Technology Fusion Architecture
Eight core technologies organically integrated to deliver differentiated investment insights
ARETE AI Trade is not just a list of technologies, but 8 core technologies integrated in real-time to provide optimal investment analysis services
Knowledge Intelligence
Ontology + RAG hybrid core foundation
Microservices
21 independent services, zero-downtime scaling
AI Agents
10+ specialized agents MoA collaboration
EOD Analysis
Investment opinion, AI score, success rate
AI Auto Trading
Real-time market analysis based auto trading
Strategy Trading
User-defined strategy-based trading execution
ARETE Store
Strategy & plugin marketplace
AI Learning
Personalized investment education system
Organic Integration → Differentiated Investment Insights
8 technologies seamlessly integrated in one platform
↓ See detailed information about each technology below
Microservices Architecture
21 independent services for unlimited scaling (11 Application + 4 AI/ML + 6 Infrastructure + 1 Gateway)
Application 5 · AI/ML 4 · Pipeline 3 · Infrastructure 6 · Gateway 1 = 19 services
API Gateway
(1 services)Kong API Gateway
:8080DB-less declarative API Gateway with JWT RS256, Rate Limiting, Circuit Breaker
Application Services
(5 services)Frontend
:3000Next.js 15 App Router with 100+ pages, real-time WebSocket, interactive Recharts
Admin Panel
:3001Next.js 15 admin panel for operations, monitoring, configuration
Backend API
:8000FastAPI core with 85+ SQLAlchemy models, 80+ endpoints, 4-tier API layers (v1/v2/internal/legacy)
Universal Gateway
:5001Multi-broker API unified interface (KIS, etc.) for trading operations
Private Trading
:5004AI-driven auto-trading engine with strategy execution and risk management
AI / ML Services
(4 services)AILab
:81005 AI architectures (ISSE/EOD/XAI/PAI/PerspectiveLens) with Intelligence Bus Stigmergy
Knowledge Intelligence
:5002Graph RAG pipeline with Neo4j ontology + pgvector HNSW + Hybrid Retriever
FinBERT Service
:8200Financial NLP specialized in sentiment analysis for news and disclosures
Hybrid Retriever
:8201BM25 + Vector hybrid search with score fusion for high-precision retrieval
Data Pipeline
(3 services)Observatory
:5003Data collection orchestrator with 26 scheduled jobs (KRX, KIS API, news)
Apache Pulsar
:6650Message broker with 4 namespaces (financial-data, intelligence, news-social, system)
Pulsar Manager
Web UI for Pulsar topic and tenant management
Infrastructure
(6 services)PostgreSQL
:5432Primary database with 85+ models, ACID transactions, pgvector HNSW for vector search
TimescaleDB
:5433Time-series database with hypertable optimization for OHLCV stock data
QuestDB
:9000High-frequency time-series database for sub-second tick data
Neo4j
:7474Graph database for stock-sector-news-indicator relationship traversal
Redis Cache
:6379LRU cache with 2GB memory for hot data and API response caching
Redis Session
:6380Persistent session/token storage with 512MB memory
5 AI Architectures
ISSE↔EOD↔XAI↔PAI↔Perspective Lens auto-connect via 6 paths through Intelligence Bus Stigmergy, forming collective intelligence.
ISSE
Intelligent Stock Screening Engine
Self-evolving stock screening with embeddings
Two-phase pipeline (Phase A: rules-based + Phase B: pgvector embedding + self-reinforcement) with badge system for screening success visualization.
- Phase A: Rule-based Screening — Multi-factor screening with customizable criteria
- Phase B: Embedding + Self-evolution — pgvector HNSW embeddings drive iterative improvement
- CrossArchFeedbackMixin — Receives signals from EOD/XAI/PAI for refinement
EOD
End-of-Day Pipeline
6-phase post-market analysis with conditional activation
Phase A-F orchestrated pipeline (with B/C conditional) for real-time/daily/weekly/monthly analysis. Integrated with XAI and Learning systems.
- 6-Phase Orchestration — A → B (cond) → C (cond) → D → E → F sequential analysis
- Multi-timeframe Analysis — Real-time / Daily / Weekly / Monthly
- 11 EOD Frontend Components — Rich visualization of Phase outputs
XAI
Explainable AI
SHAP-based explainability with 3-tier fallback + multilingual
SHAP TreeExplainer → DeepExplainer → KernelExplainer fallback chain. Phase 4 multilingual support (EN/KO/JP/ZH). 18 frontend components.
