Paper: arXiv 2602.00196

Authors: Keywan Christian Rasekhschaffe

Abstract

We study whether generative AI can automate feature discovery in U.S. equities. Using large language models with retrieval-augmented generation and structured/programmatic prompting, we synthesize economically motivated features from analyst, options, and price-volume data. These features are then used as inputs to a tabular machine-learning model to forecast short-horizon returns. Across multiple datasets, AI-generated features are consistently competitive with baselines, with Sharpe improvements ranging from 14% to 91% depending on dataset and configuration. Retrieval quality is pivotal: better knowledge bases materially improve outcomes. The AI-generated signals are weakly correlated with traditional features, supporting combination. Overall, generative AI can meaningfully augment feature discovery when retrieval quality is controlled, producing interpretable signals while reducing manual engineering effort.

Complexity vs Empirical Score

  • Math Complexity: 2.5/10
  • Empirical Rigor: 8.0/10
  • Quadrant: Street Traders — practical and empirical, lighter on theory

Why this score: This paper presents a strong empirical framework for using LLMs (with RAG and DSPy) to generate executable code for tabular feature engineering in U.S. equities. While mathematically lightweight, it is highly rigorous, utilizing strict point-in-time constraints, execution lags, and multiple robust financial datasets.

Research Flowchart

  flowchart TD
    A["Research Goal: Automate Equity Feature Discovery via Generative AI"] --> B["Multi-Source Data Inputs: Analyst, Options, & Price-Volume Data"]
    B --> C["LLM + RAG System with Programmatic Prompting"]
    C --> D["Synthesize Economically Motivated Features"]
    D --> E["Tabular ML Model Return Forecasting"]
    E --> F["Outcomes: +14% to 91% Sharpe Improvement & Weak Correlation with Traditional Features"]