Economy Assessment Standard v1.0

§ 01

Scope

The GhostPlay Economy Assessment Standard defines how GhostPlay assesses live-service game economy stability against a defined set of stress conditions. An assessment reports whether the economy model maintains predictive accuracy within specified thresholds when subjected to exactly three named stress vectors.

Assessment boundary: economy model predictive accuracy under named shock conditions.

Out of scope — this standard does NOT assess

All possible emergent failure modes not covered by the three named vectors
Gameplay balance or moment-to-moment player experience
Monetization ethics or fairness judgments
Store pricing strategy or real-world transaction value
User acquisition costs or marketing effectiveness

Publishers commissioning an assessment accept that a within-tolerance result is specific to the defined test conditions and does not constitute a guarantee of economy stability under all circumstances.

§ 02

Stress Vectors

All three vectors are applied at Day 180 of the simulation window. The Day 180–360 period constitutes the post-shock observation window used for MAPE scoring.

Vector A

Exploit Shock

+300% transaction volume injected at Day 180 across all Asset Classes simultaneously.

Models coordinated duplication exploits or viral economy hacks.

Vector B

Whale Flight

Top 10% spender cohort removed at Day 180.

Models sudden high-value player churn event (e.g., competitor release, controversy).

Vector C

Progression Overdrive

Class 2 (Velocity Catalyst) velocity ≥400% increase; Class 4 (Wildcard Faucet) velocity ≥200% increase, beginning Day 180.

Models aggressive content drop or battle pass acceleration.

§ 03

Systemic Asset Classes

The standard defines four Systemic Asset Classes. The validated signal today is the backtested mint-to-burn ratio (MBR), which rates each class's economy health directly. Forecast-accuracy scoring (MAPE per Asset Class per stress vector) is currently under methodological revision and is not used as a pass/fail gate. See § 04 below.

Class	Label	Definition
1	Fiat	Hard currency (real-money purchases)
2	Velocity Catalyst	Soft currency — earned in-game, spent on progression
3	Progression Sink	Consumables, upgrades, cosmetics sold for Class 2
4	Wildcard Faucet	Free-to-earn items with unpredictable supply dynamics

§ 04

Validation Criteria

1.50

Mint-to-Burn Ratio (MBR) collapse threshold MBR = cumulative currency inflow / cumulative currency sinks, computed from the simulated 360-day trajectory. MBR at or above 1.50 flags collapse risk, the structural signature of faucet/sink hyperinflation.

Within tolerance

MBR below 1.50: currency sinks are keeping pace with inflow, no structural collapse signature.

Out of tolerance

MBR at or above 1.50: currency is entering circulation faster than sinks remove it. Flagged in the findings report with its severity.

Forecast-Accuracy Scoring

Forecast-accuracy scoring (MAPE per Asset Class per stress vector, historically measured against a ≤5% per-class threshold over the Day 180–360 post-shock window) is currently under methodological revision and is not used as a pass/fail gate. It is reported in the delivered audit report as supporting detail, not as the validated economy-health signal.

§ 05

Agent Architecture

Simulations run 5,000 synthetic agents over 360 simulated days. Two persona archetypes drive all agent behavior:

Exploiter

Exploiter Persona

Finds and exploits economy inefficiencies via a scripted parametric rate model. Drives Vector A (Exploit Shock) scenarios and contributes to baseline economy stress.

Whale

Whale Persona

High-spend, high-volume, high-churn-risk archetype. Drives Vector B (Whale Flight) scenarios and constitutes the top 10% spender cohort.

Determinism Guarantee

Agent behavior is fully deterministic. Given an identical game binary and parameter set, simulation-stage telemetry is byte-identical run to run. Forecast-model scoring is not currently seeded and is not reproducible to 6 significant figures. This property enables independent verification of any issued findings report.

No Machine Learning in Agent Logic

Agent decision logic uses a deterministic parametric rate model. No machine learning is used in agent decision-making (ADR-0001) — the forecast model is a separate, server-side stage. All model weights are server-side forecast components only — not embedded in agents.

§ 06

Audit Deliverables

Upon completion of an economy assessment, the following deliverables are provided to the publisher within a 48-hour SLA from binary submission.

PDF

Audit Findings Report Severity-rated per-class economy-health read, scored on the mint-to-burn ratio, with forecast-accuracy detail included per Asset Class across all three stress vectors. Board-ready format.

JSON

Raw Telemetry Per-day actual vs. predicted economy state per stress vector. Complete machine-readable record enabling independent verification.

URL

Economy Replay Interface Interactive divergence timeline showing economy behavior before and after each stress vector injection. Hosted link, 12-month retention.

DOC

Findings Report — Investor Edition On request, the same findings report rendered for a non-technical M&A or fundraising diligence reader: it leads with the severity-rated risk summary and the fun-to-money revenue read, and sets the per-class forecast-accuracy detail aside as an appendix. A view of the same findings for investors and publishers — not a separate certification or seal.

SLA: 48 hours from binary submission to delivered report.

§ 07

Standard Lifecycle

v1.0 standard frozen for Phase 1 — valid for Months 1–18 from standard effective date.
Version updates require 90-day notice and must maintain backward compatibility for all previously issued assessment reports.
Report versioning: each issued findings report references the standard version under which it was evaluated. A report issued under v1.0 remains valid against v1.0 criteria regardless of subsequent standard updates.