Equity Return Recall on Blind U.S. mkt excess monthly close-to-close
100Return Parse MetricOpus 4.7
Evaluation Results
| Method | Links | ||||
|---|---|---|---|---|---|
| Opus 4.7rho_FF=1.00, n (sample size)=40, Answer Format=Variant-A2026.05 | 100 | 68 | 0.954 | 98 | |
| GPT-5.4rho_FF=1.00, n (sample size)=40, Answer Format=Variant-A2026.05 | 100 | 33 | 0.77 | 85 | |
| GPT-5.4-minirho_FF=1.00, n (sample size)=40, Answer Format=Variant-A2026.05 | 100 | 10 | 0.54 | 70 | |
| GPT-5.4-nanorho_FF=1.00, n (sample size)=40, Answer Format=Variant-A2026.05 | 100 | 5 | 0.18 | 65 | |
| DeepSeek-V3.2rho_FF=1.00, n (sample size)=40, Answer Format=Variant-A2026.05 | 100 | 15 | 0.42 | 65 | |
| Llama-3.3-70Brho_FF=1.00, n (sample size)=40, Answer Format=Variant-A2026.05 | 100 | 10 | 0.08 | 60 | |
| Sonnet 4.6rho_FF=1.00, n (sample size)=40, Answer Format=Variant-A2026.05 | 62 | 20 | 0.92 | 100 | |
| Llama-3.1-8Brho_FF=1.00, n (sample size)=40, Answer Format=Variant-A2026.05 | 53 | 0 | 0.13 | 33 |