菜单
Per-regime guardrail tables, correlation/factor/Greeks concepts, heat metric, and the risk decision taxonomy.
用 Codex 或 Claude 帮你安装 复制这段 Prompt，粘贴到 Codex、Claude 或其他助手里，让它检查 Skill 页面并帮你完成安装。
基于 SOC 职业分类
How to read market regime classifier output (HMM + crisis overlay) and what each regime implies for trading.
How to pick single-leg options for directional setups. Long premium only for now — SELL-to-open is hard-blocked at the order tools until multi-leg combos get atomic sizing (R5b/R5c will govern short legs when the block lifts).
Framework for trading around earnings — detecting the earnings lock (R6), picking IV-aware structures, and recording post-earnings lessons.
Hard sizing rules (R1-R7) for every proposed paper order. The order tools re-validate these numerically inside the moomoo MCP server — if you skip them, the order is blocked.
Invalidation discipline, exit rules, and portfolio heat — how to manage risk on open paper positions.
End-to-end order flow — thesis → sizing → paper order. Any step failing aborts the whole flow.
| name | portfolio-risk |
| description | Per-regime guardrail tables, correlation/factor/Greeks concepts, heat metric, and the risk decision taxonomy. |
The Active Risk Agent applies deterministic per-regime caps, then optionally invokes a 3-step LLM council on borderline cases. LLMs can only DOWNGRADE — never widen — the deterministic envelope.
| Cap | BULL | RANGE | TRANSITION | BEAR | CRISIS |
|---|---|---|---|---|---|
heat_pct_max | 6% | 6% | 3% | 3% | 1.5% |
single_underlying_delta_dollar_max | 15% | 15% | 7.5% | 7.5% | 0% |
portfolio_net_delta_max | 35% | 35% | 15% | 15% | 0% |
vega_10vol_max (10vol shock) | 1% | 1% | 0.5% | 0.5% | 0.25% |
gamma_2sigma_max (2σ shock) | 1.5% | 1.5% | 0.75% | 0.75% | 0.25% |
avg_pairwise_corr_max | 0.65 | 0.65 | 0.55 | 0.55 | n/a |
same_expiry_premium_max | 2% | 2% | 1% | 1% | 0% |
factor_exposure_max | 30% | 30% | 20% | 20% | 5% |
APPROVE — all guardrails clear; trade proceeds at requested qty.DOWNSIZE — one or more guardrails exceeded by ≤1.5×; qty multiplied by suggested_qty_factor.VETO — guardrail breach exceeds 1.5× cap, OR real_trading_marker, OR regime forbids new entries.DEFER — data_quality.degradation_level ≥ 2 (missing critical greeks/quotes).Sum of (option premium at risk + stock distance-to-stop × shares) divided by equity. Long premium uses full premium as max loss. Stocks without explicit stops use 5% implicit.
Pairwise Pearson on daily returns over the last 60 sessions. The avg_pairwise_corr_max is checked against the WORST off-diagonal correlation in the open-position set.