Formal Specification
Bare-hand Basque Pelota — Formal Specification
Spain · ball-team · Confidence: low
← Back to game pageImplementation Suitability
Suitability: adaptation-requiredFidelity: discretizedGap-fills: 3
# Esku huskako pilota — AI-optimized specification
## General idea
A two-agent or two-team wall-rally ball game where agents alternate striking a ball against a shared front wall above a foul line. Each rally alternates possession, requiring the active agent to reach the ball before its second floor bounce and execute a valid wall return. Strategic play revolves around shot placement (angles off the side wall, deep rebounds, drop shots) and court positioning.
## Formal model
- **Abstraction taken:** The continuous 3D aerodynamics and hand-striking biomechanics are discretized into a 2D court grid with positional zones (*cuadros* 1–10). Shot outcomes (landing zone, wall rebound, and error probability) are resolved via a stochastic skill-and-position resolver.
- **Player/agent count:** 2 individual agents (singles) or 2 teams of 2 agents (forward/back).
- **Turn model:** Sequential alternation of returns during an active rally; point-by-point progression.
## Formal rules
R1. Setup: The match starts at score 0–0. A coin toss designates the initial serving agent (). Target score is 22 points.
R2. Service: Server initiates the point by choosing a serve action from . The serve outcome lands in the legal zone . If , fault is scored (point to ). If , one re-serve is granted; a second long serve awards the point to .
R3. Rally return: After a legal serve, the non-serving side becomes active. The active agent must select a valid return shot before the ball completes 2 floor bounces.
R4. Shot execution: An agent's return is resolved with . A successful shot impacts the front wall above height and rebounds into legal court space.
R5. Error scoring: An agent fails the point if its return impacts on/below the foul strip (), lands outside court bounds, or fails to reach the front wall before 2 floor bounces. The opposing side receives point.
R6. Service retention: The agent/team winning the point serves the next point.
R7. [gap-fill] Obstruction resolution: If both agents occupy the exact same discrete zone during a stroke, an *estorbo* (let) is declared and the point is replayed without score change.
R8. Win condition: The first agent/team to reach exactly 22 points wins the match immediately.
## Edge cases
- **Tie at the end condition:** Not applicable; play proceeds point-by-point until one agent reaches 22 first. No deuce/two-point advantage rule exists in traditional regulations.
- **A player with no legal move:** Not applicable; on each active turn, an agent always has a set of physical shot choices (drop shot, deep lob, alley drive).
- **Repetition / cycles:** Bounded by stochastic error rate per stroke (), ensuring every rally terminates in finite strokes almost surely.
- **Simultaneous claims on the same resource or position:** Resolved by R7 ( [gap-fill]); physical interference triggers a neutral let (*estorbo*).
- **Player elimination mid-round:** Not applicable; match retains constant agent roster until 22 points are reached.
- **Resource exhaustion:** Not applicable; court layout and point tally are non-depleting.
- **Termination guarantee:** Because each rally ends with probability 1 in finite turns and every point adds exactly 1 to one side's score, the match strictly terminates when either score hits 22 (maximum theoretical rallies bounded stochastically).
## State representation
- **Persistent state:** `score_A` (0..22), `score_B` (0..22), `server_id` (A or B), `service_fault_count` (0..1).
- **Per-rally ephemeral state:** `ball_pos` , `ball_velocity` , `ball_bounces` (0..1), `active_team` (A or B), `agent_positions` .
- **Observable information:** Fully observable to all agents.
## Action space
- `serve(target_zone, power)`: Executed by the server from service line; parameters target with chosen trajectory.
- `return_stroke(shot_type, target_wall_x, target_wall_z, power)`: Where `shot_type` .
- `reposition(target_x, target_y)`: Movement action during rally to anticipate the opponent's rebound.
## Playability notes
- **Single-player-vs-AI:** Natural; the human player directly commands agent A while an AI policy controls agent B with matching physics/skill parameters.
- **Group games (n>4):** Not applicable; 1v1 singles or 2v2 doubles are the closed canonical forms.
- **Physical-skill games:** Modeled via parameterized execution noise: [gap-fill] .
- **What is lost in the adaptation:** Tactile sensation of leather-on-palm impact, fine aerodynamic slice spin, and physical stamina/hand bruising are abstracted into statistical parameters.
## Implementation hints
Low-to-medium complexity. A discrete 2D spatial grid corresponding to the 10 *cuadros* of the frontón simplifies spatial tracking. First milestone: Implement state machine for 1v1 rally loop with deterministic wall rebounds and uniform random shot errors, then layer in spatial positioning and shot selection heuristics.
Download as JSON
Download the full game corpus entry for this game as structured JSON.
Download JSONCite this page
Bare-hand Basque Pelota (Spain). (n.d.). Games of Countries. https://gamesofcountries.com/games/es-spain/pelota-mano
@misc{gamesofcountries-es-spain-pelota-mano,
title = {Bare-hand Basque Pelota (Spain)},
year = {n.d.},
publisher = {Games of Countries},
url = {https://gamesofcountries.com/games/es-spain/pelota-mano}
}