Spunk Odds Reference

Hi-lo guide

Jungle Hi-lo: infinite deck or one real deck?

We call this page "Jungle" for indexing reasons only. What it covers is a modelling choice that every hi-lo game makes, often without saying so: is the next card drawn from an endless supply of ranks, or from one physical deck?

Two ways to deal the next card

In the infinite-deck model, each new card is one of 13 ranks with chance 1/13 each, no matter what came before. Hash-based games that turn a random number into a rank with floor(float x 13) + 1 work this way. In the single-deck model, the cards come from one shuffled 52-card deck with four of each rank, and every card dealt is gone until the next shuffle.

The fresh-deck difference

Even before any card is dealt, the two models disagree slightly. In a fresh deck the face card itself is out of the pack, so only three cards of that rank remain among 51. Ties become a little less likely, and both higher and lower become a little more likely. With a 10 showing, higher is 23.08% in the infinite model and 23.53% from a fresh deck.

P(higher) for four face cards: infinite model against a fresh 52-card deck
Face cardInfinite modelFresh 52-card deckTie chance (infinite / deck)
284.62%86.27%7.69% / 5.88%
561.54%62.75%7.69% / 5.88%
930.77%31.37%7.69% / 5.88%
Q7.69%7.84%7.69% / 5.88%

After a few cards are gone

Once cards leave a real deck, the gap can grow in either direction. The table below keeps a 10 as the face card and assumes a Q, an A and a 6 were dealt earlier. Removing one high card and two low cards nudges the odds, but only by a little, because three cards out of 52 is a small share of the pack.

Face card 10: how the deck model changes the odds (seen cards: Q, A, 6)
Deck modelP(higher)P(lower)
Infinite deck (13 ranks)23.08%69.23%
Fresh 52-card deck23.53%70.59%
52 cards minus 3 seen22.92%70.83%

Why the model matters

A game's payouts are only fair to the stated edge if they match the deck model it really uses. A game that prices calls from the 13-rank table but deals from a real deck is slightly mispriced, and the error changes from card to card. A game that deals from an endless supply has no such drift. When you check a game's maths, find out which model it uses before you compare its multipliers with any table.