Research case study
AlphaCycle — Stock Prediction Framework
Modular data/model/report pipeline for multi-horizon equity prediction.
2015–24
Walk-forward window
Focus
Python · XGBoost
Methods
Pipeline · Walk‑forward
Repository
Private
01 / Context
The research question
How to build a reproducible, config-driven research stack for systematic equity signals?
02 / Method
Research design
Built a config-driven research framework spanning data ingestion, feature storage, model training, and reporting, with walk-forward validation and regime labels.
03 / Evaluation
How it was tested
- Walk-forward validation (2015–2024), 12-month rolling refit
- Metrics: directional accuracy, IC, Sharpe ratio
- Feature importance tracked via SHAP across horizons
04 / Findings
What the work showed
Clean separation of data/model/report layers; config-driven pipelines; evaluation with robust metrics across multiple horizons.
05 / Next iteration
Where I would take it next
Add live paper-trading integration and expand to alternative data sources.