ð¬How Does the AI Evaluate?
Our analysis is not a black box. We use 8 independent data sources and a 2-stage scoring system to neutrally analyze fake reviews and biased ratings.
1. Fraud Score Formula
Final Fraud Score =
Statistical Score à 0.7 + AI Judgment à 0.3
Blends statistical anomaly detection (70%) from playtime, review length, and owned games with Gemini NLP analysis (30%) to minimize AI hallucination risk.
2. 6-Axis Quality Rating
ðšGRAPHICS
Art & Visual Quality
ð®GAMEPLAY
Controls & Systems
ðSTORY
Narrative & World
ð°VALUE
Price vs Content
â±ïžVOLUME
Playtime & Density
ð§STABILITY
Performance & Bugs
Visualizes each game's strengths and weaknesses across 6 quality axes (0â100 pts).
3. Integrated Data Sources
ð® Steam API
ð Google Search
ð Metacritic
ð° Price History
â ïž Anomaly Detection
Cross-verifies not just Steam reviews but also external critic scores and Reddit/community sentiment for reliability.
>DATABASE
ðš
Overrated Fake
High Steam ratings but AI flagged as fake
â
My Watchlist
Track price and trust of saved games
Best Under $10
Best value picks under $10
Developer Trust
Developer trust score ranking
Game VS Battle
Compare two games head-to-head
ð®
Gamer Passport
Swipe to find your gamer type
ð¥
Health Check
AI game health diagnosis
ðš
Fraud Report
Weekly suspicious review ranking
ð
Controversy Archive
Historic Steam controversies
ð
Global News
Overseas controversies & indie discoveries
Lab Notes: Latest Analysis & Development Columns
View All Articles â
ð¬ Discovery Tools
Advanced exploration powered by AI analysis
