Trading Components
This document covers the core trading functionality of the Signal trading system, including the three-stage AI decision pipeline, signal generation and validation, position management, risk assessment, and execution components. This page provides an architectural overview of how trading decisions are made and executed.
For detailed AI agent consultation capabilities, see Agent System Reference. For configuration and deployment details, see Configuration and Deployment. For mathematical analysis specifics, see Mathematical Analysis Services.
Overview
The trading system employs a sophisticated three-stage AI pipeline that processes market data through specialized services to generate, validate, and execute trading decisions. The architecture consists of AI decision engines, mathematical analysis services, validation layers, and execution components that work together to automate cryptocurrency trading.
The core components include:
- Signal Generation: AI-driven market entry analysis using
SignalOutline - Position Management: Intelligent position monitoring and closing via
CloseOutline - Risk Assessment: Market condition evaluation through
RiskOutline - Validation Services: Multi-layer signal and close validation
- Execution Pipeline: Automated order execution and notification systems
Three-Stage AI Trading Pipeline
Core Decision Flow Diagram
The pipeline operates with dynamic Time-To-Live (TTL) intervals:
- RiskOutline: 15-minute intervals for market condition assessment
- SignalOutline: Dynamic TTL based on market conditions (1.5-5 minutes)
- CloseOutline: Adaptive TTL based on P&L momentum (1.0-5 minutes)
Service Architecture and Component Integration
Trading Components Architecture Diagram
The architecture employs dependency injection through a container system that binds services and ensures loose coupling between components. The TYPES constant defines service identifiers used throughout the system.
Component Integration and Data Flow
Signal Generation Process
The signal generation process begins with RiskAutoService evaluating market conditions every 15 minutes. If conditions are favorable (action: "trade"), SignalAutoService activates with dynamic TTL intervals. The SignalOutline analyzes multi-timeframe data using mathematical services and generates trading recommendations.
Key validation steps include:
- Signal strength validation: Ensures minimum TP/SL thresholds via
SignalValidationService - Overlap protection: Prevents conflicting positions in same price range
- Feature flag control: Granular behavior modification through
SettingsConnectionService
Position Management Workflow
Once a position is opened via SignalCoreService.commitBuySignal(), CloseAutoService begins monitoring with adaptive TTL intervals based on P&L momentum:
- PROFIT_RISING_TTL: 2.5 minutes (profit growing)
- PROFIT_FALLING_TTL: 1.0 minutes (profit declining - fast check)
- LOSS_RISING_TTL: 5 minutes (loss shrinking - slow check)
- LOSS_FALLING_TTL: 5 minutes (loss growing - slow check)
The CloseOutline specializes in scalping-oriented early closure on reversal signals, particularly in bearish market conditions.
Mathematical Analysis Integration
Six specialized mathematical services provide comprehensive technical analysis:
| Service | Timeframe | Specialization | Data Window |
|---|---|---|---|
ShortTermMathService |
15-minute | Scalping | 144 candles (36 hours) |
SwingTermMathService |
30-minute | Swing trading | 96 candles (48 hours) |
LongTermMathService |
1-hour | Position trading | 48 candles (48 hours) |
VolumeDataMathService |
Variable | Support/Resistance | 220/96 hourly candles |
SlopeDataMathService |
1-minute | Momentum analysis | 120 candles (2 hours) |
WhaleDetectionMathService |
5-minute | Institutional activity | 72 candles (6 hours) |
Each service calculates specific technical indicators optimized for its timeframe and trading strategy.
Configuration and Feature Control
The system implements sophisticated feature flag architecture with guaranteed consistency between UI controls and backend business logic. Configuration is managed through SettingsConnectionService with Redis-based persistent storage.
Key Configuration Areas
- Signal Strength Control: 5 levels with corresponding TP/SL thresholds
- AI Behavior Features: 6 smart filters for risk management
- LLM Selection: 8 AI model options for decision making
- Telegram Notifications: Granular control over message types
The configuration system uses bidirectional transforms between user-friendly UI controls and backend boolean flags, ensuring no configuration conflicts.
Execution Pipeline and Notifications
The execution pipeline integrates with external systems through WebhookService for order execution and TelegramWebService for user notifications. Each trading event triggers specific notification sequences based on feature flag settings.
The TelegramWebService implements six notification methods corresponding to different trading events:
- Buy signal notifications
- Take profit notifications
- Stop loss notifications
- Close signal notifications
- Wait signal notifications
- Watch (monitoring) notifications
All notifications include markdown formatting for professional presentation of trading data and P&L information.