep the solver keeps a genuine open-limp range, often around 15-30% of hands, to protect the BB’s option to check behind, realize equity cheaply with marginal holdings, and keep its raising range from being too uncapped. That limp range grows as stacks get deeper and shrinks as rake rises. For most players raise-or-fold is a fine default; just know you give up a small, real piece of equilibrium when you drop the limp entirely. As with the BvB small blind earlier, the decision to simplify the limp away carries over here, even though this in-position button limps and raises a far wider set.

Standard button raise sizing in HU is small, around 2x to 2.5x the big blind, because you are raising so wide that you want a cheap price on your weakest hands, and because position lets you outplay the opponent postflop regardless of how much went in preflop. A min-raise (2x) is extremely common in solver-approved HU strategy. The out-of-position BvB small blind sizes up to around 3x for the same underlying reason: lacking postflop position, it needs the larger raise to deny the BB a cheap call.

Big blind strategy: defend wide, 3-bet polarized

Facing a button that raises 80%+ of hands, the big blind cannot fold its way to profit. The BB must defend a huge fraction, often 60-70% against a min-raise, because the pot odds to call are excellent and the button’s range is weak. BB defense splits into:

• Calling with hands that play well postflop but aren’t strong enough to 3-bet for value: suited connectors, suited gappers, weak-to-medium aces, small pairs, and broadways that prefer to see a flop.
• 3-betting, whose default backbone is polar: only strong value and bluffs, with the medium hands held back. The value is big pairs, AK, AQ, and strong suited aces; the bluffs come from the bottom of your continuing range (suited wheel aces like A♠4♠, suited connectors like 7♥6♥, hands with blockers and backdoor potential). Your middling, decent-but-not-great holdings generally prefer to flat and keep the button’s bluffs in.

This polar structure is one of the most important concepts in heads-up, but its strictness depends on how wide the button actually is. The wider and weaker the opener, the more your value 3-bets merge downward. Against a typical 80-90% HU button, a strong-but-not-premium hand is comfortably ahead of the opener’s range and benefits from seizing the initiative, so you keep the polar backbone (premium value plus blocker bluffs) and add a layer of medium-strong hands to the value side, mixing in some 99, AJs, KQs, ATs, and AJo. Your truly medium hands still prefer to flat: K♥T♠ doesn’t want to get 4-bet off its equity and realizes equity fine by calling, while A♦3♦ blocks the button’s strong aces, can win by folding the button out, and can make the nut flush when called. The working rule is to lean polar and then merge value downward as the open widens.

WarningCommon mistake

Reflexively 3-betting every “good but not great” hand out of the big blind, always as pure value and in size. The merge described above is a measured addition (a slice of 99, AJs, KQs, ATs, and AJo), not a license to blast your entire upper-middle range. The genuinely medium holdings, such as K♥T♠, K♦J♠, and weak suited aces that flop well, still prefer to flat. When you 3-bet them indiscriminately you fold out the hands you dominate and bloat an out-of-position pot against the hands that have you crushed. Keep the bulk of your medium hands flatting, put your 3-betting energy into the polar extremes (premium value and well-chosen blocker bluffs), and merge in a controlled share of strong-medium hands only as the button’s open gets wider and weaker.

40.5 Fast reads and counter-adjustment

Because you play the same opponent for thousands of hands, heads-up is an arms race of adjustment. The theoretically sound (GTO, or game-theory optimal) baselines above are your default and your defense; they guarantee you cannot be heavily exploited. The money is made by deviating from them to punish a specific opponent, while watching for their counter-deviation.

A practical adjustment loop:

1. Establish a read from frequencies, not single hands. One hero-fold tells you little; a pattern over 50-100 hands tells you a lot.
2. Deviate to exploit it. Pick the single largest leak and attack it directly.
3. Watch for the counter. A thinking opponent will notice and adjust. The moment your exploit stops working, the read has flipped.
4. Re-center toward GTO, then re-read. When in doubt, return to the unexploitable baseline rather than guessing.

