🔬 Quant Lab — advanced analytics量化实验室 — 进阶分析

🌐 Cross-asset concentration — are the six markets one bet?跨资产集中度 — 六大市场是否同一笔押注？ ?How much the six markets (US equity, crypto, China, HK, commodities, the dollar) are really moving as ONE risk/liquidity bet vs independent bets. The absorption ratio is the share of total variance explained by the single dominant factor: 1/6≈0.17 means independent, →1 means everything moves together (a fragile, crowded regime that has historically preceded drawdowns). Stacking correlated views feels like diversification but doubles one exposure.六大市场（美股、加密货币、中国、香港、大宗商品、美元）在多大程度上其实是同一笔风险／流动性押注，而非独立押注。吸收比率是单一主导因子所解释的总方差占比：1/6≈0.17 表示独立，→1 表示一切同向（脆弱、拥挤的状态，历史上常先于回撤出现）。叠加相关的观点感觉像分散，实则在加倍同一笔敞口。

CONCENTRATED

3 of 6 markets are one bet (US, Commodities, Dollar). Cross-asset correlation is in the top 13% of its 5y range: stacking these views is doubling one liquidity/risk exposure, not diversifying.

Correlation matrix (63-day, risk-on-oriented)相关性矩阵（63日，按风险偏好定向）

Absorption ratio = PC1 variance share (Kritzman-Page 2010): high/rising absorption marks a fragile one-factor regime that has historically led drawdowns. Daily closes across US/Asia/24-7 crypto carry a timezone lead/lag — a coarse regime gauge, not a hedge ratio.

⚖️ Risk budgeting across the six markets六大市场的风险预算 ?If you expressed all six market views as one book, how should they be weighted? Equal-weight (1/6 each), inverse-vol (more in the calm markets), and equal-risk-contribution (ERC — each market contributes the SAME share of portfolio risk). The 'risk' column is each market's share of total portfolio variance: equal-weight quietly concentrates risk in the most volatile market; ERC equalizes it. This is the honest 'size as uncorrelated bets' weighting — a reference, not a trade list.若把全部六个市场观点合成一个组合，该如何配权？等权（各 1/6）、逆波动率（在平静市场配得更多），以及等风险贡献（ERC — 每个市场对组合风险贡献相同份额）。'风险'列是各市场对组合总方差的占比：等权会悄悄把风险集中在波动最大的市场；ERC 将其均衡。这是诚实的'按不相关押注定量'配权 — 仅作参考，非交易清单。

Equal-weighting all 6 market views puts 36% of portfolio RISK in one market (vs 17% if truly balanced); ERC rebalances to ~equal risk (heaviest weight: Dollar 0.439). Diversification ratio 1.66.

Diversification ratio分散化比率: EW 1.62 · ERC 1.66 (higher = more genuine diversification越高 = 越真实的分散化). Equal-weight puts等权将 36% of risk in one market; ERC caps each at的风险放在一个市场；ERC 将每个限制在 ~17%.

🌊 Crisis stress replay危机压力回放 ?What an equal-weight book of the then-available markets would have done through past crises. The point: in a stress everything correlates toward 1, so the 'diversified' six-view book still craters — the tail cost of the concentration the absorption ratio flags in calm times.等权持有当时可用市场的组合在过去危机中的表现。要点：在压力中一切相关性趋于 1，因此'分散'的六观点组合依然崩塌 — 这正是吸收比率在平静期所警示的集中度的尾部代价。

🧮 Factor IC scorecard — leak-free (point-in-time)因子 IC 记分卡 — 无前视（时点） ?Each equity factor's Information Coefficient: the cross-sectional rank correlation of its score vs forward returns, per quarter, built point-in-time (EDGAR as-of fundamentals + prices truncated to the date) so no factor ever sees a not-yet-filed report. IC-IR is mean/vol (the signal-to-noise); t_HAC is a Newey-West t-stat (overlapping windows autocorrelate); q_FDR is the Benjamini-Hochberg false-discovery rate ACROSS the factor panel. Judge survivors vs ~0 — post point-in-time fix the spreads are honestly weak.每个股票因子的信息系数：其分数对未来回报的横截面秩相关，按季度、以时点方式构建（EDGAR 截至日基本面 + 截断至当日的价格），因此任何因子都不会看到尚未提交的报表。IC-IR 是均值／波动（信噪比）；t_HAC 是 Newey-West t 统计量（重叠窗口存在自相关）；q_FDR 是跨因子面板的 Benjamini-Hochberg 错误发现率。以约 0 来评判幸存者 — 时点修正后价差确实偏弱。

Span区间 2011-03-31..2025-12-31 · 60 quarterly rebalances · forward次季度再平衡 · 前瞻 63d · ~1154 names只股票.

