PolyBench

Size

38,666 binary markets across 4,997 events; 36,165 predictions from 7 models (2026-02-06 to 2026-02-12)

Public split

Pipeline code on GitHub; dataset via a OneDrive link in the README

Grading

Directional accuracy, Confidence-Weighted Return (CWR), APY and Sharpe from simulated order-book execution

Licence

Not stated

Submission

none

Products allowed

Not stated

Latest version

arXiv v1 (2026-04-03)

What it does not cover

Equities and other asset markets; the evaluation window is one week

Each score shows who reported it: the benchmark's authors (quoted from their paper), the benchmark's official board, or Moneda (run with the benchmark's own grader and rules). We do not copy other aggregators.

1. 17.6MiMo-V2-Flash17.6
2. 6.2Gemini-3-Flash6.2
3. -9.2Trinity-Large-9.2
4. -11.4deepseek-v3.2-11.4
5. -12.1 Grok-4.1-Fast-12.1
6. -17.8 GPT-OSS 120B-17.8
7. -25.2MiniMax-M2.5-25.2

1. 1MiMo-V2-Flash17.617.6
2. 2Gemini-3-Flash6.26.2
3. 3Trinity-Large-9.2-9.2
4. 4deepseek-v3.2-11.4-11.4
5. 5 Grok-4.1-Fast-12.1-12.1
6. 6 GPT-OSS 120B-17.8-17.8
7. 7MiniMax-M2.5-25.2-25.2

PolyBench leaderboard (Confidence-Weighted Return (CWR) %, higher is better)

Rank	Model	Organisation	Confidence-Weighted Return (CWR) %	Setting	Source
1	MiMo-V2-Flash	Xiaomi	17.6	base lot $10, 2026-02-06 to 2026-02-12	Paper Table 5 Reported by the authors, 3 Apr 2026
2	Gemini-3-Flash	Google	6.2	base lot $10	Paper Table 5 Reported by the authors, 3 Apr 2026
3	Trinity-Large	Not stated	-9.2	base lot $10	Paper Table 5 Reported by the authors, 3 Apr 2026
4	deepseek-v3.2	DeepSeek	-11.4	base lot $10	Paper Table 5 Reported by the authors, 3 Apr 2026
5	Grok-4.1-Fast	xAI	-12.1	base lot $10	Paper Table 5 Reported by the authors, 3 Apr 2026
6	GPT-OSS 120B	OpenAI	-17.8	base lot $10	Paper Table 5 Reported by the authors, 3 Apr 2026
7	MiniMax-M2.5	MiniMax	-25.2	base lot $10	Paper Table 5 Reported by the authors, 3 Apr 2026

• Paper
• Code

Last checked 1 Nov 2026. Something out of date or wrong? Send a correction.