BacktestBench - SQL Execution Accuracy: leaderboard

The SQL generation stage of BacktestBench (2026), a quantitative-backtesting benchmark: the share of SQL queries a model writes to fetch market data for a natural-language strategy whose result set matches the reference result set, ignoring row order. Percent; higher is better.

Metric: Execution accuracy (%) of the intermediate SQL: the share of generated queries whose result set matches the ground-truth result set ignoring order, 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: Estimated already saturated. 23 models tracked. Updated 7 Oct 2026.

Top models

#ModelSettingScoreOverall rank
1Gemini 3 Pro98.7#145
2Seed 1.897.45#252
3Qwen 3 235B A22B 2507 (Thinking)97.11#466
4GPT-OSS-120B96.58#704
5Kimi K2 (Thinking)96.06
6MiniMax-M2.195.02#575
7GLM-4.794.86#380
8Qwen 3 Next 80B A3B (Thinking)94.79#594
9Qwen 3 Max94.73#415
10Qwen 3 Coder Plus94.03#720
11DeepSeek V3.292.86#419
12Qwen 3 30B A3B 2507 (Thinking)92.86#764
13Qwen 3 32B90.72#972
14GPT-OSS-20B90.56#1183
15Seed-OSS-36B-InstructNon-reasoning87.96#851

No result here: #1 Claude Fable 5.2, #3 Gemini 4 Neon, #4 GPT-6.2 Astra.

Interactive version: theaggregate.ai/benchmark?slug=backtestbench-sql-execution-accuracy · How It Works · Data refreshed daily, snapshot 2026-11-21.