Formal Specification
Armenian equestrian race and agility — Formal Specification
Armenia · equestrian-team · Confidence: high
← Back to game pageImplementation Suitability
Suitability: adaptation-requiredFidelity: discretizedGap-fills: 5
# Dzian-mrcum — AI-optimized specification
## General idea
A two-discipline mounted contest: an open cross-country distance race with a stamina-management/drafting layer, and a separate mid-gallop agility trial where riders must snatch ground targets without slowing or falling. What is structurally distinct is the stamina/pacing economy in the race (drafting and pacing are named explicitly as tactics, not flavor) combined with a lexicographic scoring rule in the agility trial — most targets collected first, fastest time only as a tiebreak.
## Formal model
- **Abstraction taken:** Continuous terrain, horse physiology, and rider skill are abstracted into a per-rider `stamina` resource that depletes with pace and terrain and a stochastic success check for each agility pickup attempt, since the entry names these dynamics qualitatively (drafting, uphill conservation, leaning to scoop a target) without numeric models.
- **Player/agent count:** 5-30 independent rider agents.
- **Turn model:** Synchronous ticks for the race phase (all riders choose pace simultaneously each tick); a per-attempt resolution for the agility phase (each rider's pickup attempts resolve independently as they reach each target position).
## Formal rules
R1. Setup: N riders (5 ≤ N ≤ 30) assemble at the start line; the course is a fixed sequence of waypoints (turning flags, river fords) of total length 5-15 km (entry §1.1-1.3).
R2. Start: on the signal, all riders begin moving simultaneously (entry §1.2).
R3. Per-tick pacing (race phase): each rider submits `set_pace(speed)`. `stamina` depletes each tick as an increasing function of `speed` and terrain gradient, and depletes more slowly for a rider whose `draft_status = drafting` behind another rider in the same pack (entry §2.2 names drafting and uphill stamina conservation as explicit tactics) [gap-fill: no numeric stamina-cost function is given by the entry; one must be chosen to make the tactic simulable]. A rider whose `stamina` reaches 0 has their max `speed` capped at a reduced value for the remainder of the race [gap-fill: the entry does not state the consequence of full stamina depletion, only that riders "must manage" it, so a speed-cap consequence is the minimal reading that makes the resource meaningful].
R4. Prohibited actions (entry §2.1): `push_opponent`, `grab_reins`, `whip_opponents_mount`. Any of these committed by a rider results in immediate disqualification from the race phase.
R5. Race win condition (entry §4.1): the race ends the instant one rider's horse crosses the finish line; that rider wins the distance race. All riders still on-course at that point are recorded by finishing order as they subsequently cross, but only the first-place result is a formally defined "win" per the entry.
R6. Agility trial setup (entry §3.1, a separate phase/lane from the race): a straight galloping lane is laid out with discrete target positions spaced 30 m apart.
R7. Agility attempt: as a rider reaches each target position at gallop speed, they may submit `pickup(target_id)`; precondition: rider is at that position, moving at or above a defined gallop-speed threshold [gap-fill: entry requires the rider be "galloping at full sprint," formalized as a minimum-speed precondition since no numeric threshold is given]. Resolve a stochastic success check; success collects the target and the rider continues at unchanged speed; failure triggers a stochastic fall-check [gap-fill: the entry's injury-risk flag and its description of leaning off the saddle imply fall risk on a failed attempt, though the entry does not explicitly state this consequence]. A rider who falls is eliminated from the remainder of the agility trial.
R8. Agility scoring (entry §4.2): each rider's result is the pair (`targets_collected`, `completion_time`). Results are ranked lexicographically — higher `targets_collected` always outranks a faster `completion_time`; `completion_time` is compared only to break ties in `targets_collected`. The top-ranked rider wins the agility crown.
R9. The race (R1-R5) and the agility trial (R6-R8) are independent contests, each producing its own winner; the entry gives no rule combining them into a single overall champion.
