Skip to main content
Formal Specification

Animal Game (Lottery) — Formal Specification

Brazil · gambling-stake · Confidence: high

← Back to game page

Implementation Suitability

Suitability: adaptation-requiredFidelity: probabilistic-proxyGap-fills: 5
# Jogo do bicho — AI-optimized specification ## General idea An illegal but ubiquitous Brazilian numbers lottery in which bettors pick one of 25 animals and an operator (*bicheiro*) pays out on a draw. Structurally it is a one-shot wager against a house, with no strategic content whatsoever — the animal is a label on a number. ## Formal model - **Abstraction taken:** the draw is a uniform random selection over 25 animals. The real draw mechanism is not documented and, being an unregulated operation, is not reliably uniform in practice — the spec models the idealised version and flags the gap. - **Player/agent count:** any number of bettors, each independent, against one operator. - **Turn model:** single simultaneous betting phase, then one resolution. ## Formal rules R1. A bettor selects one animal from a fixed list of 25. R2. The bettor places a monetary stake with the operator. R3. A draw is held at a scheduled time. R4. [gap-fill] The draw selects one animal uniformly at random. The entry marks the draw mechanism, its frequency, and how the winning animal is chosen as undocumented; uniformity is the modelling assumption, not an attested rule. R5. Bettors holding the drawn animal are paid. R6. [gap-fill] The payout is a fixed multiple of the stake, configurable. The entry marks payout structure undocumented. R7. [gap-fill] Bets are independent; one bettor's outcome does not affect another's. R8. [gap-fill] A bettor may place at most one stake per draw. Sources do not say, and without a limit the model is unbounded. R9. [gap-fill] The operator always pays; counterparty risk, which is real in an unregulated game, is not modelled. ## Edge cases - **Tie at the end condition:** not applicable. Multiple bettors on the drawn animal all win independently; they do not divide a pot. - **Player with no legal move:** a bettor with no funds cannot stake and simply does not participate in that draw. - **Repetition / cycles:** draws repeat indefinitely by design. Each is independent, so there is no state cycle to break — only an unbounded sequence. - **Simultaneous claims:** all bets are placed simultaneously and resolved against one draw; no two bettors contend for anything. - **Mid-round elimination:** a bettor whose funds reach zero drops out; since bets are independent, no turn order needs repair. - **Resource exhaustion:** bettor funds are the exhausting resource, and a bettor with zero funds is out. - **Termination guarantee:** a single draw always terminates. A *sequence* of draws does not terminate on its own — bound it by a draw count or by every bettor's funds reaching zero, which happens with probability 1 whenever the payout multiple is below 25. ## State representation Persistent: the 25-animal list, payout multiple. Per-bettor: funds, current stake, chosen animal. Per-turn ephemeral: the drawn animal. Information set: the draw outcome is hidden until resolution; everything else is public. Note that in the real game the operator's draw process is not observable to bettors, which is the substantive information asymmetry. ## Action space - `place_bet(animal_id, amount)` — precondition: `amount <= funds`; no stake already placed this draw. - `abstain()`. ## Playability notes - **Single-player-vs-AI:** unsuitable as a strategy problem. Under R4's uniformity assumption every animal has identical expected value, so there is no choice to make. It is useful only as a bankroll-dynamics model. - **Group games (n>2):** bettors are independent, so n reduces to 1 without any loss of structure. - **Physical-skill model:** none applies; there is no skill in this game. - **What is lost:** the social reality. This is an illegal lottery embedded in Brazilian daily life, with animal symbolism, dream-interpretation traditions for choosing numbers, and organised-crime involvement. The entry carries a `gambling-stake` flag for good reason. The spec models the wager and none of the context, and it should not be mistaken for a description of the practice. ## Implementation hints There is no search or opponent modelling to implement. If you build this, build it as a bankroll simulation to show the expected loss rate, and make the payout multiple explicit — with 25 animals and a multiple below 25 the bettor's ruin is certain, which is the single most important property of the game and the one a naive implementation hides.

Download as JSON

Download the full game corpus entry for this game as structured JSON.

Download JSON
Cite this page

Jogo do Bicho (Brazil). (n.d.). Games of Countries. https://gamesofcountries.com/games/br-brazil/jogo-do-bicho

@misc{gamesofcountries-br-brazil-jogo-do-bicho,
  title = {Jogo do Bicho (Brazil)},
  year = {n.d.},
  publisher = {Games of Countries},
  url = {https://gamesofcountries.com/games/br-brazil/jogo-do-bicho}
}