How to Play Flush Draws in Criss Cross Poker

Criss Cross Poker deals you two hole cards and lays five community cards in a cross. Your Across hand is your two hole cards plus the top, centre and bottom cards. Your Down hand is your two hole cards plus the left, centre and right cards. There is no choosing and no discarding: both hole cards are in both hands, every time. That single rule decides what a flush draw is.

A five-card flush needs five cards of one suit. Two of them are always your hole cards, so your hole cards must be suited. The centre card is the one you are drawing to. That leaves the two outer cards on the row, and both of them have to match as well. So the whole requirement is: two suited hole cards plus two outer cards of the same suit on the same row, with the centre card still to come.

The layouts that qualify

These are all of them. If your position is not in this table, you are not on a flush draw.

Four positions that look like draws and are not

Every one of these turns up regularly, and each one costs money if you price it as a draw.

At the Middle decision six of the seven cards in play are face up: your two hole cards and the four outer community cards. One card, the centre, is still down. Six known cards leave 46 unseen, and exactly one of them lands in the centre. Every one of those 46 is equally likely, which is why nothing here has to be estimated.

Nine nominal outs

Each suit has thirteen cards. On a flush draw you can already see four of them: your two suited hole cards and the two matching outers on your row. Thirteen minus four leaves nine cards of your suit still in the deck, and any of them completes the flush.

Then take off the opposite row

The two outer cards on the other row are face up too, and they came out of the same deck. Every card of your suit sitting over there is a card that can no longer arrive in the centre. Take one off your count for each one.

Hold with on the Across row and you have nine nominal outs. If the Down row shows one diamond, you are on eight. If it shows two, you are on seven. That is the whole adjustment, and it is the step most players skip.

On the cross itself the opposite row only holds two cards, so nine, eight and seven are the counts you will actually meet. Six and five are in the table so you can see where the line falls, and section three shows exactly where that is.

The Middle bet is settled against the pay table, not against a dealer. There is no qualifying hand and nothing to beat. A flush there returns 8 units for every 1 staked, and a miss loses the 1 unit. That is all you need to find the point where the bet stops costing you money.

The working, in plain arithmetic

Call the chance of hitting p. You win 8 units with probability p and lose 1 unit with probability 1 minus p. Break-even is where those two are equal.

Anything above 11.1111% and the Middle bet makes money. Now put the out counts from section two next to it.

Seven to nine live outs, which is every flush draw you will actually be dealt on the cross, clears the line easily. The margin at nine outs is more than eight percentage points.

What that is worth in money

These next two tables price the same draws in dollars. They assume the harshest possible reading of the hand: every centre card that is not your suit loses the bet outright. That is not true in practice, and section four shows how much it understates the draw, but it is the safe number to check a call against.

Note the last row of both tables. Five live outs is a losing bet by a whisker under this simplified model, and it is the only row that is. The next section shows why even that reading is harder on the draw than the game actually is.

Section three priced the draw the cautious way: flush or nothing. That is a useful check because it can only understate the hand, but it is not what happens at the table. When a centre card arrives that is not your suit, the row does not simply die. It might pair one of your hole cards. It might fill a straight you were not counting. And the other row is still being graded at the same time, because the Middle bet is settled on the better of the two completed rows, and the centre card belongs to both.

How the engine prices it

This site’s coach values the Middle bet by enumeration. It takes the six known cards, deals each of the 46 remaining cards into the centre in turn, evaluates both completed rows for every one of them, and takes the payout on whichever row finishes higher. Averaging those 46 outcomes gives an exact expected value per unit staked. Nothing is sampled and nothing is rounded on the way through, because with a single unknown card there is nothing left to approximate.

Here are four flush-draw layouts priced both ways. The middle column is the simplified model from section three. The right-hand column is the engine.

Read the gap between the columns

The rescues are worth roughly 50% more on the plain nine-out draw, taking it from +0.760870 to +1.130435. They more than double the seven-out double flush draw, from +0.369570 to +0.891304. And a draw reduced to seven live outs by two diamonds on the far row still lands at +0.760870 with rescues counted, which is exactly what an untouched nine-out draw was worth under the simplified model. Losing two outs to the opposite row costs you less than the flush-or-nothing arithmetic suggests.

