Strategy 3 is a realistic delta-hedging strategy using random walk deltas plus MSFT order book data. At regular intervals, it evaluates net delta and trades a portion to get flat, enforcing a minimum time between hedges. Hedge trades use fill-or-kill limit orders priced from synthetic mid and spread, with a fixed exchange delay before execution.
A synthetic book persists consumed liquidity. We track: Stock position and options delta; Net delta of the combined portfolio; Realized and unrealized P&L from stock trades; Equity and maximum drawdown.
Loading chart data...
Loading chart data...
| Model | pass@3↓ | pass@1↓ | Mean MAE (solved)↓ | Best run MAE↓ | Avg. attempts↓ |
|---|---|---|---|---|---|
| grok-4 | 1.00 | 1.00 | 1,482.33 | 1,013.33 | 1.00 |
| claude-sonnet-4.5 | 1.00 | 1.00 | 16,157.31 | 805.16 | 1.00 |
| qwen3-max | 1.00 | 1.00 | 329,700.43 | 1,087.64 | 1.00 |
| gemini-3-pro-preview | 1.00 | 0.80 | 1,245.48 | 1,013.99 | 1.20 |
| gpt-5.1-codex-max | 1.00 | 0.80 | 8,326.53 | 1,370.72 | 1.20 |
| deepseek-v3.2 | 1.00 | 0.60 | 12,075.80 | 1,127.14 | 1.40 |
| mistral-large-2512 | 0.80 | 0.40 | 137,773.85 | 1,351.55 | 2.00 |
| claude-opus-4.5 | 0.40 | 0.40 | 8,143.48 | 1,370.72 | 1.00 |
| command-a | 0.40 | 0.20 | 21,488.70 | 19,852.90 | 1.50 |
| llama-4-maverick | 0.20 | 0.00 | 20,418.58 | 20,418.58 | 2.00 |
| nova-premier-v1 | 0.00 | 0.00 | – | – | – |
| llama-3.1-nemotron-ultra | 0.00 | 0.00 | – | – | – |