galafis/ml-trading-signals
Machine learning system for generating trading signals using XGBoost, LightGBM, and ensemble methods. Features 40+ technical indicators, MLflow tracking, and FastAPI inference API.
What's novel
Machine learning system for generating trading signals using XGBoost, LightGBM, and ensemble methods. Features 40+ technical indicators, MLflow tracking, and FastAPI inference API.
Code Analysis
5 files read · 2 roundsA Python-based machine learning system that ingests historical data for Brazilian stocks, applies technical indicators, trains models (XGBoost/RF), and serves predictions via a FastAPI endpoint.
Strengths
Clear separation of concerns between feature engineering, model training, and API serving. The codebase is well-structured with distinct modules for data preparation, indicator calculation, and pipeline orchestration. Good use of standard libraries (pandas, scikit-learn) and MLflow for experiment tracking.
Weaknesses
Lack of comprehensive unit tests; error handling is basic (e.g., generic try/except blocks without specific exception types). The novelty is limited as it follows a standard ML pipeline pattern without unique algorithmic contributions. Some data preprocessing steps (scaling, target engineering) are somewhat generic.
Score Breakdown
Signal breakdown
Innovation
Craft
Traction
Scope
Evidence
Commits
26
Contributors
2
Files
31
Active weeks
5
Repository
Language
Python
Stars
1
Forks
1
License
MIT