Fractal Trading Analysis System

Overview

This project implements a sophisticated fractal trading analysis system that maps trade positions relative to atomic patterns (A-patterns) across multiple timeframes. The system creates comprehensive spatial-temporal datasets for neural network training to identify the most profitable trading setups.

System Architecture

Core Concept: Fractal Pattern Recognition

The system is based on the principle that market structure follows fractal patterns - the same atomic building blocks (A-patterns) exist across all timeframes, but their positioning and relationships create unique market structures.

Hierarchical Structure

📊 TRADING ANALYSIS SYSTEM
├── 🎯 STRUCTURAL LAYERS
│   ├── 📈 Symbol + Timeframe (e.g., NQU5, 5MIN)
│   │   └── 🆔 Singular ID (S1, S2, S3...)
│   │       └── 🧩 A-Patterns (A1, A2, A3, A4, A5, A6, A12-A19)
│   │           ├── 📊 Stage 1: A1 (Consolidation/Accumulation)
│   │           ├── 📈 Stage 2: A2 (Expansion/Markup)
│   │           ├── 🛡️ Support/Resistance: A5/A6
│   │           ├── 💰 Demand/Supply: A3/A4
│   │           └── 📐 VAL/INV Grid: A14/A15 (Fibonacci)
│   └── 📚 Historical Context
│       └── 🔄 Previous Singular IDs → Current Context
│
├── 🎯 TRADE POSITIONING ANALYSIS
│   ├── 📍 Spatial Mapping
│   │   ├── 🎯 Trade Inside A-Pattern (INSIDE vs RELATIVE_TO)
│   │   ├── 📏 Distance to All A-Patterns in S-Group
│   │   ├── 🎨 Anchor Points (9 for rectangles, 3 for lines)
│   │   └── 📐 VAL/INV Grid Positioning
│   ├── ⏰ Temporal Mapping
│   │   ├── 🕐 Timing Within A-Pattern Duration
│   │   ├── 📊 Pattern Sequence Timing
│   │   └── 🔄 Historical Pattern Evolution
│   └── 📊 Performance Correlation
│       └── 💰 PnL vs Positioning Analysis

A-Pattern Definitions

Atomic Pattern Types

Based on the analysis system, each A-pattern represents a specific market behavior:

1. A1 - Stage 1: Consolidation/Accumulation Phase
2. A2 - Stage 2: Markup/Distribution Phase
3. A3 - Demand: Buying Pressure Zone
4. A4 - Supply: Selling Pressure Zone
5. A5 - Support: Price Floor Level
6. A6 - Resistance: Price Ceiling Level
7. A12 - Bull S1 INV: Bullish Stage 1 Inventory/Invalidation
8. A13 - Bull S1 VAL: Bullish Stage 1 Value Area
9. A14 - Bull S2 INV: Bullish Stage 2 Inventory/Invalidation
10. A15 - Bull S2 VAL: Bullish Stage 2 Value Area
11. A16 - Bear S1 INV: Bearish Stage 1 Inventory/Invalidation
12. A17 - Bear S1 VAL: Bearish Stage 1 Value Area
13. A18 - Bear S2 INV: Bearish Stage 2 Inventory/Invalidation
14. A19 - Bear S2 VAL: Bearish Stage 2 Value Area

VAL/INV Grid System

• A14/A15: Create the Fibonacci grid boundaries for Stage 2 analysis
• A12/A13: Create simplified grid boundaries for Stage 1 analysis
• Purpose: Define validation and invalidation levels for trade positioning

Multi-Timeframe Analysis

Timeframe Hierarchy

The system analyzes patterns across multiple timeframes simultaneously:

• Higher Timeframes (HTF): Daily, 4H, 1H (market structure)
• Medium Timeframes (MTF): 30M, 15M (intermediate patterns)
• Lower Timeframes (LTF): 5M, 1M (entry/exit precision)

