Skip to main content
Formal Specification

Grand Chistera (Long Basket Basque Pelota) — Formal Specification

France · ball-team · Confidence: high

← Back to game page

Implementation Suitability

Suitability: engine-readyFidelity: discretizedGap-fills: 3
# Grand Chistera — AI-optimized specification ## General idea A high-speed 3v3 team rebound game on a large open court (60–80 m) where players use long curved wicker scoops (*chisteras*) to catch a 130g leather ball and sling it back against the front wall in one uninterrupted fluid motion. The team of three dynamically coordinates court coverage across forward, middle, and back zones to return volleys until one side fails to reach or return the ball. ## Formal model - **Abstraction taken:** Court space is discretized into a 2D coordinate grid with dimensions X∈[0,80] m (distance from wall) and Y∈[−10,+10] m (lateral width). Front wall height Z∈[0.8,12.0] m. Catching and slinging are modeled as a single composite action `catch_and_sling(target_z, target_y)` executed within a 1-second physical window without travel steps. - **Player/agent count:** 2 teams, 3 agents per team (total 6 agents on court). - **Turn model:** Alternating team possession during rallies; inside a team, the closest eligible agent intercepts the ball. ## Formal rules R1. Setup: Court length L=70 m, width W=20 m, front wall at X=0. Short service fault line at X=20 m. R2. Team roles: Each team of 3 consists of Avant (X∈[0,25]), Demi (X∈[20,50]), and Arrière (X∈[45,75]). R3. Service: Server initiates rally by slinging ball against front wall; ball must rebound beyond X≥20 m within lateral bounds |Y|≤10 m. R4. Return legality: The defending team must intercept the ball with a chistera before its second floor bounce. R5. [gap-fill] Action execution: The intercepting agent performs `catch_and_sling`. The catch succeeds with probability Pcatch=f(distance to ball,ball speed,agent skill). Upon catch, the ball must be released toward the front wall within Δt≤1.0 s without changing grid cell (no walking). R6. Front wall strike: The returned ball must impact the front wall at Z≥0.8 m without contacting the floor first. R7. Rally point award: A rally is lost and 1 point is awarded to the opponent if: - Ball bounces twice before intercept. - Catch fails (dropped ball). - [gap-fill] Excessive hold time (>1.0 s) or travel violation. - Ball strikes wall below Z<0.8 m or lands out of court bounds. R8. Turn initiative: The winning team of the rally scores 1 point and gains the right to serve the next point. R9. Match end: The first team to reach 40 points wins immediately. ## Edge cases - Tie at the end condition: Match ends the moment the 40th point is scored by either team. No deuce/two-point lead required. - A player with no legal move: If all three agents are unable to reach the ball before the second bounce, the turn times out and the point is automatically conceded. - Repetition / cycles: Bounded since each interception and throw has an error probability P(foul)>0, guaranteeing finite rally lengths. - Simultaneous claims on the same resource or position: [gap-fill] Zone priority applies: if ball lands in overlap region, the agent with lower X (closer to wall) has primary right of way unless called off. - Player elimination mid-round; turn order repair: Team rosters remain fixed; no mid-round player ejections. - Resource exhaustion (pieces, tokens, space, time): Single active ball per rally; court boundaries enforce space constraints. - Termination guarantee: Monotonic accumulation of points per rally ensures match reaches 40 points in finite time. ## State representation - Persistent state: Match points (TA,TB)∈ℕ2, current server ID. - Per-rally ephemeral state: Ball state (x,y,z,vx,vy,vz), floor bounce counter b∈{0,1,2}, active defending team, positions of all 6 agents {(xi,yi)}i=16. - Observable information: Fully visible state to all agents. ## Action space - `move(target_x, target_y)`: Moves agent within their positional zone (bounded by max sprint velocity). - `catch_and_sling(target_wall_z, target_wall_y, power)`: Intercepts ball and executes continuous whip-throw toward front wall coordinates. ## Playability notes - Single-player-vs-AI: Needs adaptation — single user controls entire team via automated teammate positioning AI, or plays 1v1 reduced-court variant. - Group games (n>4): 3v3 team structure is preserved by assigning deterministic zone-defense heuristics to non-user teammates. - Physical-skill games: Skill modeled via catch-reliability threshold and angular dispersion error on whip throws. - What is lost in the adaptation: The whip leverage generated by the 60cm chestnut basket, the auditory snap of wicker, and the physical exhaustion of 70-meter sprints. ## Implementation hints Discrete tick simulation (e.g. 50ms ticks) with agent positioning and ball trajectory physics. The zone-based agent assignment ensures straightforward pathfinding without intra-team collisions. First milestone: implement 3v3 rally simulation with automated zone positioning.

Download as JSON

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

Download JSON
Cite this page

Grand Chistera (Long Basket Basque Pelota) (France). (n.d.). Games of Countries. https://gamesofcountries.com/games/fr-france/grand-chistera

@misc{gamesofcountries-fr-france-grand-chistera,
  title = {Grand Chistera (Long Basket Basque Pelota) (France)},
  year = {n.d.},
  publisher = {Games of Countries},
  url = {https://gamesofcountries.com/games/fr-france/grand-chistera}
}