Formal Specification
Sindhi Sash Wrestling — Formal Specification
Pakistan · wrestling-combat · Confidence: low
← Back to game pageImplementation Suitability
Suitability: engine-readyFidelity: probabilistic-proxyGap-fills: 2
# Malakhro — AI-optimized specification
## General idea
A two-agent martial contest where both wrestlers maintain continuous mutual grip on a waist sash (*Sandhro*) while applying leverage, foot-sweeps, and hip-throws from a left-shoulder engagement. Each discrete exchange resolves a simultaneous choice of offensive technique, defensive counter-stance, and leverage commitment against a stochastic throw-resolution table. The bout ends decisively on the first successful back-fall.
## Formal model
- **Abstraction taken:** Continuous physical grappling is abstracted into discrete tactical rounds (exchanges). Each exchange consists of simultaneous technique commitment, grip-retention verification, and a stochastic fall-probability evaluation parameterized by wrestler attributes `(strength, balance, technique_rating)`.
- **Player/agent count:** 2 agents per bout.
- **Turn model:** Simultaneous hidden action selection per exchange, followed by deterministic reveal and probabilistic state resolution.
## Formal rules
R1. Setup: Both agents start in the `standing_locked` state with `grip_retained = True`, `fouls = 0`, and `fall_state = None`.
R2. In each exchange, both agents simultaneously select one action from the legal action set: `{sweep_ghori, hip_throw, lift_takedown, counter_brace, re_grip_twist}`.
R3. Grip constraint: An agent attempting a throw must commit at least one grip resource point. If an agent attempts an action that drops both grip points, a foul is recorded (`fouls += 1`).
R4. Matchup resolution: The chosen action pair `(action_A, action_B)` is evaluated against a fixed technique-counter matrix `M[action_A][action_B]` to produce transition probabilities:
`P(A throws B)`, `P(B throws A)`, `P(stalemate_reset)`.
R5. Win condition:
- If a throw succeeds against agent B, agent B transitions to `fallen_back`, and agent A wins the bout immediately.
- If agent B incurs 2 fouls (`fouls >= 2`), agent B is disqualified and agent A wins immediately.
R6. Stalemate / Neutral reset: If `stalemate_reset` is sampled, both agents remain standing, positions update slightly, and the next exchange begins.
R7. [gap-fill] Exchange limit: A single bout allows at most 20 exchanges. If no decisive fall occurs by exchange 20, the agent with higher accumulated leverage points is declared the winner; if tied, sudden-death exchanges continue until first score.
## Edge cases
- **Tie at the end condition:** If both agents execute a simultaneous back fall on the exact same tick, [gap-fill] the exchange is declared null (`barabar`) and replay occurs from neutral lock.
- **A player with no legal move:** Not applicable; defensive brace (`counter_brace`) is always legal.
- **Repetition / cycles:** Bounded by the 20-exchange hard cap in R7 and non-zero throw probability `P(throw) > 0` per exchange.
- **Simultaneous claims on the same resource or position:** Both agents commit simultaneously; matchup matrix evaluates bilateral interaction synchronously.
- **Player elimination mid-round:** Not applicable (1v1 bout; bout terminates on first fall or foul-out).
- **Resource exhaustion:** Grip points refresh every neutral reset.
- **Termination guarantee:** Guaranteed by R7's finite exchange horizon combined with non-zero decisive outcome probability per round.
## State representation
- **Persistent state:** `bout_id`, `scores_A`, `scores_B`, `fouls_A`, `fouls_B`, `exchange_count in {1..20}`.
- **Per-exchange ephemeral state:** `action_A`, `action_B` (private until resolution), sampled random float `r in [0, 1)`.
- **Observable information:** Fouls, current exchange counter, and opponent's physical attribute tiers are public; current-exchange technique choice is hidden until simultaneous reveal.
## Action space
- `sweep_ghori()`: Execute traditional Sandhi waist-sash trip.
- `hip_throw()`: Torso rotation throw.
- `lift_takedown()`: Two-handed belt elevation.
- `counter_brace()`: Defensive sprawl.
- `regrip_twist()`: Re-establish two-handed waist cloth grip.
## Playability notes
- **Single-player-vs-AI:** Natural; human directly commands one wrestler's technique choices while AI calculates optimal mixed strategies based on attribute differentials.
- **Group games (n>4):** Reduces via tournament bracket (e.g. 8-agent single-elimination tournament).
- **Physical-skill games:** Physical attributes are modeled as scalars: `strength` (influences lift throws), `balance` (defends against sweeps), `technique` (modifies counter matrix).
- **What is lost:** The fine continuous pressure of shoulder contact, foot friction on sand, and dynamic balance adjustments are abstracted into discrete matrix calculations.
## Implementation hints
Low state complexity. The core implementation requirement is the 5x5 technique-counter payoff matrix. Implement a deterministic prototype with fixed rock-paper-scissors counters before adding attribute-based probabilistic variance.
Download as JSON
Download the full game corpus entry for this game as structured JSON.
Download JSONCite this page
Sindhi Sash Wrestling (Pakistan). (n.d.). Games of Countries. https://gamesofcountries.com/games/pk-pakistan/malakhro
@misc{gamesofcountries-pk-pakistan-malakhro,
title = {Sindhi Sash Wrestling (Pakistan)},
year = {n.d.},
publisher = {Games of Countries},
url = {https://gamesofcountries.com/games/pk-pakistan/malakhro}
}