High-performance Algorithmic Trading using Machine Learning Building automated t.epub
大小:(49.31 MB)

只需: RMB 12元  马上下载

内容特别丰富，全部矢量文字！

Machine learning has rapidly become a transformative tool in algorithmic trading, offering capabilities that go far beyond traditional methods such as econometrics, technical analysis, stochastic calculus, portfolio optimization, and signal processing. While these established approaches have long been staples in quantitative finance, they often rely on rigid assumptions and handcrafted rules. In contrast, machine learning enables systems to learn directly from data—discovering patterns, adapting to market dynamics, and building predictive models with minimal human intervention. The idea that examples—training sets—could be transformed into models automatically is revolutionary.
This resource was written to fill a gap I observed repeatedly: the space between the first steps taken by beginners in machine learning for trading, and the more advanced, often inaccessible, expertise found in academic or institutional settings. Many newcomers begin by trying to classify returns—often unsuccessfully—due to a lack of experience with alternative prediction targets or a deeper understanding of feature engineering. My goal is to bridge that gap by introducing modern machine learning techniques that are both powerful and practical. Whether you're working on your own or within a small team, this resource focuses on approaches that are computationally efficient, applicable in real trading contexts, and capable of delivering measurable results.
This resource is designed as a hands-on journey through the key techniques of machine learning applied to real-world trading. It starts with the foundations of algorithmic strategy design, then progressively expands into supervised learning, unsupervised models, pattern mining, NLP for financial text, and ends with portfolio construction using advanced ML techniques.  The focus is entirely practical—mathematical derivations have been intentionally excluded in favor of code, tools, and examples—making the material accessible without sacrificing technical depth.
You will learn how to apply quantamental methods by integrating accounting data into predictive models, detect structural changes in time series and extract rules automatically, work with alternative and unstructured data, and engineer features that go far beyond basic OHLC inputs, filter out market noise while preserving signal, and construct volume- or volatility-based bars and leverage recent breakthroughs in AutoML and low-code ML, using tools like H2O and Microsoft FLAML. Each chapter combines clear explanations, ready-to-run code, and use cases that reflect real trading problems and constraints.
Chapter 1: Algorithmic Trading and Machine Learning in a Nutshell- This chapter introduces systematic trading strategies, key players in the industry, and how machine learning fits into modern trading systems.
Covers traditional approaches and contrasts them with ML-driven pipelines.
Chapter 2: Data Feed, Backtests, and Forward Testing- This chapter explores how to acquire macroeconomic and fundamental data via APIs, and how to prepare data for machine learning workflows. Introduces forward testing concepts and time-aware data pipelines.
Chapter 3: Optimizing Trading Systems, Metrics, and Automated Reporting- This chapter covers feature engineering, metric selection, model boosting, and creating automatic performance reports using QuantStats and other tools.
Chapter 4: Implement Trading Strategies- This chapter focuses on event-driven strategy implementation using Backtrader. Includes end-to-end ML strategy deployment, risk management, and performance evaluation.
Chapter 5: Supervised Learning for Trading Systems- This chapter covers the classification and regression algorithms relevant for trading. Emphasizes model selection, metric interpretation, and prediction targets.
Chapter 6: Improving Model Capability with Features- This chapter explores advanced feature creation: technical indicators, entropy, PCA, UMAP, tree-based features, and feature selection techniques.
Chapter 7: Advanced Machine Learning Models for Trading - This chapter presents ensemble methods (boosting, bagging, stacking), kernel-based regressors, and online learning strategies adapted to financial time series.
Chapter 8: AutoML and Low-Code for Trading Strategies- This chapter shows how to use AutoML frameworks (H2O, FLAML) to build efficient models without manual tuning. Focuses on workflow automation and reproducibility.
Chapter 9: Unsupervised Learning Methods for Trading- This chapter introduces change point detection and clustering for uncovering hidden patterns and structural shifts in financial series.
Chapter 10: Unsupervised Learning with Pattern Matching- This chapter teaches how to use recurrence plots, distance matrices, and matrix profiles to identify motifs and anomalies in time series data.
Chapter 11: Trading Signals from Reports and News- This chapter combines NLP and embeddings to extract trading signals from unstructured text. Covers GloVe, UMAP, similarity graphs, and HRP-based portfolio construction.
Chapter 12: Advanced Unsupervised Learning, Anomaly Detection, and Association Rules- This chapter explores unsupervised anomaly detection, projection-based clustering, and association rule mining for discovering hidden market structures.