The last row is in a different league entirely. A 5-6-7-8 of spades layout is worth +5.826087 per unit because two of its outs pay 100 to 1: the four of spades and the nine of spades each make a straight flush. Every other spade still makes a flush at 8 to 1, and every non-spade four or nine still makes a straight at 5 to 1. Almost nothing in that row is a clean loss. Section six takes the layout apart card by card.

Because both hole cards always play, a suited holding is attached to every row at once. If the Across row and the Down row each show two more cards of your suit, then both rows are four cards to a flush and both of them are waiting on the same centre card. Take as the hole cards, with 3♠ J♠ on the Across row and 8♠ K♠ on the Down row.

Count the suit once, not twice

The instinct is to add nine outs to nine outs and call it fourteen or eighteen. That is the one mistake this layout invites. There are thirteen spades in the deck. Six of them are already face up or in your hand: the ace and the six you hold, the three and the jack across, the eight and the king down. Thirteen minus six leaves seven live spades, and seven is the whole count for the layout, not seven per row.

Seven live outs out of 46 unseen cards is 15.22%, which clears the 11.1111% break-even the pay table demands. So the draw is playable. What it is not is a fourteen-out monster.

What the second row actually buys you

If a double draw does not add outs, why is it better than a single draw? Because it removes the ways the layout can miss badly. With four spades on each side, a blank centre card still leaves two rows of live cards rather than one, and the Middle bet is settled against the better of the two. Two chances at a pair, at a high card that pairs, or at any other rescue is worth more than one, even though the flush itself has only seven ways in.

That is exactly what the engine finds. Priced the simple way, where every non-spade centre card is treated as a total loss, seven outs is worth +0.369570 per unit staked. Priced properly, by dealing each of the 46 candidate centre cards in turn and taking the better completed row, the same layout is worth +0.891304. The second row does not double the outs. It more than doubles the value.

Both numbers say bet. The difference between them is the difference between knowing that a spot is playable and knowing how much it is worth, and it is the reason the simplified table in section 3 should be read as a floor rather than a valuation.

Suppose you hold and one row shows 5♠ and 6♠. Four cards of the same suit in sequence:

The centre card can finish this row in three different ways, and the pay table treats them very differently. A straight flush pays 100x. A flush pays 8x. A straight pays 5x. Two physical cards, the 4♠ and the 9♠, deliver the top result.

Count each physical card once

This is where the count goes wrong. There are nine spades left in the deck, because four of the thirteen are already showing. Two of those nine are the 4♠ and the 9♠, which means only seven spades remain that make a plain flush. Likewise there are eight fours and nines left in the deck, but two of them are spades, so only six of them make a plain straight.

The rule generalises to every layout on this page. Deal each of the 46 unseen cards in your head, ask what the row becomes, and put the card in one bucket only. The buckets must add up to 46. If they add up to anything else, you have either double counted a card or forgotten one that is already face up.

What it is worth

Priced across all 46 candidate centre cards with both rows evaluated, this layout returns +5.826087 per unit staked at the Middle. That is not in the same league as the other draws on this page, and the reason is arithmetic rather than luck: two of the outs pay 100 to 1. A single 100x card contributes more to the average than every flush out combined.

The practical consequence is short. When you can see four suited cards in sequence on a row, you are not deciding whether to bet the Middle. You are only deciding to bet it at the maximum, which is covered in the next section.

Everything above has been about a draw you already own: two suited hole cards, two more of the suit on a row, a centre card to come. The Across decision happens earlier than that, when the only thing you know is your own two cards. At that point a suited holding such as 5♠ 4♠ has flush potential and nothing else, and the question is whether that potential pays for the bet.

Our exhaustive solve over all 169 starting hands says it does not. Every unpaired 2-5 hand folds, suited included. Folding is not free, because you forfeit the ante as well as the bet, but its cost is fixed and known: exactly −1.000000 unit. Betting 1x costs more than that in every case.

