Factor Lab Daily Brief 2026-10-01

Data Cutoff

2026-09-30 (last trading day in prices_cache.parquet)

Run date: 2026-10-01 06:31 CST

Sample and Evaluation Methodology

• Universe: S&P 500 constituents, 502 valid stocks (474 for ROE)
• IC definition: Spearman rank correlation between factor value and subsequent 21-day return, averaged across historical cross-sectional slices. Not a single-day IC, not a forward return prediction, not a down-market probability.
• Holding period: Default 21-day forward return. The most recent 21 trading days (9/10–9/30) are not yet fully evaluable
• t-tests: Unadjusted for autocorrelation in overlapping returns; p-values are rough guides only
• ICIR: IC mean / IC std, not a strategy Sharpe ratio or annualized return
• Sector momentum: MOM factor computed separately within each GICS sector, covering 7 sectors

##本期 vs 上期 Comparison

Note: Sessions 9/30 and 10/1 share the same price cache (as of 9/30). Fundamental factors (EP/BP/FCF Yield/ROE) are numerically identical. MOM/VOL/SIZE show minor decimal differences due to cross-sectional regression weight adjustments.

Seven-Factor Summary

Key readings:

• VOL (volatility) is the most consistently positive factor this session (ICIR > 1.0) — low-vol stocks显著 outperform high-vol stocks. The “low-vol anomaly” persists.
• EP, FCF Yield, ROE all significantly negative: Low-value (high EP/BP/FCF), low-ROE stocks outperformed within the sample. This cannot be interpreted as “value investing is confirmed” — in a negative-MOM regime, prior losers (low EP, low FCF, low ROE) naturally mean-revert, which is a byproduct of momentum reversal, not intrinsic value factor effectiveness.
• MOM negative but insignificant (p=0.428): Momentum failed to establish statistical confidence in this sample. IC micro-improved from -0.048 to -0.035, well within noise.
• SIZE ineffective: No meaningful size signal in this sample period.

Sector Momentum Table

Sector readings:

• IT momentum is significantly negative, the primary drag on overall market momentum. Low-MOM stocks dramatically outperformed high-MOM stocks within IT (Q1-Q5 = +7.32%), consistent with the late-August tech correction narrative.
• Utilities momentum significantly negative (ICIR=-1.01, p<0.0001) — the most significant sector signal this session. Note: Utilities had a historical mean IC of +0.14 in ic_history; current -0.11 is z=-2.31 from that mean. However, combining with the alert log, Utilities momentum turned negative in late August; this alert is a continuation of an existing regime, not a today-only shock.
• Health Care momentum flipped from negative to positive (IC=+0.095, p=0.060 marginal), z=+4.81偏离历史均值. The low-MOM-outperforms pattern reversed in Health Care — this does not contradict the negative overall MOM: sector subgroups can have independent relative strength patterns.
• Financials and Industrials flipped from + to - but p-values are insignificant; statistically insufficient to confirm a regime shift.

Strategy Implications

1. Low-vol remains the relatively reliable factor: VOL ICIR > 1.0, validated across regimes. However, this is cross-sectional correlation, not an achievable strategy return (costs and turnover not deducted).
2. Value factors (EP/FCF Yield/ROE) currently negative: In the current negative-MOM environment, mean-reversion of prior losers pushes low-EP/FCF/ROE groups higher. This does not prove the value factor “has become effective” — once the momentum regime flips positive, the value group could quickly lag.
3. IT internal momentum breakdown: S&P 500’s largest-weight sector IT has MOM IC = -0.21 (significant), meaning the “strength begets strength” logic within tech temporarily failed. This is a warning for portfolios with high concentration in tech large-caps.
4. SIZE and BP both ineffective: Current data does not support timing or stock selection based on market cap or book-to-price.

Conclusions Not Supported

• ❌ MOM negative ≠ market must decline. It only means stocks ranked high 21 days ago underperformed relative to low-ranked stocks within the sample.
• ❌ EP/FCF Yield/ROE negative ≠ value investing confirmed. This is a byproduct of the momentum-reversal regime.
• ❌ Sector “flipped + to -” alert ≠ a today-only event.对照 ic_history, IT/Utilities momentum turned negative in late August; alerts are repeated markers of an existing state.
• ❌ Significant IC ≠ tradeable strategy. t-tests unadjusted for overlapping样本; quantile portfolios uncosted; cross-sectional correlation ≠ time-series return.
• ❌ SIZE marginal转正（-0.004 → +0.003）≠ small-cap rally underway. p=0.878, entirely within noise.

Statistical and Data Limitations

• Cache staleness: This session and the prior (9/30) use the same price cache (as of 9/30). All factor numbers reflect the cross-sectional state as of 9/30 close, not 10/1 real-time data.
• Incomplete MOM evaluation window: The 21-day forward return requires a complete 21-trading-day sequence. Any suspension or data gap during 9/10–9/30 affects IC calculation.
• Overlapping return autocorrelation: Cross-sectional regression uses rolling 21-day returns as the dependent variable; adjacent observations overlap by 20 days. Standard t-test p-values are downward-biased. All “significant” conclusions in this report should be discounted accordingly.
• Sector coverage: All 7 GICS sectors covered, but Utilities has only 31 stocks and Consumer Staples 36 — small-sector ICs are noisier.