Open Methodology

GhostPlay Economy
Assessment Standard

The open methodology defining stress vectors, validation criteria, agent architecture, and audit scope for live-service economy assessment.

Version 1.0 Effective 2026-06-13 Phase 1 — Months 1–18

§ 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. All four are scored against the MAPE gate across all three stress vectors, and the findings report rates each class's economy health accordingly.

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

≤5%

MAPE threshold per Asset Class per stress vector Measured during the Day 180–360 post-shock observation window. Mean Absolute Percentage Error between predicted and actual economy state.

Within tolerance

All four Asset Classes at ≤5% MAPE across all three stress vectors. Twelve total measurements, all within threshold.

Out of tolerance

Any single Asset Class exceeds 5% MAPE under any stress vector. Each out-of-tolerance measurement is flagged in the findings report with its drift severity.

Score Range

MAPE scores range from 0%–100%+ with no upper bound on failure. A score of 0% represents perfect prediction. Scores are reported per Asset Class per vector in the delivered audit report.

§ 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 scripted Behavior Trees. 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 output is 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 Behavior Trees exclusively. No machine learning models are used in agent decision-making (ADR-0001). 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 with a per-Asset-Class MAPE breakdown 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.

Cite this standard

GhostPlay Economy Assessment Standard v1.0 (2026). ghostplay.tech/standard