PortBench-QA - VaR Estimation: leaderboard
PortBench-QA's value-at-risk template (Zhao, Chen and Su, 2026): the model computes a historical-simulation value at risk at a stated confidence level from a trailing return series given in the question. Item score in percent over 50 test-period (2024-2025) questions, zero-shot at temperature 0 with thinking on; higher is better.
Metric: Item score (%; historical-simulation value at risk from a supplied return series, 50 test questions per model). Source: arxiv.org. Saturation forecast: Around February 2027. 10 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Qwen 3.7 Max (Thinking) | 85.9 | #142 (Qwen 3.7 Max) |
| 2 | Qwen 3.6 Plus (Thinking) | 85.8 | #217 (Qwen 3.6 Plus) |
| 3 | GLM-5.1 (Thinking) | 85.5 | #175 (GLM-5.1) |
| 4 | Qwen 3.6 35B A3B (Thinking) | 80.8 | #461 (Qwen 3.6 35B A3B) |
| 5 | Kimi K2.6 (Thinking) | 42.2 | #173 (Kimi K2.6) |
No result here: #1 Claude Fable 5.2, #2 Gemini 4 Carbon, #5 Claude Opus 5.5.
Interactive version: theaggregate.ai/benchmark?slug=portbench-qa-var-estimation · How It Works · Data refreshed daily, snapshot 2026-11-12.