- 3-tier SHAP Fallback — Tree → Deep → Kernel explainer chain
- Multilingual Explanations — EN, KO, JP, ZH (Phase 4, 2026-04-18)
- 18 Frontend Components — Confidence gauge, evidence graph, factor waterfall, counterfactual slider
PAI
Personalized AI
Contextual Bandit LinUCB + Cold-Start 3-cluster
Linear Upper Confidence Bound bandit with cold-start clustering for new users. Daily incremental learning from XAI/ISSE feedback.
(Internal design target: after 60-day PAI learning)
- LinUCB Algorithm — Linear Upper Confidence Bound for exploration/exploitation
- Cold Start 3-Cluster — New users classified into 3 archetypes
- Daily Incremental Learning — Continuous improvement from user feedback
Perspective Lens
Multi-perspective Investment Analysis
4-lens fusion: Technical, Fundamental, Sentiment, Risk
Integrates 4 analytical perspectives with auto-weighted confidence. Modularized into 5 lens packages with singleton orchestration.
- 4 Perspectives — Technical / Fundamental / Sentiment / Risk
- Auto-weighted Confidence — Each lens has dynamic weight based on regime
- EOD/ISSE Integration — Receives inputs from EOD pipeline + ISSE candidates
Intelligence Bus
Stigmergy Pattern
Meta-system auto-connecting 5 architectures via 6 paths
- ISSEEOD
- EODXAI
- XAIPAI
- PAIISSE
- EODPerspective Lens
- Perspective LensXAI
13 LLM Models Dynamic Selection
Dynamic routing based on complexity/cost/reliability + Circuit Breaker + cost ceiling + auto-fallback.
Knowledge Intelligence
A hybrid knowledge system combining Ontology and Advanced RAG to understand relationships and context in financial data
Hybrid Architecture: Ontology + Advanced RAG
Ontology
Structured knowledge system defining explicit relationships between companies, sectors, themes, news, and financials
Advanced RAG
Context-aware retrieval using Graph structure for accurate information extraction
Relationship Analysis
Graph analysis of supply chains, competitors, and subsidiary relationships
Chain Impact Tracking
Real-time tracking of how events impact related companies
Context-Enhanced Analysis
Deep analysis through semantic connections, not just keywords
EOD Analysis + Knowledge Intelligence = Organic Integrated Analysis
Two core technologies integrate in real-time to deliver ARETE's differentiated investment insights
AI Agent System
10+ specialized agents collaborate using MoA (Mixture of Agents) pattern for comprehensive analysis
MoA Workflow
Orchestrator
Task Coordination
Coordinates all agents and integrates results to generate final insights
Fundamental Analyst
Financial Analysis
Analyzes balance sheets, income statements, and cash flow to evaluate intrinsic value
Technical Analyst
Chart Analysis
Analyzes price charts, moving averages, RSI/MACD and other technical indicators
Sentiment Analyst
Market Sentiment
Analyzes news, social media, and community reactions to gauge market psychology
Risk Analyst
Risk Assessment
Evaluates investment risk through VaR calculation, drawdown analysis, and correlation
Macro Analyst
Macroeconomic Analysis
Analyzes impact of GDP, interest rates, exchange rates on markets
Quant Analyst
Quantitative Analysis
Provides data-driven insights through factor analysis, alpha generation, and backtesting
Knowledge Graph Expert
Relationship Analysis
Discovers company relationships and connections using Neo4j knowledge graph
Screener
Stock Selection
Selects and ranks stocks meeting criteria through multi-factor screening
Reporter
Report Generation
Synthesizes all analysis results into easy-to-understand investment reports
Portfolio Optimizer
Asset Allocation
Combines modern portfolio theory with AI to suggest optimal asset allocation strategies
News Crawler
Real-time News
Collects and categorizes investment-related information from global news sources in real-time
Earnings Analyst
Earnings Analysis
Analyzes corporate earnings releases and compares with market expectations to provide insights
ESG Analyzer
Sustainability Assessment
Analyzes environmental, social, and governance factors to calculate ESG scores
Options Strategist
Derivatives Strategy
Suggests hedging and profit strategies using options and derivatives
EOD Analysis System
AI comprehensively analyzes 3,620 stocks after market close to calculate investment opinions, AI scores, and expected success rates
Investment AI Score
Multi-AI models synthesize technical/fundamental analysis to calculate 0-100 investment scores
Expected Success Rate
Predicts success probability based on historical patterns + Knowledge Graph similar case analysis
AI Investment Opinion
Auto-generates 5-level opinions: Strong Buy/Buy/Neutral/Sell/Strong Sell
EOD Analysis Pipeline
Data Collection
All stocks price/financials/news after close
AI Analysis
Multi-LLM ensemble analysis
Graph Integration
Knowledge Intelligence context enhancement
Result Calculation
Final score/opinion/success rate output
EOD analysis results are integrated with Knowledge Intelligence in real-time, providing insights that reflect company-sector-market context, not just numbers
