PortBench-QA - Regime Detection: leaderboard
PortBench-QA's regime template (Zhao, Chen and Su, 2026): the model labels the market regime (bull, bear, sideways or crisis) from prices, macro series and news and gives the implied increase, decrease or hold for six asset classes, scored half on the regime and half on the directions. 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 (%; market regime plus the implied allocation direction for six asset classes, partial credit, 50 test questions per model). Source: arxiv.org. Saturation forecast: Around February 2027. 10 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Qwen 3.6 Plus (Thinking) | 76.8 | #217 (Qwen 3.6 Plus) |
| 2 | Qwen 3.6 35B A3B (Thinking) | 76.3 | #461 (Qwen 3.6 35B A3B) |
| 3 | Qwen 3.7 Max (Thinking) | 74.2 | #142 (Qwen 3.7 Max) |
| 4 | GLM-5.1 (Thinking) | 73.8 | #175 (GLM-5.1) |
| 5 | Kimi K2.6 (Thinking) | 32 | #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-regime-detection · How It Works · Data refreshed daily, snapshot 2026-11-12.