Online, your read engine is your HUD (heads-up display). The most useful HU stats and rough interpretive ranges:

• VPIP (voluntarily put money in pot): in HU this is very high for everyone, with typical competent ranges running ~70-90%. Below ~60% means a player folding the button too much (attack their blinds, and over-fold to their raises since their range is strong).
• PFR (preflop raise): in HU, VPIP and PFR sit close together for good players, who raise rather than limp or call. A large VPIP-PFR gap signals a passive limper/caller you can value-bet relentlessly and rarely need to bluff.
• 3-bet %: HU 3-bet frequencies run high relative to full ring, typically ~12-20% for solid players and higher for aggressive ones. A low 3-bet (well under that band) means their flats are wide and their 3-bets are strong, so fold to 3-bets but barrel their flatting range. A very high 3-bet means you can 4-bet lighter and flat their 3-bets in position.
• Fold to c-bet / aggression frequency: the workhorse stats for postflop exploits. A high fold-to-c-bet says barrel everything; a low one says value-bet thin and stop bluffing.

NoteDrill

Over your next 200 hands of heads-up (or a hand-history review of them), track just two numbers for your opponent: how often they fold the big blind to your button raise, and how often they fold to your flop c-bet. If either number is above roughly 65%, exploit it hard: raise every button and c-bet every flop until they adjust. Write down the hand number where the leak stops working, since that is the moment they counter-adjusted and your read must flip with it.

Note that these stat thresholds are approximate and population-dependent; treat them as directional, not precise. Small samples lie. A 70% fold-to-c-bet over 20 hands is noise; over 200 hands it is a strategy.

40.6 A fully worked heads-up hand

Let us walk a representative button-vs-big-blind hand at 100bb deep to see the concepts in motion.

Preflop. It folds to you on the button with A♣5♣ (a suited wheel ace). You raise to 2bb, a standard wide, small button open. A♣5♣ is a great hand here, with an ace blocker, flush potential, and the ability to make a wheel. The BB calls. Pot is 4bb.

Flop: K♣8♣3♦. The BB checks (out of position, they check their entire range here by default). You hold A♣5♣: bottom-end equity, but you have the nut flush draw (two clubs in hand on a two-club board, headed by the A♣) plus overcard blockers. This is a textbook semi-bluff c-bet, betting a drawing hand that can win immediately or improve. On a king-high two-tone board that hits your wide button range hard, the high-frequency size is small (around 25-33% pot, i.e. ~1.3bb into 4bb), but with a strong draw you can also choose a larger, more committal c-bet of half pot, say 2bb into 4bb, to build the pot behind your equity when called. Here you bet 2bb (half pot). The BB calls. Pot is 8bb.

Turn: K♣8♣3♦ → 7♥. The 7♥ is a blank that doesn’t complete the obvious draws. Your A♣5♣ makes no straight or straight draw here: a wheel would need both a 2 and a 4, and on K♣8♣3♦7♥ there is no one-card straight available to you. Your equity is the nut flush draw (nine clubs, roughly 19.6% to hit on the river) plus ace-high showdown value. The BB checks again.

Here is the key decision. You have two profitable options:

• Check back to realize your equity for free and keep the BB’s bluffs in on the river. This controls the pot with a drawing hand and is fine.
• Barrel again as a semi-bluff. Against an opponent who folds turns too often (your HUD read), a second barrel of ~5bb into 8bb folds out the BB’s missed connectors and middling pairs while still having the nut flush draw as backup equity.

Suppose your read is that this BB over-folds to turn barrels (fold-to-turn-c-bet running high). You barrel 5bb. The BB folds. You take down 8bb with ace-high, exactly the kind of pot that, repeated hundreds of times, separates winning heads-up players from breakeven ones. Position, fold equity, and a live backup draw made the barrel profitable here even with no made hand.

Had the BB called, the river club gives you the nuts to value-bet; a brick lets you give up having lost the minimum. This is the heads-up engine: wide preflop aggression, equity-backed continuation betting, and reads-driven barreling, all leveraged by guaranteed position.

40.7 Rake: the silent counterweight to “widen everything”

Almost every directional instruction in this chapter pushes the same way: wider, looser, more aggressive. One large, real-world force pushes the opposite way, and short-handed and heads-up games are where it bites hardest: rake.

Rake hurts these games disproportionately for structural reasons. You contest a pot almost every hand, so you pay rake on a huge fraction of deals rather than occasionally. Blind-vs-blind and heads-up pots are small, so a percentage-based rake eats a larger share of each pot, and the per-pot cap that protects you in big multiway full-ring pots rarely comes into play. The net effect is that rake compresses your edge most precisely in the games this chapter is about.

