Starting hands in Texas Hold’em

The best starting hand is a pair of aces: against one random hand it takes 85.2% of the pot on average. Then come kings (82.4%), queens (79.9%), jacks and ace-king of one suit. The weakest hand heads-up is 3-2 offsuit at 32.4%; against several opponents, 7-2 offsuit sinks to the bottom.

Two cards can be dealt in 1,326 ways, but there are only 169 different hands, because suits are interchangeable before the flop: 13 pairs, 78 suited hands and 78 offsuit ones. We played each of them two million times against one to eight random opponents with our poker engine. The chart below holds all 1,352 results.

Pairs run down the diagonal. Above it are suited hands (s), below it offsuit hands (o): AKs in the top row is ace-king of one suit, AKo in the second row ace-king of different suits, and T stands for ten. Pick the number of opponents and tap any hand.

The colour compares each hand with its fair share — the part of the pot every hand would get if all hands were equal: 50% against one opponent, 33.3% against two, 11.1% against eight. Green hands win more than their share against random cards, red ones less.

The number is equity: the share of the pot a hand wins on average if it goes to the river with no more betting. A win counts as 1, a two-way split as ½ and a three-way split as ⅓. A♠ A♥ against one random hand wins 84.9% of the time and splits 0.5%, so its equity is 85.2%.

The key word is “random”. Real opponents do not play random cards: someone who raises from the first seat usually holds something much better than average. The chart tells you how strong a hand is on its own, not how it will fare against a particular raise. To test a hand against specific cards, use the odds calculator.

Equity also ignores everything after the flop. 7♠ 6♠ rarely makes a straight or a flush, but when it does it can win a big pot; K♣ 7♦ often makes a pair of kings and loses a big pot with it. Good players call this playability, and it is why the ranges in our positions guide are not simply the top of this chart.

The groups come from our numbers. AA, KK, QQ and JJ are in the top five against any number of opponents; AK suited never drops below eighth place, AK offsuit never below twelfth. The strong hands stay inside the top 26 whatever the count. The premium group is 40 of the 1,326 two-card combinations, so you get one about once every 33 hands.

Speculative hands behave like the mirror image of weak offsuit hands. Heads-up they are worth between 0.83 and 1.14 of their fair share; against eight opponents every one of them takes 13–30% more than its share.

Against one opponent, high cards rule: a pair, or even a lone ace, is often enough. Against many, somebody nearly always makes two pair or better, so hands that can become straights and flushes gain value and lonely high cards lose it. Our numbers show it plainly:

Suited connectors climb 50 to 70 places; bare aces and kings with a small kicker fall 50 places or more. Against eight opponents A6o and K8o are already below their fair share of 11.1%.

Suits are worth more than many beginners think and less than they hope. Across all 78 pairs of suited and offsuit hands, being suited adds 2.8 percentage points of equity against one opponent and 3.9 against eight. Over many hands that is real money, but it does not turn a bad hand into a good one: 9♠ 3♠ is 126th of 169 heads-up and 114th against eight.

Even aces lose more often than they win in a crowd. A♠ A♥ takes 85.2% against one opponent and 34.6% against eight. That is still three times the fair share of 11.1%, so aces stay the best hand by a mile; they just stop being a sure thing. The practical lesson: raise with aces before the flop so that fewer players stay in.

The same hand is a raise from one seat and a fold from another. When you act first at a nine-seat table, eight players still speak after you, and any of them may hold something better. On the button only the two blinds are left. So the hands you play should depend on where you sit.

These are the hands we suggest opening with a raise when everyone before you has folded. We built them from this chart: the 169 hands sorted by their average place against one to eight opponents and taken from the top until each seat’s share was filled. The seats left out here (UTG+1, UTG+2) and a diagram of the table are in the guide to positions.

At our tables, count players rather than chairs. If only four players are dealt in at a nine-seat table, the first to act has three players behind him. That is a cutoff, not under the gun, and he can open about a quarter of his hands.

These 20 hands make up 126 of the 1,326 combinations, 9.5% of all deals. A player who plays only them folds about nine hands in ten — which is about as tight as you should be in the first seats at a full table.

• Playing any ace. A♠ 2♦ looks pretty and takes 54.9% of the pot heads-up, but against eight opponents it gets 10.4% — less than its fair share, 108th of 169. When it pairs the ace, someone else often holds an ace with a better kicker.
• Overrating suits. Being suited adds about three percentage points. It makes a good hand better; it does not make a bad hand good.
• Calling raises with hands you would not raise yourself. The chart is against random cards, and a raiser’s cards are not random. KJo or A9o look fine here, yet against a raise they often run into a bigger ace or king.
• Treating small pairs as strong hands. 2♣ 2♦ is barely above 50% against one hand and below its fair share against two to six. Its value is the 11.8% chance (about 1 in 8.5) of making a set on the flop, which only pays when the pot gets big.
• Slowplaying aces before the flop. 85.2% against one opponent, 34.6% against eight. Every player you let in for free takes a bite of your equity.
• Forgetting the seat. K♦ 10♣ is a fine raise on the button and a fold under the gun. Same cards, different seat.

We used the poker engine that deals the cards at our tables. For each of the 169 hands we took one representative — A♠ K♠ for suited hands, A♠ K♥ for offsuit ones — and played 2,000,000 random deals: eight opponent hands and a five-card board each time. Equity against k opponents uses the first k of those hands, so every deal counts in all eight columns. That is 338 million deals in total.

With two million deals, every figure is accurate to about ±0.1 percentage points with 95% confidence, so the first decimal means something. As a check, the textbook values come out exactly: aces 85.2% against one hand, ace-king suited 67.0%, ace-king offsuit 65.3%. The random seed is fixed, so running the calculation again gives the same table.