rckrdmrd 75c4d07690 feat: Initial commit - ML Engine codebase

Hierarchical ML Pipeline for trading predictions:
- Level 0: Attention Models (volatility/flow classification)
- Level 1: Base Models (XGBoost per symbol/timeframe)
- Level 2: Metamodels (XGBoost Stacking + Neural Gating)

Key components:
- src/pipelines/hierarchical_pipeline.py - Main prediction pipeline
- src/models/ - All ML model classes
- src/training/ - Training utilities
- src/api/ - FastAPI endpoints
- scripts/ - Training and evaluation scripts
- config/ - YAML configurations

Note: Trained models (*.joblib, *.pt) are gitignored.
      Regenerate with training scripts.

Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>

2026-01-18 04:27:40 -06:00

1.8 KiB

Raw Permalink Blame History

Symbol-Timeframe Model Training Report

Generated: 2026-01-07 03:40:26

Configuration

Training Data Cutoff: 2024-12-31 (excluding 2025 for backtesting)
Dynamic Factor Weighting: Enabled
Sample Weight Method: Softplus with beta=4.0, w_max=3.0

Training Results Summary

Model	Symbol	Timeframe	Target	MAE	RMSE	R2	Dir Accuracy	Train	Val
GBPUSD_5m_high_h3	GBPUSD	5m	high	0.000504	0.000592	-0.6309	98.17%	310314	54762
GBPUSD_5m_low_h3	GBPUSD	5m	low	0.000548	0.000645	-0.6558	98.88%	310314	54762
GBPUSD_15m_high_h3	GBPUSD	15m	high	0.000887	0.001025	-0.6944	98.52%	104191	18387
GBPUSD_15m_low_h3	GBPUSD	15m	low	0.000955	0.001102	-0.7500	98.90%	104191	18387

Model Files

Models saved to: /home/isem/workspace-v1/projects/trading-platform/apps/ml-engine/models/symbol_timeframe_models

Model Naming Convention

{symbol}_{timeframe}_high_h{horizon}.joblib - High range predictor
{symbol}_{timeframe}_low_h{horizon}.joblib - Low range predictor

Usage Example

from training.symbol_timeframe_trainer import SymbolTimeframeTrainer

# Load trained models
trainer = SymbolTimeframeTrainer()
trainer.load('models/symbol_timeframe_models/')

# Predict for XAUUSD 15m
predictions = trainer.predict(features, 'XAUUSD', '15m')
print(f"Predicted High: {predictions['high']}")
print(f"Predicted Low: {predictions['low']}")

Notes

Models exclude 2025 data for out-of-sample backtesting
Dynamic factor weighting emphasizes high-movement samples
Separate models for HIGH and LOW predictions per symbol/timeframe

Report generated by Symbol-Timeframe Training Pipeline

1.8 KiB Raw Permalink Blame History