That has a concrete strategic consequence: high rake should tighten you, not loosen you. When the house takes a meaningful cut of every small pot, marginal hands that are barely break-even pre-rake become losers, so:

• BB defense narrows. The “defend ~40-50%” and “60-70% vs a min-raise” figures assume modest rake. In a high-rake pool you simply cannot profitably call as wide, because you must overcome both the button’s range and the rake every time you continue.
• Marginal SB opens and limps narrow. The thinnest steals and completes lose value first when each pot is taxed.
• Thin value and marginal bluff frequencies shrink. Bets that win a tiny pot net of rake are worth less; the rake quietly raises the bar for a profitable thin value bet or a marginal barrel.

The solver-derived “wide” baselines throughout this chapter assume a modest, realistic online rake. In a high-rake environment, treat them as ceilings and shade tighter. And because rake is a fixed drag on every hand, game and site selection plus rakeback can decide whether widening pays at all. Choosing a lower-rake site, a better cap structure, or a meaningful rakeback deal can swing a marginal short-handed strategy from losing to winning without your changing a single in-game decision.

40.8 Variance and bankroll: the honest part

Short-handed and especially heads-up cash carry the highest variance of any cash format. The reasons are structural, not bad luck: you play far more hands to showdown, you play them with wider ranges and thinner edges, stacks go in lighter, and your entire result rides on a single opponent rather than being diversified across a table. Win rates are measured in big blinds per 100 hands, and the swings around that win rate are large.

Three honest consequences:

1. You need a deeper bankroll. These are only rules of thumb, but they anchor the gap: a full-ring or 6-max grinder is often comfortable on roughly 30-50 buy-ins for their stake, whereas heads-up and very short-handed players typically want 50-100+ buy-ins for the format and stake, because the downswings are longer and deeper even when you are clearly winning. Move down a level the moment your roll dips below the floor; high variance punishes under-rolled play far faster here than at full ring.
2. Edge can be small even when you’re better. Two competent heads-up players can have a win rate so close to zero that variance dominates over thousands of hands. Game selection, meaning who you choose to play, often matters more than your absolute skill.
3. Tilt control is non-negotiable. Because you are locked in with one opponent who is actively trying to get under your skin, heads-up is psychologically brutal, and the format rewards the player who stays on their A-game longest. (The mental-game tools are covered in the psychology chapters; here, just know that heads-up tests them harder than any other format.)

TipKey idea

In heads-up, game selection is a skill on par with your in-game strategy. Strong players sit out or avoid tables where a tougher player is waiting, and seek out spots where they hold a clear edge. There is no shame in declining a match against a better player; at heads-up you cannot hide from them once seated, so the choice of whether to play is itself strategic. Refusing bad matchups and accepting good ones is how professionals protect their win rate from variance.

40.9 Putting it together

The throughline from six-max to heads-up is a single, intensifying logic: as players are removed, ranges widen, position becomes more decisive, blind-vs-blind frequency rises toward 100%, and aggression is increasingly rewarded. Master the progression and you master a family of the most profitable and most demanding games in poker.

A compact summary to carry to the tables:

• Six-max: open wide from late position, defend the big blind aggressively, and win the blind-vs-blind battle. The button opening ~45% and the BB defending ~40-50% against a 2.5x button open (closer to 50% against smaller 2x/min opens, and remembering the SB co-defends) are the load-bearing numbers.
• Four/three-max: widen everything further, lean on 3-bets over flats out of position, and barrel relentlessly against capped ranges. Folding the bottom of your button range is fine; folding it too tight is the real sin.
• Heads-up: every hand is BvB. Raise the button relentlessly, defend the big blind wide, 3-bet on a polar backbone (premium value plus blocker bluffs) while merging your value downward as the open widens, and treat the game as an adjustment arms race against one opponent. Win from the button; minimize losses from the big blind.
• Throughout: respect the variance, keep a deeper bankroll (roughly 50-100+ buy-ins for HU), watch the rake (it tightens every “wider” instruction above), and select your games, because at the small-handed tables, the math of your edge and the discipline of your mental game are exposed for everyone, including you, to see.