Chapters

Explore the complete ML3T curriculum. Each chapter includes hands-on code examples, quizzes, and downloadable resources.

Part 1

Foundations of ML-Driven Trading

5 chapters

Process Is Your Edge

Getting started with machine learning in financial markets

• Identify the recent structural breaks and market shocks that render older, static trading models obsolete

• Describe the seven distinct stages of the ML4T Workflow, from ideation to live monitoring

• Explain how the workflow addresses real-world constraints imposed by team structure, regulations, and research economics

Part 2

The Core Quantitative Workflow

7 chapters

Engineering Alpha Factors

• Distinguish between raw data, features, alpha factors, and signals.

• Explain the importance of a hypothesis-driven approach to avoid the pitfalls of the "factor zoo."

• Use high-performance libraries like Polars and TA-Lib to create factors efficiently.

Evaluating Alpha Factors

• Apply a full factor analysis using the Alphalens library to measure a factor's predictive power and risk characteristics using Information Coefficient (IC),

• Explain the danger of multiple testing and apply the Benjamini-Hochberg procedure to control the False Discovery Rate (FDR)

• Calculate and interpret the Deflated Sharpe Ratio to assess the risk of backtest overfitting.

Simulating Performance: The Backtesting Engine

• Explain why a "falsification mindset" is critical for avoiding common backtesting biases like look-ahead bias and survivorship bias

• Analyze the impact of data quality issues (survivorship bias, lookahead bias, synchronization errors) on downstream modeling performance and backtest validity

• Select the appropriate backtesting architecture (vectorized vs. event-driven) based on the research objective, simulation fidelity requirements, and computational constraints

Generating Signals with Linear Models

• Differentiate between inferential and predictive modeling and articulate the role of the bias-variance tradeoff.

• Justify the selection of a regularization technique (Ridge, LASSO, Elastic Net) based on the underlying investment hypothesis and data characteristics

• Frame a return forecasting problem as a classification task and build a regularized logistic regression model.

Portfolio Construction & Management

• Articulate the "Markowitz Curse" and explain why classical Mean-Variance Optimization often fails out-of-sample.

• Apply **Ledoit-Wolf shrinkage** to build robust covariance matrices.

• Construct portfolios using the graph-based **Hierarchical Risk Parity (HRP)** algorithm.

Modeling Transaction Costs & Market Impact

• Decompose total trading costs into their explicit and implicit components, using the **Implementation Shortfall** framework

• Explain the microstructural origins of market impact, including the distinction between **temporary** and **permanent** price impact

• Implement and critically evaluate the industry-standard benchmark execution algorithms, **TWAP** and **VWAP**

Quantitative Risk Management

• Decompose portfolio risk into systematic (factor-driven) and idiosyncratic components.

• Analyze the assumptions and limitations of VaR and CVaR models, particularly during periods of market stress.

• Explain the Model Risk Management (MRM) framework as defined by regulatory guidance (SR 11-7) and its application to quantitative trading strategies.

Part 3

Language, Knowledge, and Reasoning

3 chapters

From Text to Features: The Transformer Breakthrough

• Articulate the historical evolution of NLP feature engineering, explaining the specific limitations of each major paradigm (bag-of-words, word embeddings, and transformers)

• Explain the core conceptual breakthrough of the self-attention mechanism, including the intuition behind Query, Key, and Value matrices

• Implement a best-practice, three-stage workflow: leveraging a general pre-trained model, adapting it to the financial domain via fine-tuning, and extracting features for downstream ML

The Reasoning Engine: Generative AI in Finance

• Implement a RAG-based system that performs multi-hop reasoning over financial documents with verifiable citations.

• Analyze the sources of error in a RAG pipeline (retrieval relevance, reranking effectiveness, synthesis accuracy) to diagnose performance bottlenecks.

• Select and justify the critical components of a financial RAG system (embedding models, vector databases, rerankers) based on specific use case requirements.

Knowledge Graphs & Graph Neural Networks

• Articulate the limitations of vector-only RAG for complex financial queries and explain why Knowledge Graphs provide superior structured reasoning

• Design a financial knowledge graph schema with appropriate nodes, relationships, and properties.

• Implement a modern, end-to-end pipeline to construct a KG from unstructured text (like SEC filings) using an LLM-based extraction framework

Part 4

Advanced Predictive Modeling

6 chapters

Classical and Bayesian Time Series Models

• Execute a complete diagnostic workflow on a time series, testing for stationarity, autocorrelation, and structural breaks

• Build, validate, and critically assess ARIMA-GARCH models, understanding their primary utility as risk forecasting tools

• Develop Bayesian time series models using PyMC to move beyond point estimates to full posterior distributions

Deep Learning for Time Series

• Implement and train core deep learning architectures (MLPs, 1D-CNNs, LSTMs/GRUs) for time series forecasting, comparing their performance on a financial benchmark.

• Critically evaluate the suitability of core deep learning architectures for specific financial time series tasks.

• Implement and compare decomposition-based (e.g., N-BEATS) and attention-based (e.g., Transformer) architectures on a financial forecasting task, analyzing their respective trade-offs in accuracy, interpretability, and computational cost.

