Formal Specification
Sardinian Wrestling — Formal Specification
Italy · wrestling-combat · Confidence: high
← Back to game pageImplementation Suitability
Suitability: engine-readyFidelity: probabilistic-proxyGap-fills: 4
# S'Istrumpa — AI-optimized specification
## General idea
A two-agent stand-up grappling contest where both agents begin locked in a symmetrical upper-body chest-to-chest grip. Agents simultaneously select balance-breaking techniques (sweeps, hip throws, leg hooks, lifts, counters) across discrete round exchanges, where leverage matchups and endurance stats determine whether a clean back/shoulder takedown (*atterramento*) is scored.
## Formal model
- **Abstraction taken:** Continuous physical pressure, dynamic weight distribution, and foot positioning are abstracted into discrete grappling rounds. In each round, both agents simultaneously select a technique from an enumerated martial set (Sweep, Hip-Throw, Leg-Trip, Lift, Defensive-Brace, Counter).
- **Player/agent count:** 2 agents in a 1v1 duel.
- **Turn model:** Simultaneous selection per exchange, followed by stochastic resolution. Best of 3 falls (first to 2 points).
## Formal rules
R1. State setup: Both agents start with `falls` , `fouls` , and `stamina` . Grip state is initialized as locked.
R2. In each exchange , both agents simultaneously choose an action .
R3. Technique outcome resolution:
- [gap-fill] An interaction payoff matrix determines the base leverage advantage .
- [gap-fill] Takedown probability is evaluated: .
- Outcome: Either Agent 1 scores a fall, Agent 2 scores a fall, or no fall occurs (neutral clinch).
R4. Fall registration: A successful throw awards fall to the attacker. Both agents reset to the standard standing grip with stamina partially restored.
R5. Foul rules: [gap-fill] If an agent attempts an illegal strike or grip release, they receive foul warning. Accumulating 3 fouls forfeits the current round/fall.
R6. Win condition: The first agent to achieve 2 falls wins the match immediately.
R7. [gap-fill] Maximum exchange limit: If no agent reaches 2 falls after exchanges (equivalent to 3 minutes), the agent with more falls wins. If falls are equal, the agent with fewer fouls wins; if still tied, higher remaining stamina wins.
## Edge cases
- Tie at the end condition: Resolved deterministically by tie-breaking hierarchy (Falls Fouls Stamina) as specified in R7.
- A player with no legal move: Not applicable; the 6 combat actions are always available.
- Repetition / cycles: Bounded strictly by the 30-exchange round clock limit in R7.
- Simultaneous claims on the same resource or position: Both agents commit actions simultaneously; R3's resolver explicitly computes joint outcome probabilities, preventing ordering conflicts.
- Player elimination mid-round; turn order repair: In tournament brackets, the defeated agent is eliminated and the winner advances.
- Resource exhaustion (pieces, tokens, space, time): Stamina depletes with each high-effort technique; bounded by round time limit.
- Termination guarantee: Clock monotonicity () guarantees that every bout terminates in at most 30 steps.
## State representation
- Persistent state:
- `falls_1`, `falls_2`: integers .
- `fouls_1`, `fouls_2`: integers .
- `stamina_1`, `stamina_2`: floats .
- `exchange_count`: integer .
- Per-exchange ephemeral state:
- `action_1`, `action_2`: chosen technique IDs (private until reveal).
- Observable information: Public fall counts, foul warnings, stamina gauges, and past action history. Current round intent is private.
## Action space
- `select_technique(action_type: str)`:
- `action_type` .
## Playability notes
- Single-player-vs-AI: Engine-ready; standard rock-paper-scissors dynamic with stamina resource management and opponent action frequency tracking.
- Group games (n>4): Contested as standard single-elimination or round-robin tournament brackets.
- Physical-skill games: Physical balance and torso leverage are abstracted to the technique interaction matrix and skill parameters.
- What is lost in the adaptation: Continuous bodily friction, micro-adjustments of grip fingers, and facial cheek contact.
## Implementation hints
Lightweight 2-agent simultaneous zero-sum combat model.
First milestone: Build the technique payoff matrix, stochastic throw resolver, and best-of-3 fall manager before adding stamina decay curves.
Download as JSON
Download the full game corpus entry for this game as structured JSON.
Download JSONCite this page
Sardinian Wrestling (Italy). (n.d.). Games of Countries. https://gamesofcountries.com/games/it-italy/s-istrumpa
@misc{gamesofcountries-it-italy-s-istrumpa,
title = {Sardinian Wrestling (Italy)},
year = {n.d.},
publisher = {Games of Countries},
url = {https://gamesofcountries.com/games/it-italy/s-istrumpa}
}