Research map / Large language models / Agents, code and math
AlphaBench: Benchmarking Large Language Models in Formulaic Alpha Factor Mining
Haochen Luo, Ho Tin Ko, Jiandong Chen, David Sun, Yuan Zhang, Chen Liu
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| add chain-of-thought prompting (ablation) | GPT-4.1-Mini · Factor generation | Overall: 0.65 -> 0.59 | Hurt |
| add chain-of-thought prompting (ablation) | Gemini-2.5-Flash · Factor generation | Overall: 0.71 -> 0.69 | Hurt |
| add chain-of-thought prompting (ablation) | DeepSeek-R1-Distill-Qwen-32B · Factor generation | Overall: 0.23 -> 0.32 | Improved |
| add chain-of-thought prompting (ablation) | LLaMA3.1-8b-Instruct · Factor generation | Overall: 0.48 -> 0.51 | Improved |
| add chain-of-thought prompting (ablation) | Gemini-2.5-Flash · Factor ranking | Precision: 0.25 -> 0.14 | Hurt |
| add chain-of-thought prompting (ablation) | Gemini-2.5-Flash · Factor evaluation | Overall: 0.47 -> 0.44 | Hurt |
| add chain-of-thought prompting (ablation) | Gemini-2.5 Pro · Pairwise Selection CSI300 | accuracy: 0.40 -> 0.52 | Improved |
| add chain-of-thought prompting (ablation) | DeepSeek-V3 · Signal Classification CSI300 | accuracy: 0.46 -> 0.40 | Hurt |
| add in-market supervised fine-tuning on CSI300 (ablation) | GPT-4.1-Mini · Pairwise Selection CSI300 | accuracy: 0.48 -> 0.83 | Improved |
| add fine-tuning on CSI300 (ablation) | GPT-4.1-Mini · Pairwise Selection SP500 | accuracy: 0.44 -> 0.64 | Improved |
| add fine-tuning on SP500 (ablation) | GPT-4.1-Mini · Pairwise Selection CSI300 | accuracy: 0.48 -> 0.50 | No clear effect |
| add fine-tuning on CSI300 (ablation) | GPT-4.1-Mini · Signal Classification CSI300 | accuracy: 0.39 -> 0.48 | Improved |
Closest papers by meaning
- AlphaPROBE: Alpha Mining via principled retriveval and on-graph biased evolution (NeurIPS 2026 · similarity 0.89)
- FinBen: A Holistic Financial Benchmark for Large Language Models (NeurIPS 2024 · similarity 0.88)
- AlphaForge: A Framework to Mine and Dynamically Combine Formulaic Alpha Factors (AAAI 2025 · similarity 0.87)
- metabench - A Sparse Benchmark of Reasoning and Knowledge in Large Language Models (ICLR 2025 · similarity 0.86)
- MASS: Mathematical Data Selection via Skill Graphs for Pretraining Large Language Models (ICML 2025 · similarity 0.86)
- Efficient Evaluation of LLM Performance with Statistical Guarantees (NeurIPS 2026 · similarity 0.86)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
