Rigged Backtest Audit - Clean-Control False Positive Rate: leaderboard
Metric: False positive rate on clean controls (%; share of the 48 clean backtests the model flags as flawed; clean-aware prompt on the code surface of a 96-item paired set, 48 flawed backtests over eight flaw classes and 48 matched clean controls, deterministic scorer). Source: arxiv.org. Saturation forecast: Estimated already saturated. 4 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Gemini 2.5 Flash Lite | 0 |
| 2 | DeepSeek V4 Flash | 0 |
| 3 | GPT-4.1 Mini | 16.7 |
| 4 | GPT-4o Mini | 79.2 |
Interactive version: theaggregate.ai/benchmark?slug=rigged-backtest-audit-clean-control-false-positive-rate · How It Works · Data refreshed daily, snapshot 2026-09-26.