Data Science for Finance Course

Lesson 1 • Financial Data Types and Sources

Distinguishes structured, unstructured, and alternative financial data. Learners identify appropriate sources for each analytical use case.

Lesson 2 • Python Environment for Financial Analysis

Sets up a reproducible Python environment with finance-relevant libraries. Learners write their first financial data scripts using pandas and NumPy.

Lesson 3 • Core Financial Markets and Instruments

Covers equities, fixed income, derivatives, and FX markets as data-generating systems. Provides the financial context required for all subsequent analytical work.

Lesson 1 • Building Reproducible Data Pipelines

Teaches modular pipeline design using functions and configuration files. Ensures data workflows are auditable and reusable across projects.

Lesson 2 • Data Normalisation and Transformation

Covers scaling, log-transformation, and return calculation for financial series. Prepares data in the formats required by downstream statistical and ML models.

Lesson 3 • Retrieving Market Data Programmatically

Demonstrates API-based retrieval of price, volume, and fundamental data. Connects data acquisition to the analytical pipeline introduced in Chapter 1.

Lesson 1 • Hypothesis Testing in Finance

Applies t-tests, bootstrap methods, and multiple-testing corrections to financial hypotheses. Guards against spurious findings in large financial datasets.

Lesson 2 • Correlation and Covariance Analysis

Measures linear and rank-based relationships between financial assets. Lays the groundwork for portfolio construction and factor analysis.

Lesson 3 • Time-Series Patterns and Seasonality

Identifies trends, cycles, and calendar effects in financial data. Builds awareness of data artefacts that can mislead predictive models.

Lesson 4 • Financial Data Visualisation

Creates candlestick charts, return distributions, and drawdown plots using matplotlib and plotly. Effective visualisation accelerates hypothesis generation.

Lesson 1 • Backtesting Risk Models

Validates VaR models using exception counting and statistical tests. Ensures risk models meet internal and regulatory performance standards.

Lesson 2 • Value at Risk and Expected Shortfall

Derives VaR and ES using parametric, historical, and Monte Carlo methods. Connects risk metrics to regulatory capital and internal risk limits.

Lesson 3 • Portfolio Risk Decomposition

Decomposes total portfolio risk into factor and idiosyncratic components. Enables attribution of risk to individual positions and systematic exposures.

Lesson 4 • Volatility Estimation Methods

Computes historical, EWMA, and GARCH volatility estimates for financial assets. Volatility is the foundational input to all risk and pricing models.

Lesson 1 • Linear Models for Financial Prediction

Applies OLS, ridge, and lasso regression to return forecasting and factor modelling. Linear models serve as interpretable baselines for all subsequent ML work.

Lesson 2 • Gradient Boosting for Tabular Finance Data

Implements XGBoost and LightGBM on structured financial datasets. Gradient boosting consistently outperforms simpler models on tabular financial data.

Lesson 3 • Interpretability and Model Explainability

Uses SHAP values and partial dependence plots to explain model predictions. Interpretability is essential for regulatory compliance and stakeholder trust.

Lesson 4 • Model Evaluation in Financial Contexts

Applies cross-validation, walk-forward testing, and financial performance metrics. Prevents look-ahead bias and overfitting in time-series model evaluation.

Lesson 1 • Classical Time-Series Models

Fits ARIMA and SARIMA models to stationary and seasonal financial series. Establishes statistical forecasting baselines before introducing ML-based approaches.

Lesson 2 • Volatility Forecasting with GARCH Models

Extends GARCH to EGARCH and GJR-GARCH to capture asymmetric volatility. Volatility forecasts feed directly into option pricing and risk management.

Lesson 3 • Forecast Combination and Ensembling

Combines forecasts from multiple models to reduce variance and improve accuracy. Ensemble methods are standard practice in production financial forecasting.

Lesson 4 • Recurrent Neural Networks for Sequences

Builds LSTM and GRU networks for sequential financial data prediction. Deep sequence models capture long-range dependencies missed by classical models.

Lesson 1 • Strategy Performance Evaluation

Measures strategy quality using Sharpe, Sortino, Calmar, and drawdown metrics. Distinguishes genuine alpha from luck using statistical significance tests.

Lesson 2 • Position Sizing and Portfolio Construction

Applies Kelly criterion, volatility targeting, and mean-variance optimisation to size positions. Proper sizing determines realised risk-adjusted returns.

Lesson 3 • Signal Generation and Alpha Research

Constructs price-based, fundamental, and ML-derived trading signals. Signal quality is measured by information coefficient and decay analysis.

Lesson 4 • Overfitting and Strategy Robustness

Detects and mitigates overfitting through walk-forward testing and parameter sensitivity. Robust strategies maintain performance across unseen market regimes.

Lesson 1 • Model Monitoring and Drift Detection

Monitors deployed models for data drift, concept drift, and performance degradation. Proactive monitoring prevents silent model failures in live systems.

Lesson 2 • Communicating Results to Stakeholders

Translates complex model outputs into executive-ready reports and interactive dashboards. Effective communication drives adoption of data science in finance teams.

Lesson 3 • Fraud Detection Systems

Designs real-time fraud detection pipelines using anomaly detection and classification. Addresses extreme class imbalance and latency constraints in production.

Lesson 4 • Credit Risk Modelling at Scale

Builds end-to-end credit scoring pipelines using ML on large loan datasets. Covers probability of default, loss given default, and model governance.