Quantitative Trading Research Overview

Causal Impact Analysis: Uses the 2003 NYSE automation shift as an exogenous instrument to prove Algorithmic Trading (AT) causally improves market quality.

Key Findings: AT narrows spreads, reduces adverse selection, and increases quote informativeness, particularly for large-cap stocks.

Efficiency Role: Investigates HFTs' contribution to price efficiency. HFTs generally facilitate discovery by trading in the direction of permanent price changes and against transitory errors.

Execution: Their liquidity-demanding orders are most effective, while their liquidity-supplying orders often suffer from adverse selection.

Survey: A comprehensive review of Deep Learning architectures (e.g., Transformers) in Time Series Forecasting (TSF) over the last 5 years.

Scope: Proposes a new model taxonomy, reviews applications across finance and energy, and identifies future challenges like computational complexity and long-range forecasting.

Methodology: Proposes a Double Deep Q-Network (DDQN) utilizing a multi-agent GRU architecture for intraday trading in volatile commodity futures (e.g., gold).

Validation: Developed a specific futures market simulator (handling margin/clearing) to prove the model outperforms benchmarks in risk-adjusted returns.

Retail Framework: Focuses on tools for individual investors, converting moving average regression data into classification problems.

Performance: Tested six ML models; the Support Vector Machine (SVM) achieved the best results with 90.31% accuracy and a 29.57% annualized return rate.