7s, 8s, and 9s

A hi-lo split poker game in which twelve cards in the deck are natively wild, additional wild cards can be constructed by combining two ordinary cards, and the bulk of strategic action takes place before the first chip is ever bet — during four scripted passing rounds that decide what every player's final hand will be.

The game has many names. We call it 7s, 8s, and 9s because three of the four points of leverage in the game derive from those three ranks. We sometimes call it the Passing Game because the betting is, in a meaningful sense, a sideshow: the strategic heart of the game is the four-round distribution mechanism that precedes any wager. By the time players have antied in, exchanged cards through the structured passes, and chosen their declarations, the hand has often already resolved itself in everyone's head — the betting round is simply a vehicle for capitalising on that conviction.

The variant is dealer's-choice country; we have never seen it played in a casino. The version documented here is the version we have simulated and analysed, with the rules tightened where folk transmission left ambiguities. The findings page covers what we have learned from running it under a wide spread of agent archetypes.

Setup

Six or seven players. Standard 52-card deck. Each player begins with a stack of 500 tokens; we recommend chips in 5- and 10-token denominations plus a handful of single tokens for the declaration phase. There is no dealer button in this variant — the dealer rotates, but acts in turn like any other seat.

Before dealing, every player antes 5 tokens into the pot. With seven players this seeds a 35-token pot — meaningful but not yet decisive. After the ante, seven cards are dealt face-down to each player. There are no community cards. There is no further drawing once the deal is complete.

The Wild Matrix

The defining feature of this game is the unusual density of wild cards. There are three independent ways a card can become wild, they accumulate, and they are responsible for both the game's high ceiling (Five of a Kind is the routine winning high hand) and the deep strategic texture of the passing phase (because most cards in your hand have multiple potential roles, deciding what to pass is rarely obvious).

1. Native wilds — the 7s, 8s, and 9s

Every 7, 8, and 9 in the deck is a wild card. That is twelve cards in total — almost a quarter of the deck. A wild card can stand in for any other card (any rank, any suit) at the moment of resolution. The same card need not commit to a particular identity until showdown: the 7 of clubs in your hand can be the third ace in your high hand and the missing six in your low hand only if you are competing in both directions, but for any single resolution it must be exactly one specific card.

2. Additive wilds — same-suit pairs summing to 8

Two cards of the same suit whose ranks sum to exactly 8 may be combined into a single wild card. The two cards still occupy two of your seven hand slots, but at showdown they jointly play as one wild — you lose one card from your effective hand in exchange for the flexibility a wild card brings.

Because each suit contains exactly one of every rank, only two combinations satisfy this for any suit:

3. Subtractive wilds — opposite-colour pairs differing by 8

Two cards of different colours (one red, one black) whose ranks differ by exactly 8 may also be combined into a single wild card. Five such pairings exist, and any cross-colour combination counts (a heart and a club, or a diamond and a spade, etc.).

Composite wilds — both additive and subtractive — are constructed only from non-native cards. The native 7s, 8s, and 9s are already wild and cannot also participate in a composite.

The optionality rule

Consider an ace and a six of opposite colours. As a subtractive wild, this pairing produces one all-purpose wild. As two ordinary cards, it produces an ace (the lowest card available for low hands) and a six (the ceiling of the six-four). For a player competing for the low pot, leaving them uncombined is almost always strictly better.

The Passing Phase

Once the deal is complete and antes are in, the table executes four consecutive passing rounds. Each round is simultaneous: every player selects the cards they intend to pass before any cards are exchanged, then all exchanges happen at once. There is no speaking or signalling. The order of passes, and the seat-to-seat routing, is fixed.

The structure is deliberate: by varying both the count and the routing of each pass, every card in the deck circulates broadly before settling. At a seven-player table the four-pass sequence distributes cards to four different seats across the four rounds, so no two players' hands are even close to independent by the time betting begins.

Each player still has exactly seven cards in front of them after each pass — passes are net-zero. Hands are reshaped, not grown or shrunk.

The information you accumulate

Although players cannot see anyone's hand, every player has full visibility into the cards they themselves received, by seat. Over four rounds you accumulate Bayesian evidence about what your sender was building. If seat 3 sends you nothing but fives, sixes, and threes in round one, they are almost certainly not building a low hand. If they then send you a queen and a king in round two, the inference reverses. Strong players treat the incoming pass as the most important information in the game.

Reveal and Betting

Once the four passes are complete, each player privately arranges their seven cards into the order they wish to reveal them. The seventh card — the last in their ordering — is never revealed during play and stays hidden until showdown. The other six are turned over progressively, with a betting round between each flip.

Six (REVEAL → BET) cycles follow. In each cycle, every non-folded player simultaneously turns over the next card in their ordering, exposing it face-up to the table. A betting round then begins to the left of the dealer (and rotates one seat clockwise on each subsequent round):

• The opening bet and all raises are in 10-token increments.
• The bet level is capped at 25 tokens per player per round — no single player may put more than 25 additional chips in during one of the six betting rounds.
• No check-raise. A player who checks (when the action returns to them at no cost) may not subsequently raise in the same round. Their remaining options are check, call, or fold.