The arithmetic is a subtraction you can check yourself. Take the best of the three, 5-4 suited: 1.152908 − 1.000000 = 0.152908. Playing it for 1x gives away about fifteen hundredths of a unit against simply folding, every time you do it. The weakest tier gives away 0.213724, which is more than a fifth of an ante.

Suitedness is worth something, just not this much

Being suited is not a rounding error. It is the difference between a hand that can make a flush and a hand that cannot, and the engine prices it. Compare the line-bet expectation of each suited hand against the same ranks offsuit:

So suitedness buys roughly 0.085 to 0.093 per unit for these hands. Set that against the 0.15 to 0.21 the hands need in order to beat folding and the verdict writes itself. Suitedness closes roughly two fifths to three fifths of the gap. It never closes the whole of it, and it is not close enough for the decision to be marginal.

Why the two things are not in conflict

There is no contradiction between this section and the rest of the page. Flush equity is enormously valuable at the Middle, where you are looking at four cards of a suit, one card to come, and a wager that has no ante attached to it. A nine-out draw priced properly is worth +1.130435 per unit. At the Across decision you have two cards, five cards to come, and an ante already at risk. The same suit is doing a fraction of the work.

The order matters. Flush equity is a reason to bet the Middle. It is not a reason to open a hand you would otherwise have folded, because by the time the flush matters you have to have survived a decision that suitedness alone cannot pay for.

Low card hands in full

The printable strategy chart

First, the size. The maximum Middle wager is 3x the ante. It is not an all-in and it is not capped by what you have in front of you. At a $25 ante the largest Middle bet available is $75, which is why the tables on this page price everything per unit staked and then multiply.

Why 1x is never optimal

The Middle is the only wager on the layout with no ante behind it. The Across and Down bets each sit on top of an ante you posted before the deal, so declining them still costs you that ante. The Middle costs nothing at all if you leave it alone. Passing is exactly 0.000000.

Second, the value of the Middle is linear in the stake. There is no bonus for betting more and no discount for betting less. A layout worth +1.130435 per unit is worth three times that at 3x, and a layout that loses on average at 1x loses three times as much at 3x. Put those two facts together and the whole sizing decision collapses:

A 1x Middle bet on a good draw leaves two thirds of the value on the table. A 1x Middle bet on a bad draw takes a third of a loss you were free to avoid entirely. There is no count, no suit and no row that makes the middle sizes correct. This is the same conclusion the straight-draw guide reaches from the other direction, and for the same reason.

How to play straight draws

What you already staked is not part of the count

The commonest way to get this wrong is to let the Across and Down bets speak. Those chips are already committed. They will be settled against the same seven cards whether or not you bet the Middle, and nothing you do now can retrieve them. A big Across bet is not a reason to add a Middle bet on a five-out draw, and a folded row is not a reason to skip a Middle bet on a live one.

The eight-step check

Run this in order, before the chips go out. It takes about ten seconds once it is familiar, and it catches the three mistakes that cost the most: the double-counted card, the dead out on the opposite row, and the fourteen-out illusion.

1. 1 Check your two hole cards are the same suit. If they are not, there is no flush draw for you on either row, whatever the outer cards show.
2. 2 Find a row with two more cards of that suit. Four of a suit on one row is the only shape that a single centre card can complete.
3. 3 Start the count at nine. Thirteen in the suit, minus the four you can already see, leaves nine nominal outs.
4. 4 Subtract every card of that suit sitting on the opposite row. Those cards are dealt and gone, and each one takes an out with it.
5. 5 Look for a straight-flush overlap. If the four suited cards are also in sequence, two of your outs pay 100x and the layout is worth far more than the flush count alone suggests.
6. 6 Check the other row. If it is also four to the flush, you share one set of outs, not two, but the extra rescues are worth real money.
7. 7 Compare the live count with break-even. Five outs is 10.87%, just under the 11.1111% the 8-to-1 pay table requires. Six and up clears it.
8. 8 Bet 3x or pass. Do not compromise at 1x or 2x, and do not let the Across and Down chips into the decision.

Every count you make should be checkable against 46. Sort the unseen cards into buckets, put each card in exactly one, and add the buckets up. A total that is not 46 means you double counted a card, missed one that is already face up, or forgot that the centre card is dealt once into the middle of both rows.