About this program
Most people who want to use ML in finance get stuck in the same place: they understand the math but not the market context, or vice versa.
What this program covers
Covers the full pipeline from raw financial data to automated decision logic. You will work with time-series preprocessing, feature engineering on price and volume data, and building classification models that flag entry and exit conditions. Each module uses real historical datasets from equity markets.
The gap between notebooks and production
Running a model in a Jupyter notebook is very different from running it inside a live portfolio system. Spend significant time on this gap: logging predictions, handling missing data in real time, and setting up basic monitoring so you know when a model starts drifting.
Portfolio construction with ML signals
Signals alone are not a strategy. Cover how to combine multiple model outputs into a coherent position-sizing framework, including simple risk constraints like volatility targeting and drawdown limits. No guarantees on outcomes, but the logic is sound and well-documented in academic literature.
Tools used throughout
- Python with scikit-learn and LightGBM
- Pandas and NumPy for data wrangling
- Backtrader for strategy simulation
- PostgreSQL for storing tick and bar data
Who this is designed for
Works best for people with some Python experience who have already spent time reading financial statements or working with market data. Complete beginners to both fields will find the pace demanding.