Econometric Algorithmic Trading using Machine Learning With Python: From Stationarity to Execution Building Tradable Signals with Econometrics and ML (Richman Computational Economics) - Softcover

Inhaltsangabe

Build institutional-grade trading signals with econometrics and machine learning—end to end in Python. This is a dense, model-first handbook for quants who want repeatable alpha, robust risk, and friction-aware execution without the fluff. Every chapter moves from rigorous theory to end-of-chapter multiple-choice questions, and finishes with full, runnable Python code demonstrations you can adapt to your pipeline today.

Inside, you will learn to

• Stabilize noisy financial series, differentiate fractionally, and decide when to model levels vs. spreads.
• Forecast returns and volatility with ARIMA/ARFIMA, HAR/MIDAS, and advanced GARCH variants under fat tails and leverage.
• Model multi-asset risk with DCC/BEKK and factor structures for scalable portfolio construction.
• Extract stationary spreads and design error-correction trading rules with VECM and threshold dynamics.
• Track time-varying betas with Kalman filters; decode regimes with Markov-switching; manage breaks with structural change tests.
• Quantify microstructure effects, estimate efficient prices, and model order flow and jumps via Hawkes/ACD.
• Build high-dimensional alpha models using lasso/elastic net, boosting (XGBoost/LightGBM/CatBoost), kernels, GPs, and deep nets.
• Capture nonlinear dependence and tail risk with copulas and EVT; forecast quantiles and expected shortfall for risk-aware sizing.
• Identify causal effects with D-i-D, IV, RDD, and double ML; target policy to tradable subpopulations.
• Allocate across signals with online learning and bandits; trade under realistic impact with Almgren–Chriss and propagator models.
• Deploy RL for execution and market making, with proper off-policy evaluation and conservative objectives.
• Evaluate your edge correctly with Diebold–Mariano, MCS, Reality Check, SPA, and deflated Sharpe to avoid data snooping.

Who it’s for

• Quant researchers, portfolio managers, and traders upgrading from ad hoc heuristics to statistically defensible, production-ready models.
• Data scientists entering quantitative finance who need a rigorous bridge from ML theory to tradable implementation.
• Graduate students and practitioners seeking a compact, code-complete reference for model-driven trading.

How each chapter works

• Theory: assumptions, identification, estimation, diagnostics, and forecasting.
• Checkpoint: multiple-choice questions to test comprehension and common pitfalls.
• Practice: full Python code demonstrations—from data prep and estimation to validation, backtesting, and interpretation.

Note: Educational content only. Markets carry risk; no strategy guarantees profits.

Die Inhaltsangabe kann sich auf eine andere Ausgabe dieses Titels beziehen.