Advanced Models for Tabular Data

• Apply and critically compare the industry-standard GBM libraries: XGBoost, LightGBM, and CatBoost.

• Analyze complex feature interactions and dependencies within a GBM using advanced SHAP analysis (interaction values, dependence plots).

• Critique the limitations of current XAI methods (e.g., instability of explanations, correlation vs. causation) and select appropriate techniques for different stakeholders.

Discovering Latent Factors: From PCA to Data-Driven Asset Pricing

• Articulate the methodological crisis of the "Factor Zoo" and the motivation for discovering latent factors.

• Implement PCA to extract statistical risk factors (eigenportfolios) from equity returns.

• Apply PCA to decompose the yield curve dynamics into level, slope, and curvature components.

Causal Inference for Trading Strategies

• Frame causal inference as a primary tool for risk management against spurious models.

• Situate advanced methods (DML, BSTS) within the broader context of established quasi-experimental techniques like DiD and RDD.

• Implement Double ML to estimate the causal effect of a continuous factor, with practical guidance on nuisance model tuning.

Quantifying Model Uncertainty: Calibration and Conformal Prediction

• Articulate the crucial difference between aleatoric (data) uncertainty and epistemic (model) uncertainty and why it matters for trading.

• Implement Deep Ensembles and MC Dropout as practical, scalable methods for estimating epistemic uncertainty.

• Generate statistically guaranteed prediction intervals for any model using Conformalized Quantile Regression (CQR).

Part 5

Autonomous Agents and Simulation

NEW in 3rd Edition

3 chapters

RL for Optimal Execution, Market Making, and Hedging

• Formulate financial decision-making problems as Markov Decision Processes with appropriate state representations, action spaces, and reward functions

• Implement a PPO-based optimal execution agent that minimizes implementation shortfall.

• Design a market-making agent that balances spread capture against inventory risk.

Building Autonomous Financial Agents

• Design robust agentic reasoning flows using patterns like **ReAct**, **Tree of Thoughts**, and **Reflexion**.

• Implement hierarchical memory systems that provide agents with short-term context and persistent long-term knowledge

• Integrate and orchestrate a diverse toolkit of financial APIs and proprietary functions.

Generative Models for Synthetic Data

• Articulate the "Quant's Dilemma" and how synthetic data helps solve backtest overfitting.

• Understand the evolution of generative models from GANs (TimeGAN) to Diffusion Models (for time series) and LLMs (for tabular data).

• Implement a diffusion model to generate realistic financial time series that capture key "stylized facts" (e.g., volatility clustering, heavy tails).

Part 6

From Research to Production

NEW in 3rd Edition

3 chapters

From Research to Reality with Live Trading Systems

• Design a resilient, event-driven trading system architecture using Apache Kafka and microservices.

• Develop secure and reliable integrations with brokerage APIs like Alpaca and Interactive Brokers.

• Deploy trading systems on cloud platforms (AWS/GCP) using Infrastructure as Code and security best practices.

MLOps and Production Governance

• Design a practical, 6-stage model lifecycle (from research to retirement) suited for a solo trader or small team.

• Use an open-source feature store (Feast) to define features and prevent training-serving skew.

• Employ MLflow to track trading experiments and manage a Model Registry with Staging and Production stages.

The Systematic Edge: Your Path Forward

• Internalize the 7-Stage Workflow as a career-long mental model for value creation.

• Navigate the modern quant career landscape, including the rise of quantamental roles, and make strategic decisions about specialization.

• Build a personalized continuous learning system that includes community engagement and personal brand development.

Unlock Free Tier 2 Content

Get detailed section summaries, code examples, best practices, and case study introductions delivered to your inbox. No purchase required.

Section summaries
Code examples
Best practices
Case studies

Subscribe Now

Join 5,000+ readers • Free forever • Unsubscribe anytime

Total Chapters

3rd Edition (28 chapters)

New Content

10+

New chapters in 3rd edition

Reading Time

~10h

Total estimated time

Chapters

Foundations of ML-Driven Trading

Process Is Your Edge

Defining the Trading Strategy

The Financial Data Universe

Market Data and Microstructure

Fundamental and Alternative Data

The Core Quantitative Workflow

Engineering Alpha Factors

Evaluating Alpha Factors

Simulating Performance: The Backtesting Engine

Generating Signals with Linear Models

Portfolio Construction & Management

Modeling Transaction Costs & Market Impact

Quantitative Risk Management

Language, Knowledge, and Reasoning

From Text to Features: The Transformer Breakthrough

The Reasoning Engine: Generative AI in Finance

Knowledge Graphs & Graph Neural Networks

Advanced Predictive Modeling

Classical and Bayesian Time Series Models

Deep Learning for Time Series

Advanced Models for Tabular Data

Discovering Latent Factors: From PCA to Data-Driven Asset Pricing

Causal Inference for Trading Strategies

Quantifying Model Uncertainty: Calibration and Conformal Prediction

Autonomous Agents and Simulation

RL for Optimal Execution, Market Making, and Hedging

Building Autonomous Financial Agents

Generative Models for Synthetic Data

From Research to Production

From Research to Reality with Live Trading Systems

MLOps and Production Governance

The Systematic Edge: Your Path Forward

Unlock Free Tier 2 Content