Overview

This document provides a comprehensive technical overview of the Signal trading system, an enterprise-grade AI-driven cryptocurrency trading platform. The system combines multi-timeframe technical analysis, agent swarm AI consultation, and automated trading execution into a scalable, maintainable architecture.

The Signal system is designed for professional cryptocurrency trading with three core capabilities: automated signal generation through AI analysis, multi-agent consultation for manual trading decisions, and comprehensive risk management. For detailed information about specific AI decision-making components, see AI Trading Engine. For agent system implementation details, see Agent Swarm System. For configuration and deployment guidance, see Configuration and Deployment.

System Architecture Overview

The Signal trading system implements a service-oriented architecture with clear separation between AI decision-making, mathematical analysis, user interfaces, and trading execution.

High-Level Architecture

Core System Components

AI Trading Engine Pipeline

The system implements a three-stage AI decision pipeline that processes market data through specialized mathematical services and produces trading decisions with complete audit trails.

Agent Swarm Architecture

The agent swarm system provides intelligent consultation capabilities through specialized cryptocurrency agents that share common technical analysis tools.

Key Technical Characteristics

Mathematical Analysis Engine

The system employs six specialized mathematical services, each optimized for specific timeframes and trading strategies:

Service	Timeframe	Candles	Indicators	Specialization
ShortTermMathService	15min	144 (36h)	RSI(9), MACD(8,21,5), Bollinger(10,2.0)	Scalping optimization
SwingTermMathService	30min	96 (48h)	RSI(14), MACD(12,26,9), Bollinger(20,2)	Swing trading
LongTermMathService	1h	48 (48h)	RSI(14), MACD(12,26,9), SMA(50), ADX(14)	Position trading
VolumeDataMathService	1h	220/96	Pivot Points, Volume spikes (1.5x+)	Support/Resistance
SlopeDataMathService	1min	120 (2h)	VWAP, Price Slope, Momentum(10)	Micro-trend analysis
WhaleDetectionMathService	5min	72 (6h)	Volume anomalies, OBV analysis	Institutional detection

Configuration Management System

The system implements a sophisticated feature flag architecture with bidirectional UI-to-backend mapping through SettingsConnectionService:

• Signal Strength Control: 5-level slider (1-5) mapped to boolean flags isSignalStrengthMinimal through isSignalStrengthMaximal
• AI Behavior Control: 6 feature flags controlling risk filters, volume analysis, and overlap protection
• LLM Selection: 8 AI model options across HuggingFace, xAI, and Ollama providers
• Telegram Integration: Granular notification controls for buy, sell, wait, and watch signals

Multi-Provider AI Integration

The system supports dynamic AI model selection through InferenceMetaService with three provider categories:

• HuggingFace: openai/gpt-oss-120b, DeepSeek-R1, DeepSeek-V3
• xAI: grok-3, grok-3-mini
• Ollama: gpt-oss:120b, deepseek-r1:671b, deepseek-v3:671b

All completion types (RunnerCompletion, RunnerStreamCompletion, RunnerOutlineCompletion) use unified provider selection logic.

Enterprise-Grade Data Architecture

The system generates comprehensive trading datasets capturing complete decision pipelines from input analysis through AI reasoning to market execution results. Each trading decision creates a BacktestEventRow with:

• Complete audit trail: resultId linking to full LLM interaction logs
• Real-time P&L tracking: Actual percentage returns and dollar amounts
• Multi-dimensional capture: Technical indicators, volume dynamics, social signals
• Trading chain reconstruction: Complete narratives for each signal lifecycle

The Signal trading system represents a convergence of enterprise software architecture principles with cutting-edge AI technology, designed for professional cryptocurrency trading operations requiring both automated execution and human-assisted decision making.