Strategy 2 is a pairs trading mean-reversion on Coke and Pepsi. At each book update, the model calculates a mid price for both symbols and forms a spread as a linear combination. A rolling history of the spread drives the mean and z-score signals across 3 states: flat, long-spread, or short-spread. When the z-score exceeds the entry threshold, it buys one leg and sells the other; positions flatten inside the exit threshold.
Additional features: A cooldown to avoid rapid re-entry; A shared capital account; Synthetic books and VWAP tracking per symbol; Immediate-or-cancel limit orders priced from synthetic mid and spread.
Loading chart data...
| Model | pass@3↓ | pass@1↓ | Mean MAE (solved)↓ | Best run MAE↓ | Avg. attempts↓ |
|---|---|---|---|---|---|
| gemini-3-pro-preview | 1.00 | 1.00 | 52.22 | 52.22 | 1.00 |
| gpt-5.1-codex-max | 1.00 | 1.00 | 136.97 | 89.43 | 1.00 |
| claude-sonnet-4.5 | 1.00 | 0.80 | 205.36 | 70.82 | 1.20 |
| mistral-large-2512 | 1.00 | 0.80 | 267.21 | 135.24 | 1.40 |
| qwen3-max | 1.00 | 0.60 | 572,587,991.97 | 100.14 | 1.40 |
| grok-4 | 0.80 | 0.40 | 573.78 | 119.25 | 1.75 |
| deepseek-v3.2 | 0.60 | 0.60 | 132.10 | 125.87 | 1.00 |
| llama-4-maverick | 0.40 | 0.20 | 3,202.11 | 131.92 | 2.00 |
| llama-3.1-nemotron-ultra | 0.40 | 0.20 | 14,010.61 | 135.26 | 2.00 |
| claude-opus-4.5 | 0.40 | 0.00 | 138.40 | 85.56 | 2.50 |
| command-a | 0.40 | 0.00 | 7,335.00 | 6,738.19 | 3.00 |
| nova-premier-v1 | 0.00 | 0.00 | – | – | – |