Fractal Consistency

Daily (HTF): S1 → A1 → A2 → A3 → A14/A15 (VAL/INV)
4H (HTF):    S1 → A1 → A2 → A3 → A14/A15 (VAL/INV)
1H (MTF):    S1 → A1 → A2 → A3 → A14/A15 (VAL/INV)
30M (MTF):   S1 → A1 → A2 → A3 → A14/A15 (VAL/INV)
5M (LTF):    S1 → A1 → A2 → A3 → A14/A15 (VAL/INV)
1M (LTF):    S1 → A1 → A2 → A3 → A14/A15 (VAL/INV)

Key Principle: Same A-patterns exist across all timeframes, but their positioning and relationships create unique market structures.

Spatial-Temporal Analysis

Trade Positioning Metrics

For each trade, the system calculates:

Spatial Analysis

• Entry/Exit Retracement: Position within A-pattern price range (0.0-1.0)
• Distance from Validation: How far from upper boundary (A14/A15)
• Distance from Invalidation: How far from lower boundary (A14/A15)
• Closest Anchor Point: Which of 9 anchor points (rectangles) or 3 anchor points (lines) the trade is nearest to
• Spatial Score: Normalized positioning quality (0-1)

Temporal Analysis

• Entry/Exit Time Percent: Position within A-pattern duration (0.0-1.0)
• Time Distance from Anchor: Minutes from closest anchor point
• Pattern Sequence Timing: When trade occurs relative to A1→A2→A3 sequence
• Temporal Score: Timing quality relative to pattern middle (0-1)

Combined Metrics

• Combined Score: Average of spatial and temporal scores
• Relationship Type: INSIDE (trade inside A-pattern) vs RELATIVE_TO (trade outside but analyzed against A-pattern)

Anchor Point System

Rectangle Patterns (9 anchor points)

• Top Row: top_left, top_center, top_right
• Middle Row: middle_left, middle_center, middle_right
• Bottom Row: bottom_left, bottom_center, bottom_right

Line Patterns (3 anchor points)

• start: Beginning of line
• middle: Center point of line
• end: End of line

Data Flow

1. Input Data

• Trade CSV: Entry/exit times, prices, PnL, contract information
• Analysis JSON: A-patterns, S-groups, VAL/INV grids across timeframes

2. Processing Pipeline

1. Data Loading: Import trade and analysis data
2. Pattern Extraction: Parse A-patterns from analysis JSON
3. S-Group Organization: Group A-patterns by Singular ID
4. Trade Mapping: Map trades to relevant S-groups and timeframes
5. Spatial-Temporal Analysis: Calculate positioning metrics for all trade-A-pattern combinations
6. Feature Engineering: Create normalized metrics, scores, and categorical features

3. Output Dataset

Comprehensive ML-ready dataset with:

• Core Metrics: Normalized_Change, Normalized_Bars, Pattern_Strength, Volatility_Score, Pattern_Confidence, Z_Score
• Succession Features: Has_Next_Pattern, Next_Pattern, Time_To_Next_Minutes
• S-Group Features: S_Group_Size, S_Group_Avg_Strength, S_Group_Consistency
• Fibonacci Features: Fib_Type, Fib_Level, Zone (for A1 patterns)
• Spatial-Temporal Features: Trade positioning relative to S-pattern boundaries
• Categorical Features: Symbol, Timeframe, Pattern types (one-hot encoded)

Neural Network Training Objectives

Learning Goals

The neural network will learn to identify:

1. Most Profitable Setups: Which A-pattern combinations lead to profitable trades
2. Optimal Positioning: Where within VAL/INV grids to place trades
3. Perfect Timing: When Stage 1→Stage 2 transitions occur
4. Pattern Reliability: Z-scores and historical consistency for predictability
5. Multi-Timeframe Context: How HTF/MTF/LTF alignment affects profitability

