BacktestBench - Ticker Selection: leaderboard
Metric: Back-test accuracy (%) on the Ticker Selection task family (selecting the stock a natural-language strategy would pick), synthetic BacktestBench test split (the expert-crafted subset excluded), each model driving the AutoBacktest pipeline of the authors (Summarizer, SQL Retriever and Python Coder agents) over a PostgreSQL market database, temperature 0.6; higher is better. Source: arxiv.org. Saturation forecast: Around January 2027. 23 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Gemini 3 Pro | 90.37 |
| 2 | Seed 1.8 | 85.83 |
| 3 | Qwen 3 Max | 85.56 |
| 4 | Qwen 3 235B A22B 2507 (Thinking) | 82.26 |
| 5 | MiniMax-M2.1 | 81.71 |
| 6 | GLM-4.7 | 80.33 |
| 7 | DeepSeek V3.2 | 79.78 |
| 8 | Qwen 3 32B | 75.93 |
| 9 | GPT-OSS-120B | 74 |
| 10 | MiMo-V2-Flash | 73.04 |
| 11 | Qwen 3 30B A3B 2507 (Thinking) | 71.25 |
| 12 | Qwen 3 Next 80B A3B (Thinking) | 67.4 |
| 13 | Kimi K2 (Thinking) | 66.57 |
| 14 | Qwen 3 Coder Plus | 63.41 |
| 15 | Qwen 3 14B | 62.04 |
Interactive version: theaggregate.ai/benchmark?slug=backtestbench-ticker-selection · How It Works · Data refreshed daily, snapshot 2026-10-07.