Factor overlap因子重叠: mean |corr|平均|相关| 0.137 raw →原始 → 0.055 orthogonalized. Most-correlated:正交化后。相关性最高： low_vol↔low_beta (0.74); quality↔accruals (0.71)

Deep ~2011-2026 re-test on a SURVIVORSHIP-BIASED price panel: delisted names are absent (yahoo serves only currently-listed), so this is an OPTIMISTIC bound — a clean test needs delisting-recovered prices. FINRA short-interest is omitted (no point-in-time history). Fundamentals stay point-in-time from the EDGAR panel. Compare to the shallow 2023-2025 read: factors that survived on ~2.5y (notably SUE) weaken on deep history. Point-in-time (EDGAR panel + prices truncated to asof); IC = quarterly cross-sectional rank corr of factor z vs forward return; t_hac = Newey-West; q_fdr = Benjamini-Hochberg across the factor panel. Survivors judged vs ~0.

The two dials behind the regime周期背后的两个刻度盘 ?The regime label comes from two scores between -1 and +1, built from market prices (which lead) plus a small weight on economic data (which confirms).周期标签来自两个介于 -1 与 +1 之间的分数，由市场价格（领先）加上经济数据（确认）的小幅权重构建。

Growth:增长： RISING上行 (+0.07 of ±1 · agreement／±1 · 一致度 4%)

Inflation pressure:通胀压力： RISING上行 (+0.40 of ±1 · agreement／±1 · 一致度 30%)

What's voting for this regime哪些指标在为此周期投票

Weakest link最薄弱环节

Watch how many S&P 500 stocks are in uptrends — of everything supporting the current call, it's the one closest to flipping sides. If it fades, the growth dial (and possibly the regime) goes with it.关注 how many S&P 500 stocks are in uptrends — 在支撑当前判断的所有因素中，它最接近翻转。若其转弱，growth 刻度（乃至周期）也会随之改变。

Does the market agree with the regime?市场是否认同此周期？

framework and tape agree

Market internals — risk appetite gauges市场内部结构 — 风险偏好计 ?Simple ratios between two assets; each answers a question about investor mood. Direction over the last 20 days matters more than the level.两类资产之间的简单比率；每一个都回答一个关于投资者情绪的问题。过去 20 天的方向比绝对水平更重要。

US equity — size & style美股 — 规模与风格 ?Is the advance broad or narrow? Small/mid caps vs the S&P 500; growth vs value style rotation. Direction over the last 20 trading days.上涨是广泛还是狭窄？小／中盘对标普500；成长对价值的风格轮动。看过去20个交易日的方向。

Small-cap participation小盘参与度: 69% of S&P 600 vs的标普600 对 62% for the S&P 500的标普500 — small caps participating MORE than large caps — healthy, broad-based tape小盘参与度高于大盘 — 健康、广泛的盘面.

🐳 Accumulation watch — sector ETFs增持监测 — 板块 ETF ?Each sector SPDR's top weights, with the part of the weight change beyond what price alone explains ('active'). On passive ETFs this is index reconstitution / float-weight flow, not manager conviction. Uncalibrated until more snapshots accrue.每只板块 SPDR 的主要权重，以及权重变动中超出价格本身所能解释的部分（'主动'）。在被动 ETF 上这是指数再平衡／流通权重资金流，而非经理人信念。在积累更多快照前未经校准。

Star fund managers' real moves & ETF money flows明星基金经理的真实动作与 ETF 资金流 ?Daily holdings of watchlist funds (ARK etc.) with mechanical flow stripped out, plus daily money in/out of each sector ETF. History builds from June 2026.观察名单基金（ARK 等）的每日持仓（已剔除机械性资金流），以及每个板块 ETF 的每日资金进出。历史自 2026 年 6 月起累积。