## Edge cases
- Tie at the end: in the race (R5), a tie for first (simultaneous finish-line crossing) is not addressed by the entry [gap-fill: treated as a shared win, since no photo-finish or replay procedure is described]. In the agility trial (R8), an exact tie in both `targets_collected` and `completion_time` is likewise unaddressed and treated as a shared win.
- A player with no legal move: not applicable — a moving rider always has `set_pace` available (race phase) and `pickup`/no-op available at a target position (agility phase).
- Repetition / cycles: not applicable — the race course is a fixed finite sequence of waypoints terminating at a finish line, and the agility lane has a fixed finite number of target positions; no state can recur indefinitely.
- Simultaneous claims on one position: not applicable — riders occupy independent lanes/positions on an open course; there is no exclusive-tile contest, only the prohibited-contact rule (R4) governing rider-to-rider interference.
- Elimination mid-round and turn-order repair: a disqualified (R4) or fallen (R7) rider is simply removed from further consideration in that phase; all other riders' state is unaffected, and there is no turn order to repair since movement is simultaneous, not sequential.
- Resource exhaustion: `stamina` depleting to 0 (R3) does not end a rider's race, only caps their speed — so this is a soft constraint, not a hard elimination; explicitly not treated as a loss condition since the entry never says stamina exhaustion ends a rider's participation.
- Termination guarantee: the race has a fixed finite course length, and every rider's speed is bounded below by the reduced-but-nonzero post-stamina cap (R3), so every non-disqualified rider reaches the finish line in bounded time; the race phase terminates. The agility trial has a fixed finite number of target positions and every rider either passes or is eliminated (R7) at each one, so it also terminates in bounded steps.
## State representation
- Persistent (event) state: `race_result` (finish order), `agility_result` (ranked (targets, time) pairs).
- Per-rider race state: `position`, `stamina`, `draft_status`, `speed`, `disqualified` (bool).
- Per-rider agility state: `position`, `targets_collected`, `elapsed_time`, `fallen` (bool).
- Observable information: fully observable — all riders see all positions, paces, and stamina states; no hidden information.
## Action space
- `set_pace(speed)`: any active rider, race phase; precondition: not disqualified, race in progress.
- `pickup(target_id)`: any active rider, agility phase; precondition: rider at `target_id`'s position, speed ≥ gallop threshold, not fallen.
- Prohibited (never in the legal set, but must be recognized to trigger R4): `push_opponent`, `grab_reins`, `whip_opponents_mount`.
## Playability notes
- Single-player-vs-AI: natural for both phases — a human rider competes against AI-paced opponents in the race, or against an AI-timed ghost result in the agility trial.
- Reducing N > 4 to 2 agents: straightforward for the race (fewer competitors, same course); the agility trial is already effectively single-rider-at-a-time in its scoring (individual best result), so it reduces trivially.
- Physical-skill games: real riding balance, pacing intuition, and the physical skill of leaning from a moving horse are entirely collapsed into `stamina` depletion and the pickup stochastic check; no genuine horsemanship is modeled.
- What is lost: the actual bond between rider and horse, real terrain-reading and drafting tactics, the ceremonial festival framing (banners, silver bridle prizes), and the specific danger and skill of a full-gallop ground pickup.
## Implementation hints
State is modest (per-rider position/stamina/speed plus a small target list), so the design difficulty is entirely in calibrating the stamina-cost function (R3) and pickup/fall probabilities (R7), none of which the source quantifies. First milestone: implement the race phase alone with a flat (non-drafting) stamina model to verify finish-order determination (R5), then add drafting and the agility trial once base pacing is validated.
Download as JSON
Download the full game corpus entry for this game as structured JSON.
Download JSONCite this page
Armenian equestrian race and agility (Armenia). (n.d.). Games of Countries. https://gamesofcountries.com/games/am-armenia/dzian-mrcum
@misc{gamesofcountries-am-armenia-dzian-mrcum,
title = {Armenian equestrian race and agility (Armenia)},
year = {n.d.},
publisher = {Games of Countries},
url = {https://gamesofcountries.com/games/am-armenia/dzian-mrcum}
}