How to Play Ace-Deuce in Criss Cross Poker

The Across bet is the first of the three betting decisions, and it is the only one you make with no community card showing. That makes it the one decision in Criss Cross Poker that can be solved outright, and we have solved it: every one of the 169 starting hands was evaluated against all 19,600 three-card boards available to it. For an ace with a deuce the answer never moves off the minimum.

Only fourteen starting hands are raised to 3x: the thirteen pocket pairs and Q-J suited. Everything between the folds and the raises is a 1x, and A-2 lives there with about a hundred other hands. The full grid is on the printable strategy chart.

What you have committed at that point

At a $10 table the Across decision leaves $30 on the felt: the $10 Ante Across, the $10 Ante Down and the $10 Across bet. Both antes are posted before you see a card, so two of those three chips were never yours to withhold. The Down bet and the optional Middle bet are later decisions, made with community cards showing, and they are not part of the opening commitment.

Hold and you have no made hand. An unpaired ace is a high card, and a high card is paid nothing on any of the three lines. What you have is two ways to improve, and they are not worth the same.

• Pair the ace and you are paid. A pair of aces is a high pair, so the Across, Down and Middle bets pay 1 to 1 and each ante pays 1 to 1 as well. This is the outcome the hand is played for.
• Pair the deuce and you lose. A pair of deuces sits below the push band, so the line bet loses and the ante loses with it. Improving your hand and losing your money are the same event here.

That second row is the one people underrate. In most poker games any pair is progress. On this pay table a pair of deuces is a dead hand that has cost you a bet, and half of your starting cards can only ever make one. The deuce is not a bad card because it is small. It is a bad card because the pay table has a floor, and the deuce sits underneath it.

Two pair counts, and aces and deuces is two pair. A paired deuce is only dead while the ace stays unpaired.

Everything above 1 to 1 needs the board to do the work. A flush needs three more cards of your suit in one direction, which is only available to you if the two hole cards are suited, and a straight needs the board to bring exactly the three ranks that fill it. Those outcomes are real, and section 3 prices how little they add. The ante is a separate matter again: it pays 1 to 1 on a high pair or better, pushes on a low pair and loses below that, which is why an unimproved ace loses both antes without ever being beaten by anything. The complete pay tables

The correction first, because it is the point of this page. A-2 is the bottom tier of ace hands. It sits behind A-7 by 0.098 per unit on the Across bet despite holding a wheel draw that A-7 cannot make, because pairing a seven pushes while pairing a deuce loses. A-2 is not a connected hand with hidden potential. It is the weakest shape an ace comes in, and the wheel does not change that.

The usual case for A-2 runs the other way: the deuce is connected, A-2-3-4-5 is a straight, and A-7 can never make one at all. Every word of that is true and the emphasis is wrong. We evaluated every suited ace-x hand against all 19,600 three-card boards, and the results sort into four tiers. The tier is set by what happens when the kicker pairs, not by straight potential.

Read the table down the middle two columns and the point is unavoidable. A-7 has no straight available to it and beats A-2, which has one. The tiers track the last column, not the second: the hands whose kicker pairs into a winner are at the top, the hands whose kicker pairs into a push are in the middle, and the four hands whose kicker pairs into a loser are at the bottom on their own.

Both of the smaller figures come straight out of the same table. The straight route is worth about 0.044 per unit, which is the gap between A-10, with one straight and a kicker that pushes, and A-9, with no straight and a kicker that pushes. Being a dead card when it pairs costs about 0.143 per unit, which is the push-versus-lose penalty once that straight bonus has been accounted for. The straight is genuine, and it is roughly a third the size of the penalty it is being asked to offset. Subtract the one from the other and you get the 0.098 that separates A-2 from A-7.

How often the wheel actually arrives

A straight is possible with A-2, which is more than a pocket pair can say. The board has to bring exactly 3-4-5 in one direction, and a straight lands on 0.32% of boards. Three hands in a thousand, paying 5 to 1 when it comes.

Offsuit the picture is worse in the same shape. A-2 offsuit returns −0.376122 per unit on the Across bet, against −0.273265 for A-7 offsuit and −0.131224 for A-K offsuit. The ante is negative throughout as well: −0.458367 holding A-2 suited and −0.475102 holding A-2 offsuit, because the bare ace does not qualify and the paired deuce does not either.

