3rd Edition

The ML4T Curriculum

27 chapters across 6 parts — from data infrastructure and feature engineering through machine learning models, strategy backtesting, and live deployment.

27

Chapters

720+

Pages

446+

Notebooks

9

Case Studies

Part 1

Introduction & Financial Data

The Process Is Your Edge

Why process discipline matters
Introducing the ML4T workflow
Causal inference and generative AI in the workflow
Keeping up with changing market regimes
Independent versus institutional workflows in the real world

6 sections 1 notebooks

The Financial Data Universe

A Modern taxonomy of financial data
The asset-class market data landscape
A due diligence framework for data sourcing
Storing data

5 sections 16 notebooks

Market Microstructure

How microstructure impacts price formation
The anatomy of modern market data feeds
From raw messages to the limit order book
The art of sampling
Detecting price jumps in intraday returns

7 sections 17 notebooks

Fundamental and Alternative Data

The point-in-time pipeline
Entity resolution and mapping
Fundamentals across the asset-class spectrum
Understanding alternative data
Using text data for NLP features

6 sections 13 notebooks

Synthetic Financial Data

The quant’s dilemma
Evaluating synthetic financial data
Classical simulation baselines
Generative model taxonomy
GANs for financial time series

9 sections 8 notebooks

Part 2

Research Design and Feature Engineering

Strategy Research Framework

From idea to evidence with the ML4T workflow
Mapping strategies and sources of edge
Defining the trading setup
Setting objectives and evaluation metrics
Evaluation protocol for time series

8 sections 4 notebooks

Defining the Learning Task

Data preprocessing and encodings
Label engineering
Univariate feature–label evaluation
Search accounting and multiple testing
From correlation to causality

6 sections 10 notebooks

Financial Feature Engineering

Capturing and configuring the economic drivers
Price-derived features
Structural and cross-instrument features
Contextual and slow-moving features
Cross-cutting feature types and the limits of direct aggregation

7 sections 8 notebooks

Model-Based Feature Extraction

Diagnostics and stationarity features
Transforming signals to uncover hidden structure
Volatility Features
Uncertainty features
Regime features

7 sections 22 notebooks

Text Feature Engineering

Lexical and statistical models
Static embeddings
Sequential models
Transformers
The modern feature extraction workflow

6 sections 16 notebooks

Part 3

Model Development

The ML Pipeline

From inference to prediction
Regularized regression
Predicting direction with logistic regression
Interpreting models with SHAP
Quantifying predictive uncertainty

7 sections 20 notebooks

Advanced Models for Tabular Data

From decision trees to ensembles
Gradient boosting machines
Deep learning alternatives for tabular data
Advanced hyperparameter tuning with Optuna
Model explainability with SHAP

7 sections 12 notebooks

Deep Learning for Time Series

Recurrent networks and their limits
N-BEATS and explicit decomposition
Attention for time series
Linear baselines versus transformers
Modern transformer variants

10 sections 18 notebooks

Latent Factor Models

Making the case for latent factors
Extracting latent factors with PCA
Eigenportfolios for equity strategies
Decoding the yield curve
Bridging economics and statistics with advanced models

9 sections 19 notebooks

Causal Machine Learning

From theory to estimation
Identification and validation
Validation and refutation
Isolating factor effects with DML
Measuring event impact with Bayesian structural time-series

8 sections 19 notebooks

Part 4

Strategy Implementation

Strategy Simulation

Backtesting as falsification
Specifying your backtest protocol
Vectorized and event-driven backtesting
An auditable non-ML baseline
Understanding performance metrics

8 sections 14 notebooks

Portfolio Construction

Defining the allocation problem
A portfolio construction workflow
Portfolio evaluation metrics
Defining baseline allocators
Mean-variance optimization and the Markowitz curse

9 sections 15 notebooks

Transaction Costs

Where costs enter the ML4T workflow
A cost taxonomy for practitioners
The microstructure regime link
Baseline backtesting cost models
Execution algorithms as controls

9 sections 12 notebooks

Risk Management

Turning your backtest winner into a tradable system
A practical risk taxonomy for quant strategies
Measuring the tail – VaR and CVaR
Drawdowns, path risk, and time to recovery
Decomposing factor, sector, and macro exposures

9 sections 13 notebooks

Strategy Synthesis

First-pass results across nine case studies
Setup and feature evaluation
Signal quality and prediction uncertainty
From signals to strategies
Portfolio allocation across the case studies

10 sections 6 notebooks

Part 5

Advanced AI

Reinforcement Learning

The sequential decision-making paradigm in finance
Financial markets as Markov decision processes
Core algorithms – From DQN to actor-critic
Application I – Optimal trade execution
Application II – Market making

9 sections 8 notebooks

RAG for Financial Research

From feature extraction to generation
Grounding LLMs with retrieval-augmented generation
Intelligent document ingestion
Domain-specific embeddings
Hybrid retrieval and vector databases

10 sections 8 notebooks

Knowledge Graphs

When relational structure justifies a graph
Constructing financial knowledge graphs
Deterministic relational reasoning with graph RAG
From graphs to machine learning features
Correlations and portfolios in financial networks

8 sections 10 notebooks

Autonomous Agents

From prediction functions to agentic workflows
Cognitive architectures – How agents reason
Agent memory – State, persistence, and replay
Tool integration – Contracts, controls, and context engineering
The engineering stack – Frameworks and migration

11 sections 10 notebooks

Part 6

Production

Live Trading Systems

The unified research-to-production framework
Integrating with Interactive Brokers
Integrating with Alpaca
QuantConnect and managed platforms
Order lifecycle management

8 sections 17 notebooks

MLOps and Governance

Two sources of live trading failure
Performance monitoring
Drift detection
Safe model updates
Circuit breakers and safety

7 sections 7 notebooks

The Systematic Edge

The systematic edge – From techniques to philosophy
The modern quant career
Building a learning practice
Navigating the frontiers – Quantum, DeFi, and ethical AI
Building your path forward