Predictive Elements

• A1/A2 Averages & Z-Scores: Current vs historical performance comparison
• Pattern Reliability: Z-score proximity indicates predictable patterns
• Breakout Timing: When Stage 1 consolidation leads to Stage 2 expansion
• Contextual Analysis: How previous S-patterns influence current setups

Expected ML Outputs

• Entry/Exit Recommendations: Optimal trade positioning
• Probability Scores: Win/loss likelihood
• Timing Windows: Best entry/exit times
• Pattern Recognition: Identify profitable setups
• Risk Assessment: Position sizing and management

Key Insights for ML Training

Pattern Evolution

• Stage 1 (A1): Consolidation/accumulation phase
• Stage 2 (A2): Expansion/markup phase
• Transition Timing: When A1 approaches average bar count, A2 breakout is likely
• Pattern Reliability: Z-scores close to 0 indicate predictable patterns

Spatial-Temporal Relationships

• Perfect positioning doesn't guarantee profits: 0.950 combined score can still result in losses
• Pattern relationships matter: Trade inside A1 but poorly timed relative to A2
• Temporal context is crucial: Trade timing relative to pattern sequences
• Multiple pattern analysis provides richer context: Single pattern analysis is insufficient

Multi-Timeframe Context

• Timeframe independence: Same A-patterns exist across all timeframes
• Timeframe relationships: Higher timeframe context influences lower timeframe decisions
• Pattern alignment: Successful trades often align across multiple timeframes
• Hierarchical importance: HTF > MTF > LTF for decision making

File Structure

Trade journal/
├── analysis_dashboard.py          # Main analysis application
├── data/
│   ├── trades_export (2).csv     # Trade data
│   ├── trades_export (3).csv     # Additional trade data
│   └── indexeddb_complete_export (1).json  # Analysis patterns
├── models/                       # Neural network models (future)
├── notebooks/                    # Jupyter notebooks for analysis
├── trade_analysis_output/        # Generated analysis outputs
├── visualizations/               # Charts and graphs
└── README.md                     # This documentation

Usage

1. Upload Data: Import trade CSV and analysis JSON files
2. Run Analysis: Execute the dashboard to process all data
3. Review Results: Examine spatial-temporal analysis and pattern metrics
4. Export ML Dataset: Download comprehensive dataset for neural network training
5. Train Models: Use the exported data to train profitability prediction models

Technical Notes

Score Calculations

• Spatial Score: 1 - (closest_anchor_price_distance / a_price_range)
• Temporal Score: 1 - abs(entry_time_percent - 0.5)
• Combined Score: (spatial_score + temporal_score) / 2

Data Requirements

• Trade Data: Must include Id, ContractName, EnteredAt, ExitedAt, EntryPrice, ExitPrice, PnL, Type
• Analysis Data: Must include A-pattern labels, chartValues, metrics, and timeframe information
• Symbol Matching: Trade ContractName must match analysis symbol for proper mapping

Limitations

• Pattern Type Assumption: Currently assumes all patterns are rectangles (can be enhanced for lines)
• Timeframe Matching: Requires exact symbol/timeframe matches between trade and analysis data
• Historical Context: Limited to available analysis data (no external market context)

Future Enhancements

1. Multi-Timeframe Integration: Enhanced analysis across all timeframes simultaneously
2. Pattern Type Detection: Automatic detection of line vs rectangle patterns
3. Historical Context: Integration with external market data for broader context
4. Real-Time Analysis: Live pattern recognition and trade positioning
5. Advanced ML Models: Deep learning models for pattern prediction

Conclusion

This fractal trading analysis system provides a comprehensive framework for understanding market structure through atomic patterns. By mapping trade positions relative to A-patterns across multiple timeframes, the system creates rich datasets for neural network training to identify the most profitable trading setups.

The key insight is that market structure follows fractal patterns - the same atomic building blocks create infinite variations through different arrangements. The neural network learns to recognize which combinations and positioning strategies lead to profitable trades within this fractal framework.