Allocation Strategy · in depth仓位策略 · 深度
How much S&P 500 — vs T-bills — to hold right now当前应持有多少标普500 — 相对短期国债
As of截至 Jul 16, 2026 · SPY $750.72
A drawdown / Sharpe engine — not a CAGR-beater回撤 / 夏普引擎 — 并非跑赢年化收益的工具
Recommended allocation right now当前建议仓位
S&P 500 (SPY)标普500 (SPY)
3-mo T-bills (cash carry)3个月国债（现金套息）
Last change上次调整: 2026-05-22 — 66% → 100%
Next de-risk: a sustained score above 25 trims equity to 66%.下一步降险：分数持续高于 25 时，股票权重降至 66%。
Macro de-risk score宏观降险评分 (0–100)
13
LOW — fully invested低 — 满仓
Higher = more risk-off. The score sets the equity weight via graded bands (below); a Fed-put-gated re-deploy buys validated washouts.分数越高越偏避险。该分数通过分级区间（见下）设定股票权重；并由美联储看跌期权门控的再部署逻辑买入经验证的恐慌底。
Fed-put美联储看跌: present在场
Dislocation错位: calm平静
Capitulation投降: 0/3
LLM context / veto overlayLLM 背景 / 否决叠加
LLM context — Fed guidance on_hold
No veto — the mechanical recommendation stands.无否决 —— 以机械建议为准。 Context overlay — NOT in the backtest; advisory + tracked forward.
What is driving the score分数由什么驱动
Each leg adds points to the 0–100 score (renormalised weighted mean). The bar below is the current score, split by contribution — so you can see exactly what is risk-on or risk-off today.各分项为 0–100 分数贡献分值（归一化加权平均）。下方进度条为当前分数，按贡献拆分 — 让你一眼看清今天是什么在加风险或降风险。
Filled width = the score out of 100. Empty = headroom (calm).填充宽度 = 满分100中的得分。空白 = 余量（平静）。
| Risk leg风险分项 | Intensity强度 | Weight权重 | Pub lag发布滞后 | Points贡献分 |
|---|
| Macro-stress drawdown gauge宏观压力回撤计 |
19/100 |
1.0 |
10d |
5.4 |
| Recession risk衰退风险 |
5/100 |
1.0 |
22d |
1.3 |
| Credit stress (HY-OAS widening)信用压力（高收益利差走阔） |
42/100 |
0.5 |
3d |
5.9 |
| Financial conditions (tight & tightening)金融条件（紧且收紧） |
0/100 |
0.5 |
5d |
0.0 |
| Net-liquidity contracting净流动性收缩 |
0/100 |
0.5 |
0d |
0.0 |
| Composite score综合分数 | | | | 13 |
The strategy in action · S&P 500 with the model on top策略实战 · 标普500 叠加模型
Green = fully invested, amber = trimmed, clear = defensive. ▲/▼ mark every allocation change. The strategy equity (blue) is rescaled onto the price axis so you can read protection vs participation directly. Use the range buttons to zoom.绿色 = 满仓，琥珀 = 减仓，空白 = 防守。▲/▼ 标注每次仓位变动。策略净值（蓝）已缩放至价格轴，可直接对比保护与参与度。用范围按钮缩放。
Growth of $1 (log) — vs buy & hold1美元增长（对数）— 对比买入持有
Underwater (drawdown) — vs buy & hold水下回撤 — 对比买入持有
Honest framing: the headline CAGR includes the T-bill carry earned on the de-risked sleeve; net of carry it roughly matches buy & hold. The robust, verified edge is the higher Sharpe and the ~40% shallower max drawdown. It stays mostly invested and WILL lag the index in prolonged bulls — that give-up is the premium paid for crash protection, banked when the cycle breaks.诚实说明：标题年化收益包含降险部分赚取的国债套息；扣除套息后大致与买入持有持平。稳健、已验证的优势是更高的夏普比率与约浅 40% 的最大回撤。它大部分时间在场，在长牛市中必然跑输指数 — 这一让步是为崩盘保护支付的保费，在周期破裂时兑现。
Backtest scorecard · 33.5 years vs buy & hold回测记分卡 · 33.5 年对比买入持有
CAGR
12.51%
buy & hold买入持有 10.81% · ex-carry扣息 12.08%
Sharpe
0.88
buy & hold买入持有 0.65 · CI 0.58–1.2
Max drawdown最大回撤
-36.2%
buy & hold买入持有 -55.2%
Time invested在场时间
98.9%
1.62x turnover/yr换手/年
Sortino
1.14
buy & hold买入持有 0.83
After-tax CAGR (taxable, 35%)税后年化（应税，35%）
8.26%
buy & hold买入持有 10.81% · cum. tax累计税 6.71×
Carry contribution套息贡献
+0.43pp
rate-regime dependent (~0 in ZIRP)取决于利率环境（零利率时≈0）
The rule-book · risk band → equity weight规则手册 · 风险区间 → 股票权重
| Risk band风险区间 | Score分数 | S&P 500 weight标普500权重 | T-bills短债 |
|---|
| Low低 ◀ |
< 25 |
100% |
0% |
| Elevated偏高 |
25–50 |
66% |
34% |
| High高 |
50–75 |
33% |
67% |
| Extreme极端 |
≥ 75 |
0% |
100% |
A new band must hold 5 trading days before it takes effect (hysteresis kills whipsaw). On a buyable washout (capitulation with the Fed-put present) the model over-rides back to fully invested and holds ~42 days to ride the recovery. Bands are pre-registered, not tuned per result.新区间须连续保持 5 个交易日才生效（滞后机制抑制频繁换仓）。在可买入的恐慌底（投降且美联储看跌在场）时，模型会强制回到满仓并持有约 42 天以参与反弹。区间为预先登记，非按结果调参。
Independent bear episodes · the true sample size独立熊市事件 · 真实样本量
| Peak峰值 | Trough谷底 | S&P drawdown标普回撤 |
|---|
| 2000-03-24 | 2002-10-09 | -47.5% |
| 2007-10-09 | 2009-03-09 | -55.2% |
| 2020-02-19 | 2020-03-23 | -33.7% |
| 2022-01-03 | 2022-10-12 | -24.5% |
Only ~4 independent ≥20% bears govern a get-out / get-back-in switch — effective-N, not the day count, bounds confidence. The edge is validated by leave-one-crisis-out + an AQR-style permutation null (real Sharpe & drawdown beat every random-timing shuffle, p=0.0), and holds under point-in-time (ALFRED) macro vintages.只有约 4 次独立的 ≥20% 熊市决定一次离场/再入场切换 — 决定置信度的是有效样本量，而非天数。该优势通过留一危机检验 + AQR 式置换零假设验证（真实夏普与回撤击败每一次随机择时洗牌，p=0.0），并在时点（ALFRED）宏观数据下依然成立。
Taxable-account warning: switching realises short-term capital gains that buy & hold defers — best run in a tax-advantaged account. The low ~1.5×/yr turnover keeps this manageable.应税账户提示：切换会实现买入持有可递延的短期资本利得 — 最好在税收优惠账户中运行。约 1.5×/年 的低换手使其可控。