After six reveal-bet cycles, every active player has shown six of their seven cards face-up. Declaration follows. The hidden seventh card is what gives Piggy declarations their final dose of risk: you have watched everyone's hands take shape for six rounds, but you have not seen the card they reserved.

The Declaration Phase

After betting, each non-folded player makes a simultaneous, blind declaration of which side of the split pot they are competing for. The mechanism is physical: each player takes the appropriate number of small tokens (we use 1-token chips for this purpose) into a closed fist beneath the table, and on a count of three all fists are revealed.

There is no order to declarations; players cannot use the information of others' declarations to adjust their own. This is structurally important. The blindness of the declaration is what keeps the Piggy declaration genuinely risky.

Resolution

Hands are revealed and evaluated according to the standard seven-card-best-five rule, with one wrinkle: each player decides privately, at the moment of resolution, which (if any) of their available wild-card combinations to activate.

The high hand

The high hand follows ordinary poker rankings, extended upward to include Five of a Kind, which is achievable thanks to the wild cards. The hierarchy, from best to worst:

1. Five of a Kind
2. Straight Flush (including royal)
3. Four of a Kind
4. Full House
5. Flush
6. Straight
7. Three of a Kind
8. Two Pair
9. One Pair
10. High Card

Because of the wild density, Four of a Kind is not particularly impressive in this game. A typical winning high hand is Five of a Kind; a Four of a Kind that does not contain an ace as a kicker is frequently in trouble.

The low hand

The low hand is ace-to-six lowball: aces play as 1, and straights and flushes count against a low hand. The best possible low is therefore A–2–3–4–6, universally called a six-four at the table. (The 5 is excluded because A–2–3–4–5 is a wheel straight, which disqualifies the hand as a clean low.) "I have the six-four" is the only thing anyone says when they have the perfect.

Low-hand classes, from best to worst:

1. No-pair, no straight, no flush (clean low)
2. One pair
3. Two pair
4. Three of a kind
5. Straight (no flush)
6. Flush (no straight)
7. Full house
8. Four of a kind
9. Straight flush

A clean low always beats a paired low. Within the clean class, hands are compared from the highest card down; lower wins.

The Piggy rule

The Piggy bust rule is the central engine of strategic tension in the game. A player with a strong-but-not-overwhelming hand has the constant temptation of declaring Piggy — doubling their upside — while risking total loss if even one tie occurs. A conservative player can sometimes profitably declare a single direction with a marginal hand purely to create a tie that busts an opponent's Piggy.

The Odd Chip rule

Because the pot is split 50/50 between high and low, a pot that contains an odd 5-token increment cannot be split exactly. In that case, the surplus 5 tokens go to the high hand winner. (If the pot is itself a multiple of 10, no Odd Chip applies; this rule only matters when ante and bet arithmetic produces a remainder.)

What happens when no one competes for a direction

Occasionally every non-folded player declares the same direction. In that case the entire pot is awarded to the best hand in the declared direction, with the unclaimed half simply rolling over. If a side ties, that side's portion is split evenly among the tying claimants.

Worked Example

Suppose at a seven-player table you find yourself holding, after the four passes are complete, the following hand:

A♠   2♥   3♦   7♣   8♥   Q♠   K♣

You have two native wilds (the 7 and 8) plus an ace-low, two-of-a-kind-like-ace lowering potential. Your wild matrix inspection turns up one composite candidate: the A♠ and K♣ are opposite colours and differ by 1 — not a subtractive wild. No additive combinations exist either, because no same-suit pair in your hand sums to 8. So you have only the two native wilds.

Now look at the cases:

• If you declare High: your five non-wild ranks are all distinct (A, 2, 3, Q, K), so you cannot reach Four of a Kind or higher — that would require at least two non-wilds sharing a rank, since with two wilds you can fill only two of the four "same rank" slots. What you can reach is an A–K–Q–J–10 broadway straight (the wilds become the missing J and 10). A decent hand, but Five of a Kind and Four of a Kind together account for almost two-thirds of all winning High hands at a seven-seat table, so a broadway straight is frequently second-best in this game.
• If you declare Low: your wilds become the missing 4 and 6, giving you the six-four. This is essentially uncontested unless another player also has it — in which case you split, which is fine.
• If you declare Piggy: you commit to both. The same two wild cards play as J and 10 on the high side and as 4 and 6 on the low side — they can do double duty because high and low evaluate independently. Your high (broadway straight) is good but not premium; your low (the six-four) is perfect. The risk is twofold: a Piggy must scoop both sides outright, and (a) someone else may have a stronger high than a broadway straight, and (b) the six-four is commonplace enough that ties on the low side are routine. Either condition voids the Piggy.

The correct declaration depends on the betting history (did anyone signal a low build, did anyone signal a Five build?) and on what was passed to you (did anyone telegraph holding aces or sixes?). On most tables the straight Low declaration is the safest play here.

Further reading

• The Five Archetypes — the agent personalities we developed to study the meta-game, in approximate order of how often they appear at our table.
• Findings — empirical results — what the simulator has told us so far. Currently in progress; placeholder until the first batch of analyses completes.
• About the project — what we are looking for, and what counts as a finding.