StakeBench - Revealed Side: leaderboard
Revealed-side identification in StakeBench (2026): given a comment by a user who holds a prediction-market position, infer which side that position is on. The score is strict accuracy in percent, abstentions counted wrong, macro-averaged over topic and platform splits; higher is better.
Metric: Strict accuracy (%) on G2: which side (YES or NO) a positioned commenter holds, abstaining counted as wrong, Polymarket and Manifold comments of resolved markets, macro-average over 18 topic and platform splits, greedy decoding; higher is better. Source: arxiv.org. Saturation forecast: Around 2031. 15 models tracked. Updated 7 Oct 2026.
Top models
| # | Model | Setting | Score | Overall rank |
|---|---|---|---|---|
| 1 | Llama 3.2 3B Instruct | 53.4 | #7241 | |
| 2 | Fino1-8B | 52.4 | #6784 | |
| 3 | FinMA-30B | 47.6 | ||
| 4 | GPT-5.5 | 44.8 | #54 | |
| 5 | Fin-Chat-7B | 44.5 | ||
| 6 | Qwen 3 32B | Non-reasoning | 43 | #972 |
| 7 | Qwen 3 14B | Non-reasoning | 42.3 | #1269 |
| 8 | Gemma 2 9B | 39.7 | #3201 | |
| 9 | Gemini 2.5 Flash | 38.4 | #463 | |
| 10 | Claude Haiku 4.5 | 38.1 | #583 | |
| 11 | Fino1-14B | 36.3 | ||
| 12 | DeepSeek R1 Distill Llama 8B | 34.5 | #7059 | |
| 13 | Qwen 3 30B A3B | Non-reasoning | 34.4 | #1156 |
| 14 | Mistral 7B Instruct (v0.3) | 33.5 | #6750 | |
| 15 | Qwen 3 8B | Non-reasoning | 32.5 | #1873 |
No result here: #1 Claude Fable 5.2, #3 Gemini 4 Neon, #4 GPT-6.2 Astra.
Interactive version: theaggregate.ai/benchmark?slug=stakebench-revealed-side · How It Works · Data refreshed daily, snapshot 2026-11-21.