None of this makes A-2 unplayable. A 1x Across bet still beats surrendering the ante, and section 1 sets out why. It makes A-2 an ordinary hand that you back for the minimum, rather than a hand you look forward to. Our figures come from our own exhaustive computation of this game; for the published house-edge analysis see Wizard of Odds. The game carries a 4.33% house edge on the ante, or 1.48% as element of risk, and nothing on this page changes that.

Your Across hand is your two hole cards plus the three cards of the horizontal row: left, centre and right. After the Across bet two of those three are exposed, so your row is settled apart from one card. Nothing you can do changes the Across bet now. What the reveal gives you is information, and with A-2 there are five cases worth telling apart.

Two blanks is the case you will meet most often, and it is the one to be honest about. An unimproved ace with a deuce beside it has one card left to hit and a pay table that gives nothing for high cards. That is the ordinary shape of this hand, and it is the shape the numbers in section 3 are averaged over.

The Down bet is made on the vertical column: your two hole cards plus the top, centre and bottom cards. Only the centre card is shared with the row, and it is still face down. So the column you are about to back is three cards you have seen nothing of, joined to the same two hole cards you started the hand with. Treat it as a fresh decision, because that is what it is.

Put the rows together and the practical answer is short. Your hole cards have not moved and the column is still three unseen cards, so the same reasoning that made the Across bet a 1x makes the Down bet a 1x with A-2. The exception is the ace that has already appeared in the row: you hold a paying pair, one of your outs is gone from the deck, and the column is being backed by a hand that is now worth less than it was, not more. Do not raise the column because the row is winning.

That is a limit on the arithmetic, not on the advice. What can be said with confidence is what each later card does to the hand you hold, and the next sections count those cards one by one rather than reaching for a percentage that does not exist. How the Middle bet is graded

The centre card is turned last, after every betting decision has been made. When you reach the Middle bet you have seen your two hole cards, the left and right cards of the row and the top and bottom cards of the column. Six cards are known and 46 are not. The centre card sits where the row and the column cross, so it lands in your Across hand and your Down hand at the same time, and it is the only card that can still change either of them.

That is why this decision does not need a fixed percentage. You can look at the four exposed cards, hold your A-2 against them and sort the 46 unseen cards into groups. Every centre card falls into exactly one of five.

• It pairs your ace. Three aces are unseen. An ace-high pair is the good outcome: a pair of aces pays 1x on the line it lands in, and it clears the ante requirement, so both antes are paid rather than lost. Because the centre card is shared, it does this in both hands at once.
• It pairs your deuce. Three deuces are unseen. A pair of deuces is below the low-pair band, so it loses on the line and loses the ante. This is the outcome that separates A-2 from the aces above it: the second card pairing is a result you do not want.
• It pairs a card already showing. A five-card hand is built from your two hole cards plus that line’s three community cards, so a centre card that matches an exposed card makes a pair inside the hand even though neither card is yours. Jacks through aces pay 1x, sixes through tens push, and anything lower loses. Read the exposed ranks before you decide what a pair there is worth.
• It completes a straight or a flush. With A-2 the only straight available is the wheel, A-2-3-4-5, and it needs the exposed cards of one arm to supply two of the three missing ranks. A flush needs a suited A-2 and four cards of your suit already lying in one arm. Both are rare, and neither is something to bet on before you can see it.
• It is a blank. Most of the 46 are. The line finishes as ace-high, which loses on the line and loses the ante. On the large majority of boards this is where A-2 ends up.

Two worked boards

The same hand, the same decision, two very different counts. Nothing here is estimated: you are reading four face-up cards and dividing 46 unseen ones.

Expected value on a betting line is linear in the stake. The ante is already posted and does not move, and each unit you add to the bet earns exactly the same amount as the unit before it. So the value of the whole Across decision is one number plus a multiple of another:

Read that formula and the strategy falls out of it. If the line EV is positive, every extra unit adds value, so you want as many units as the table allows: bet 3x. If the line EV is negative, every extra unit subtracts value, so you want as few as possible: bet 1x, or fold if even 1x finishes below the −1.000 that folding costs. A straight line has its best point at one end or the other. It never has its best point in the middle.

