QuantCode-Bench: A Benchmark for Evaluating the Ability of Large Language Models to Generate Executable Algorithmic Trading Strategies

TL;DR

QuantCode-Bench is a new benchmark designed to evaluate large language models' ability to generate executable algorithmic trading strategies from natural language descriptions, emphasizing domain-specific logic and API usage.

Contribution

The paper introduces QuantCode-Bench, a comprehensive benchmark with 400 tasks for assessing LLMs in generating trading strategies, and analyzes current models' limitations in this domain.

Findings

Current models struggle with operationalizing trading logic and API usage.

Success depends on aligning natural language, financial logic, and data behavior.

Most failures are not due to syntax but to semantic and operational errors.

Abstract

Large language models have demonstrated strong performance on general-purpose programming tasks, yet their ability to generate executable algorithmic trading strategies remains underexplored. Unlike standard code benchmarks, trading-strategy generation requires simultaneous mastery of domain-specific financial logic, knowledge of a specialized API, and the ability to produce code that is not only syntactically correct but also leads to actual trades on historical data. In this work, we present QuantCode-Bench, a benchmark for the systematic evaluation of modern LLMs in generating strategies for the Backtrader framework from textual descriptions in English. The benchmark contains 400 tasks of varying difficulty collected from Reddit, TradingView, StackExchange, GitHub, and synthetic sources. Evaluation is conducted through a multi-stage pipeline that checks syntactic correctness, successful backtest execution, the presence of trades, and semantic alignment with the task description using an LLM judge. We compare state-of-the-art models in two settings: single-turn, where the strategy must be generated correctly on the first attempt, and agentic multi-turn, where the model receives iterative feedback and may repair its errors. We analyze the failure modes across different stages of the pipeline and show that the main limitations of current models are not related to syntax, but rather to the correct operationalization of trading logic, proper API usage, and adherence to task semantics. These findings suggest that trading strategy generation constitutes a distinct class of domain-specific code generation tasks in which success requires not only technical correctness, but also alignment between natural-language descriptions, financial logic, and the observable behavior of the strategy on data.