BehaviorBench (Behavioral Traces) - Trade Amount Error (ProfileGen): leaderboard
Metric: Median absolute error of the predicted trade amount (paper units, unbounded); history interface ProfileGen (a released structured profile summarizing the wallet's pre-target behavior); 2,000 pseudonymous prediction-market wallets reconstructed from public Polymarket metadata and Polygon on-chain logs, about 59,560 chronologically held-out test transactions; unparsable outputs count as failures; context-only baseline 31.79; lower is better. Source: arxiv.org. Saturation forecast: Around June 2027. 11 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Llama 4 Scout Instruct | 10.3 |
| 2 | Gemma 3 12B (IT) | 10.99 |
| 3 | GPT-5.4 Mini | 12.96 |
| 4 | Claude 3.7 Sonnet | 14.12 |
| 5 | GPT-5.4 | 14.61 |
| 6 | Gemma 3 27B (IT) | 15.18 |
| 7 | Claude Opus 4 | 15.35 |
| 8 | GPT-OSS-20B | 15.5 |
| 9 | NVIDIA-Nemotron-3-Nano-30B-A3B-BF16 | 17.07 |
| 10 | GPT-5.4 Nano | 20.83 |
| 11 | Ministral-3-14B-Instruct-2512 | 35.48 |
Interactive version: theaggregate.ai/benchmark?slug=behaviorbench-behavioral-traces-trade-amount-error-profilegen · How It Works · Data refreshed daily, snapshot 2026-09-29.