Skip to main content
Formal Specification

Shuttlecock Kicking / Featherball — Formal Specification

Vietnam · dexterity-throw · Confidence: low

← Back to game page

Implementation Suitability

Suitability: adaptation-requiredFidelity: probabilistic-proxyGap-fills: 5
# Đá cầu — AI-optimized specification ## General idea Đá cầu is a traditional Vietnamese foot-dexterity sport where players volley a weighted, feathered shuttlecock across a central net using solely their feet, knees, chest, and head. In competitive match play, agents alternate tactical shot selection—ranging from delicate drop shots to high aerial spikes—seeking to force the opponent into a handling error or land the shuttlecock on the opposing court. ## Formal model - **Abstraction taken:** The continuous 3D aerial flight mechanics and acrobatic body strikes (overhead bicycle kicks, heel flicks) are abstracted into a discrete rally state machine. Each shot choice has a success and return difficulty parameterized by player skill, positioning, and stamina [gap-fill: continuous aerial foot acrobatics and aerodynamics are discretized into a rally state machine where each shot is chosen from a discrete action set and resolved via stochastic execution probabilities]. - **Player/agent count:** 2 agents (1v1 singles match). - **Turn model:** Alternating sequential shot execution within an active rally. ## Formal rules R1. Match and Court Structure: - Match is played as best-of-three sets. - A set is won by the first player to reach 21 points with a minimum 2-point lead. - Court is divided into two halves (Zone A and Zone B) separated by a net. R2. [gap-fill] Shot Taxonomy and Probability Model: - Shot action set 𝒜={SERVE,LOB,DROP,DRIVE,SPIKE}. - Each shot a∈𝒜 has an intrinsic execution difficulty D(a) and defensive pressure Q(a): - LOB: D=0.10,Q=0.20 (safe, easy to return). - DRIVE: D=0.25,Q=0.45 (medium speed). - DROP: D=0.35,Q=0.60 (short net placement, risky). - SPIKE: D=0.50,Q=0.80 (high-power offensive kill, high error risk). - Probability of clean execution: P(clean)=clamp(skillstriker−D(a)−Q(aincoming),0.05,0.98). R3. Rally Flow: - Server initiates the rally with SERVE. - If execution fails (sample Bernoulli(P(clean))=0): the striker commits a fault; opponent scores +1 point; rally terminates. - If execution succeeds: turn passes to the receiving agent, who must select a return shot from 𝒜∖{SERVE}. - Maximum touches per side: in singles (1v1), maximum 2 consecutive touches per player before the shuttlecock must cross the net. R4. [gap-fill] Deuce and Set Cap: - If score reaches 20–20, play continues until one player leads by 2 points, or until a hard ceiling of 25 points is reached (at 24–24, first to 25 wins). R5. [gap-fill] Rally Length / Loop Termination: - If a single rally exceeds 30 consecutive shot exchanges without fault, fatigue penalty reduces P(clean) by an additional 5% per subsequent shot, guaranteeing stochastic rally termination. R6. Set and Match Victory: - Winning a set increments the player's set score. - The first player to win 2 sets wins the match immediately. ## Edge cases - Tie at the end condition: Handled by R4; sets require a 2-point margin or resolve at the 25-point cap. - A player with no legal move: The active receiver must choose a shot action from 𝒜; failing to execute is evaluated as a point loss. - Repetition / cycles: Bounded by R5 (stamina decay on rallies >30 exchanges) and finite set winning points. - Simultaneous claims on the same resource or position: Not applicable; shots strictly alternate across the net. - Player elimination mid-round; turn order repair: Not applicable in 1v1 match play. - Resource exhaustion: Points and sets accumulate monotonically; no consumable physical resource. - Termination guarantee: why the game provably ends: Guaranteed by R4 (point caps) and R5 (fatigue-induced stochastic rally termination). ## State representation - Persistent state: - `sets_won_A`, `sets_won_B`: integers ∈[0,2]. - `score_A`, `score_B`: integers ∈[0,25]. - `serving_player`: enum `{PLAYER_A, PLAYER_B}`. - Per-round ephemeral state: - `rally_shot_count`: integer ∈[0,∞). - `last_shot_type`: nullable enum ∈𝒜. - `active_striker`: enum `{PLAYER_A, PLAYER_B}`. - Observable information: - Fully observed by both agents: scores, current rally length, previous shot type, and public player attributes. ## Action space - `select_shot(shot_type)`: shot_type ∈{SERVE,LOB,DROP,DRIVE,SPIKE}. - `set_power(level)`: [gap-fill] continuous or discrete power multiplier influencing risk vs reward. ## Playability notes - Single-player-vs-AI: Highly natural; human player engages in a tactical turn-based or timed shot-selection duel against an AI opponent. - Group games (n>4): The match model naturally extends to 2v2 or 3v3 team play by adding a touch-distribution sub-phase (pass to teammate before clearing net). - Physical-skill games: Acrobatic flexibility and foot coordination are mapped to the skill scalar and shot difficulty matrix D(a). - What is lost in the adaptation: The visual spectacle of aerial backflips, spinning jump kicks, and the crisp metallic click of coin-weighted shuttlecocks. ## Implementation hints - Model as a simple discrete Markov Decision Process (MDP) or two-player zero-sum game. - Suggested first milestone: Implement the 1v1 rally state machine with R2 shot difficulty tables and verify that 100 simulated sets terminate with realistic score distributions.

Download as JSON

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

Download JSON
Cite this page

Shuttlecock Kicking / Featherball (Vietnam). (n.d.). Games of Countries. https://gamesofcountries.com/games/vn-vietnam/da-cau

@misc{gamesofcountries-vn-vietnam-da-cau,
  title = {Shuttlecock Kicking / Featherball (Vietnam)},
  year = {n.d.},
  publisher = {Games of Countries},
  url = {https://gamesofcountries.com/games/vn-vietnam/da-cau}
}