AI Auto Trading System
Next-generation trading technology where EOD analysis, Knowledge Intelligence, and AI agents converge to execute personalized automated trading
Tech-Fusion Trading
EOD analysis results, Knowledge Intelligence context, and AI agent real-time analysis combine to capture optimal trading moments
User Customization
Freely configure risk tolerance, investment strategy, and trading frequency to match your personal style
Safe AI Trading
AI analyzes and suggests, but the final decision remains in user control for safe automated trading
Real-time Performance
Monitor detailed metrics including equity curve, drawdown, and win rate in real-time
Technology Fusion Architecture
Three core technologies integrate in real-time to achieve precision automated trading impossible with any single technology
Your Personal AI Trader
Choose from 4 risk levels - conservative to aggressive - for strategies matching your style
Directly adjust parameters like max positions and minimum confidence thresholds
Test safely with paper trading, then switch to live trading when ready
Currently in Development
Core features complete, beta service planned for H1 2025
"AI becomes your personal expert trader, monitoring markets 24/7 and executing optimal trades"
Strategy Trading System
Complete strategy trading pipeline from quant strategy building to backtesting and live automated trading
Strategy Trading Pipeline
Strategy Builder
No-code strategy design
Visual Editor
Build strategy logic with drag & drop
200+ Technical Indicators
RSI, MACD, Bollinger Bands and more
Condition Builder
Complex conditional logic operations
Backtesting Engine
Thorough validation before live trading
10+ Years Historical Data
Long-term performance simulation
Detailed Performance Metrics
Returns, Sharpe Ratio, MDD analysis
Parameter Optimization
Auto-discover optimal settings
Live Automated Trading
Execute validated strategies in real-time to never miss opportunities
In Development
Strategy builder and backtesting engine complete
ARETE Store
Premium marketplace ecosystem for trading investment tools, strategies, and plugins
Total Products
Downloads
Avg Rating
Two-Sided Platform Ecosystem
Strategy Store
45Purchase AI-powered investment strategies and apply them instantly
Plugin Store
18Extensions like Telegram alerts, auto-rebalancing, and more
Widget Store
24Various widgets to customize your dashboard
Template Store
12Expert-configured dashboard templates
Resource Store
156Icons, components, and UI/UX resources
Platform Features
Developer Ecosystem
Open platform where anyone can develop and sell strategies/plugins
Security Verification
All products pass dependency validation, conflict detection, and security review
Revenue Sharing
Sustainable model where developers and platform share profits
App Store for Investment Tools
An ecosystem for safely trading verified investment tools, like the Apple App Store
AI Personalized Learning System
AI analyzes user's investment style, learning progress, and performance to provide personalized learning paths
AI Insights
Personalized investment advice through style analysis and confidence-based recommendations
Personalized Learning Path
Auto-generated optimized curriculum matching your skill level and goals
Gamification
Motivate learning with XP/level system and badge achievements
Learning Modules
Investment Basics
Understanding stocks, bonds, and ETFs
Technical Analysis
Chart patterns and indicator interpretation
Fundamental Analysis
Financial statements and valuation analysis
AI-Powered Investing
Interpreting and utilizing AI signals
Growth System
Earn XP to level up and unlock new content
Earned Badges
AI Learning Pipeline
User Analysis
Collect investment style, learning history, performance data
AI Diagnosis
Analyze strengths/weaknesses and identify growth areas
Path Generation
Auto-generate personalized learning curriculum
Real-time Feedback
Dynamic adjustments based on learning progress
The AI learning system integrates with actual investment records, connecting theoretical learning to practical experience and supporting continuous growth
Future Technology Roadmap
The blueprint for investment technology ARETE AI Trade is building
Global Stock Support (2027~)
From 2027, real-time analysis and investment for US, European, and Japanese markets via Korean brokers
Social Trading
Follow successful traders' strategies and auto-replicate trades
Cryptocurrency Integration (2028~)
From 2028, analysis and trading support for Bitcoin, Ethereum, and more
AI Strategy Marketplace
Ecosystem for trading expert AI strategies and sharing profits
Building the Future of Investment Together
ARETE AI Trade is creating a future where anyone can access professional-grade investment tools through AI (2027: Global stocks, 2028: Crypto, 2029: All assets)
An Evolving Platform, ARETE's Worldview
An AI investment platform that evolves through recursive learning. Continuously growing through the cycle of analysis, prediction, execution, verification, and learning.
Explore Our Philosophy