Skip to main content
Formal Specification

Pitch and Toss — Formal Specification

Bosnia & Herzegovina · gambling-stake · Confidence: low

← Back to game page

Implementation Suitability

Suitability: engine-readyFidelity: probabilistic-proxyGap-fills: 2
# Čize-bliže — AI-optimized specification ## General idea A wagering and target-pitching game where agents compete to land a coin closest to a target line. Proximity determines rank priority in a subsequent waterfall coin-flipping phase, where higher-ranked agents flip the pooled coins and pocket all coins that land on the target face, passing any remaining unflipped coins to lower-ranked players. ## Formal model - **Abstraction taken:** Physical coin tossing to a 1D target line at distance Xtarget=4.0 m is modeled as a normal distribution Xland∼𝒩(Xtarget,σskill2). The coin-flipping resolution is modeled as independent fair Bernoulli trials with p=0.5. - **Player/agent count:** 2 to N agents (standard: N=4). - **Turn model:** Phase 1: Sequential throwing. Phase 2: Ranked priority waterfall resolution. ## Formal rules R1. **Setup:** Target distance D∗=4.0 m. Initial bankroll per agent Bi=10 coins. Ante per round A=1 coin. Round active pot size P=N. R2. **Phase 1 — Pitching:** Each participating agent i deducts 1 coin to fund pot P, then executes `pitch_coin(target_offset)`: - Landing position xi=D∗+offseti+ϵi, where ϵi∼𝒩(0,σi2) and σi=0.4⋅(1.1−skilli). - Distance to line: di=|xi−D∗|. - Agents are sorted in ascending order of di to form priority ranking R=[r1,r2,…,rN]. - [gap-fill] Exact ties in di are broken by a single random fair coin flip. R3. **Phase 2 — Waterfall Flip:** - Pot coins remaining Crem=P. - For each agent k∈R: - If Crem==0: Break loop (round over). - Agent k flips all Crem coins. Each coin lands HEADS with probability 0.5. - Number of HEADS obtained: Hk∼Binomial(Crem,0.5). - Agent k claims Hk coins: Bk←Bk+Hk. - Remaining pot updates: Crem←Crem−Hk. - [gap-fill] If Crem>0 after the lowest-ranked player flips, the remaining coins carry over to the pot of the next round. R4. **Elimination and match win:** Agents with bankroll Bi=0 are eliminated from subsequent rounds. The game ends when one agent holds all coins (or after a predetermined round limit Rmax⁡=20). ## Edge cases - **Tie at the end condition:** If match round limit Rmax⁡ is reached, the agent with the highest bankroll Bi wins; if tied, a sudden-death single pitch to the line breaks the tie. - **A player with no legal move:** Not applicable; pitching within valid bounds is always available while solvent. - **Repetition / cycles:** Coins are conserved and shift toward more skilled agents over successive rounds; the round limit Rmax⁡ prevents infinite cycling. - **Simultaneous claims on the same resource:** Handled strictly by the priority queue R established in Phase 1. - **Player elimination mid-round / turn order repair:** Bankrupted players are removed between rounds; surviving player ranking dynamically rebuilds each round. - **Resource exhaustion:** Total coins in the economy ∑Bi+Crem=N⋅10 is strictly invariant. - **Termination guarantee:** Each round strictly executes N pitch actions and at most N binomial evaluations. Match concludes either upon bankroll monopolization or at Rmax⁡=20. ## State representation - **Persistent state:** `bankrolls: Dict[AgentID, int]`, `round_number: int`, `pot_carryover: int`. - **Per-round ephemeral state:** `distances: Dict[AgentID, float]`, `priority_order: List[AgentID]`, `coins_in_pot: int`. - **Observable information:** Fully observable state; all coin counts, distances, and flip outcomes are public. ## Action space - `pitch_coin(target_offset: float)`: Parameter offset∈[−0.5,0.5] representing deliberate undershoot/overshoot strategy. ## Playability notes - **Single-player-vs-AI:** Natural as a competitive multi-agent wagering game with distinct AI risk profiles. - **Group reduction:** 4-player game reduces cleanly to 2 players (Duel) where Rank 1 flips 2 coins, and any tails are flipped by Rank 2. - **Physical-skill games:** Pitching accuracy is modeled by standard normal dispersion parameterized by skill. - **What is lost:** The micro-tactics of spin, surface bounce, coin weight friction against concrete, and cheating maneuvers (like spinning the coin flat) are abstracted. ## Implementation hints - Very simple probabilistic math; fast execution and highly suitable for Monte Carlo analysis. - Obvious trap: Not ensuring the total coin count across all bankrolls and pot remains constant. - Suggested first milestone: Implement Phase 1 ranking and Phase 2 binomial waterfall in a script to verify bankroll transitions.

Download as JSON

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

Download JSON
Cite this page

Pitch and Toss (Bosnia & Herzegovina). (n.d.). Games of Countries. https://gamesofcountries.com/games/ba-bosnia-herzegovina/cize-blize

@misc{gamesofcountries-ba-bosnia-herzegovina-cize-blize,
  title = {Pitch and Toss (Bosnia & Herzegovina)},
  year = {n.d.},
  publisher = {Games of Countries},
  url = {https://gamesofcountries.com/games/ba-bosnia-herzegovina/cize-blize}
}