That is the whole case against 2x. It is never uniquely optimal. Whenever 2x looks reasonable, either 3x is better or 1x is better, and 2x is simply the halfway house between them. There is no board texture and no read that changes this, because there is nobody to read: the game is banked by the house against a fixed pay table, and no opponent is watching your bet size. Varying your stake buys you nothing except variance.

A-2 through the formula

Suited A-2 has an Across line EV of −0.295969 per unit and an ante EV of −0.458367. Substituting each stake into the formula gives the full ladder, and the shape of it is the argument:

The order is 1x, then fold, then 2x, then 3x, and it is strictly monotonic once the fold is set aside. 1x beats folding by 0.245664 of a unit, which is why A-2 is played rather than thrown away. 2x sits exactly halfway between 1x and 3x because that is what a straight line does, and 3x costs a third of a unit more again. Offsuit A-2 runs the same ladder from a lower start: a line EV of −0.376122 and an ante EV of −0.475102 put 1x at −0.851224 and 3x at −1.603468.

One caution about applying the formula further down the hand. The Across decision is the one we have solved exactly, because you make it knowing only your own two cards. The Down and Middle decisions come after cards have been revealed, so their value depends on what is showing and cannot be reduced to a single number. The linearity argument still holds at every decision, and it is the part that carries: pick the end of the range, not the middle.

What passing actually costs

Passing costs you exactly one thing: the bet you did not make. Nothing else. If the centre card turns over and your better hand is a straight, you do not collect 5x on a Middle bet you never placed, and that is the entire penalty. Your Across bet still collects its 5x. Your Down bet is still paid on whatever the column made. Both antes are still settled on their own terms. The Middle bet is a separate, self-contained wager on the same two hands, and declining it leaves the rest of the hand untouched.

This is worth being precise about, because the two things people mix up look similar and behave very differently:

• Folding Across or Down forfeits that line’s bet and its ante. That is real money already on the felt, and it is the reason the fold decision is scored against −1.000 rather than zero. The other line is not affected and is paid as normal.
• Passing the Middle forfeits nothing at all, because there is no Ante Middle to lose. You simply do not have that wager working. It is a decision to add money or not, taken with four community cards already face up.

One structural point sits underneath all of this. The Middle bet is only offered while at least one of Across and Down is still live. If you folded both, it is not available, and only the Five Card Bonus can still pay. So the question of passing the Middle only ever arises when you have money working on the board already.

When to make it, and for how much

The Middle bet is graded on whichever of your two hands finishes better, which makes it the most forgiving of the three: it takes the best of two chances rather than one. It is also the only decision you make with four of the five community cards showing, so read them before you decide. The sizing rule from the previous section still applies, so the real choice is 3x, 1x or pass.

With A-2 specifically, you will pass the Middle often, and that is a normal outcome rather than a mistake. The hand is a 1x opener on both lines, so the top row of that table is where you usually are. Passing keeps your exposure down without costing you a thing on the money already committed. How the Middle bet is settled

A suited A-2 can make a flush and an offsuit A-2 cannot, so the two versions are genuinely different hands. The gap is measurable, and we measured it across all 19,600 three-card combinations for each.

Sharing a suit is worth 0.080153 per unit on the Across bet and another 0.016735 on the ante, a little under a tenth of a unit in total. That is a real edge and it is not noise. It is also nowhere near large enough to move the decision, because both versions sit in the same place: negative on the line, so the minimum rather than the maximum, and comfortably above the −1.000 of folding, so played rather than thrown away.

Two things follow from that. The first is that you should not treat a suited A-2 as an upgrade worth chasing: it is the same minimum bet, made for the same reason. The second is that the flush route only starts to matter after the reveal, when you can see whether an arm is actually running your suit. Four cards of your suit in one line, with the centre card still face down, is a live flush draw and worth counting. Two cards of your suit is scenery.

It is also worth noting where the flush route puts A-2 relative to hands that have none. A pocket pair cannot make a flush or a straight, because two cards of one rank leave at most four distinct ranks and sit in two different suits. A-2 can make both. It is still the weaker holding by a wide margin, which is a good reminder that counting the ways a hand can improve is not the same as measuring how much it is worth. The full hand ranking ladder