From Natural Language to Executable Option Strategies via Large Language Models

Battle Count: 50

Highly relevant to quantitative trading. The paper directly addresses automated option strategy design from natural language, a core task in derivatives trading. It introduces a practical framework (OQL) that bridges the gap between trader intent and executable strategies, with empirical validation through backtesting. The neuro-symbolic approach ensures logical validity and reduces hallucinations critical in high-stakes financial settings. The work covers directional, volatility, and income/hedging strategies with Greeks-based constraints, directly applicable to options desk operations and systematic derivatives trading.

Moderate to high complexity. Requires: (1) LLM integration for semantic parsing with OQL grammar injection, (2) A deterministic execution engine with AST parsing, vectorized filtering over option chains, Cartesian product leg assembly, and aggregate constraint evaluation, (3) Option chain data infrastructure with Greeks computation (BSM model), (4) Backtesting framework. The OQL grammar is well-defined (EBNF), and the backend is Python-based (NumPy/Pandas), but the full pipeline requires significant engineering for production deployment. The role-based schema and scoped filtering add logical complexity.

The paper provides detailed OQL grammar (EBNF), formal definitions, dataset construction methodology, and evaluation metrics. However, no GitHub repository is mentioned. The dataset construction relies on human expert annotation which may be difficult to reproduce exactly. The backend is Python-based (NumPy/Pandas) which aids reproducibility. Market data from 2025 is used but specific data sources are not fully detailed in the extract.