Factor Lab Daily Brief 2026-09-12
S&P 500 factor analysis as of 2026-09-11: Momentum reversal deepens (IC=-0.20), value factors turned negative, size effect persists. Sector decomposition shows IT momentum IC=-0.48. All reversal alerts are regime continuation, not new events.
Factor Lab Daily Brief 2026-09-12
Data Cut-off and Sample Definition
- Price data as-of: 2026-09-11 (Friday, US market close)
- Price cache: 502 stocks, 639 trading days (2024-02-26 to 2026-09-11)
- Fundamental panel: combined_panel.pkl (TTM rolling, quarterly filings)
- Evaluation window: 60 cross-sectional observations for fundamental factors; 39 for price factors (MOM/VOL/SIZE)
- MOM holding period: Price[t-21]/Price[t-273] - 1 (21-day return, 273-day lookback)
- Key: ic_mean is the mean of historical cross-sectional rank correlations, not a daily IC, not forward returns, not a downside probability. The latest 21 trading days cannot be fully evaluated for the MOM factor given the holding period convention.
Current vs Previous vs Recent Trend
Full 7-Factor Table
| Factor | IC Mean | ICIR | p-value | Long-Short (Q1-Q5) | Previous IC | Delta | Valid Obs |
|---|---|---|---|---|---|---|---|
| EP | -0.0480 | -0.52 | 0.0002⭐ | +2.98% | -0.0480 | 0.0000 | 60 |
| BP | +0.0030 | +0.05 | 0.7276 | +1.13% | +0.0030 | 0.0000 | 60 |
| FCF Yield | -0.0352 | -0.54 | 0.0001⭐ | +2.48% | -0.0352 | 0.0000 | 60 |
| ROE | -0.0248 | -0.68 | <0.0001⭐ | +0.76% | -0.0248 | 0.0000 | 60 |
| MOM | -0.1973 | -1.38 | <0.0001⭐ | +5.99% | -0.2037 | +0.0064 | 39 |
| VOL | +0.0543 | +0.33 | 0.0500 | -2.03% | +0.0476 | +0.0067 | 39 |
| SIZE | -0.0763 | -1.06 | <0.0001⭐ | +4.20% | -0.0760 | -0.0003 | 39 |
⭐ = p < 0.05, statistically significant (not corrected for overlapping samples)
6-Run Trend (Price factors only show changes)
| Factor | 09-07 | 09-08 | 09-09 | 09-10 | 09-11 | 09-12 | Trend |
|---|---|---|---|---|---|---|---|
| MOM | -0.2225 | -0.2226 | -0.2201 | -0.2141 | -0.2037 | -0.1973 | Slightly收敛, still strongly negative |
| VOL | +0.0340 | +0.0339 | +0.0379 | +0.0415 | +0.0476 | +0.0543 | Slowly turning positive, p=0.05 borderline |
| SIZE | -0.0659 | -0.0660 | -0.0706 | -0.0747 | -0.0760 | -0.0763 | Continues weakening |
Fundamental factors (EP/BP/FCF/ROE) are identical since 09-07 because the TTM panel does not change with single-day price updates.
Sector Momentum Table (MOM by Sector)
| Sector | IC Mean | ICIR | p-value | Long-Short (Q1-Q5) | Status |
|---|---|---|---|---|---|
| Information Technology | -0.4845 | -1.90 | <0.0001⭐ | +19.21% | 🔴 Strong reversal |
| Industrials | -0.2207 | -1.36 | <0.0001⭐ | +8.13% | 🔴 Strong reversal |
| Financials | -0.2717 | -1.31 | <0.0001⭐ | +5.15% | 🔴 Strong reversal |
| Consumer Staples | -0.0910 | -0.46 | 0.0067⭐ | +4.64% | Significant negative |
| Utilities | -0.0701 | -0.47 | 0.0061⭐ | +0.97% | Significant negative |
| Consumer Discretionary | -0.0500 | -0.32 | 0.0580 | +2.46% | Not significant |
| Health Care | -0.0497 | -0.18 | 0.2654 | -0.21% | Not significant |
| Real Estate | +0.1540 | +0.47 | 0.0538 | -2.56% | Marginally significant |
Sector Momentum Alert Status (vs ic_history.json baseline)
The following alerts are derived from historical positive-momentum baselines in ic_history.json. All are regime continuation, not new today:
- 🔴 IT/Industrials/Financials/CDC/C-Staples/Utilities: IC flipped from positive to negative — historical baseline mean +0.16 to +0.52, all currently negative
- ⚠️ Financials/IT: Deviated 2σ+ from historical mean (z-score -2.33 to -2.42)
- ⚠️ Health Care: IC偏高 (z=+2.23), because historical baseline was negative (-0.12), current -0.05 deviates upward
- ⚠️ Real Estate/Utilities/CDC: p-value flipped from significant to not significant
These alerts have been firing continuously since momentum regime turned negative in late August. They are not new events today.
Strategic Implications
Momentum reversal has deepened. MOM IC = -0.20, negative for 6 consecutive runs, with IT sector IC=-0.48 being the deepest cross-sectionally. This means yesterday’s 21-day winners significantly underperform over the next 21 days — reverse momentum is currently the most statistically robust signal. Cannot conclude “market must fall,” as this is a cross-sectional ranking relationship, not a directional forecast.
Value factors broadly weak. EP, FCF Yield, and ROE all significantly negative. Low valuation, low FCF, low ROE stocks outperform relatively. This is not “value rotation confirmed” — the market is pricing fundamental factors in reverse under the current regime.
SIZE continues negative: Small-cap outperforms large-cap within S&P 500. Consistent with the trend since early September, though the magnitude worsened from -0.066 to -0.076.
VOL is the only marginally improving factor, IC rising slowly from +0.034 to +0.054, p=0.050 at the significance boundary. Low-volatility stocks continue to slightly outperform.
Conclusions That Cannot Be Drawn
- ❌ MOM negative ≠ market must drop tomorrow. This is cross-sectional correlation, not a directional prediction.
- ❌ SIZE negative ≠ broad small-cap rally. This is a relative size effect within S&P 500.
- ❌ EP/FCF/ROE negative ≠ value style confirmed. These factors are reversely effective in the current regime; direction needs specific validation.
- ❌ “Significant” p-value does not equal profitable strategy. No overlapping-sample autocorrelation correction applied; ICIR is IC mean/std, not strategy Sharpe.
- ❌ Number of negative sectors ≠ new reversals today. Baseline includes positive-momentum period from May-June; alerts persist for weeks after regime flip.
- ❌ BP not significant (IC=+0.003, p=0.73) = book value has no predictive power in this regime.
Statistical and Data Limitations
- t-tests are not corrected for overlapping samples (the 21-day forward returns for MOM/VOL/SIZE have substantial overlap), p-values may be biased low.
- ICIR = ic_mean / ic_std, measuring factor stability, not annualized returns or strategy Sharpe.
- Fundamental factors have 60 observations covering ~3 months of cross-sections; price factors have 39 due to the 273-day lookback requirement for MOM.
- In sector decomposition, Health Care (n=59) and Real Estate (n=31) have small samples; IC estimates are noisier.
- Real Estate MOM has only 20 valid observations; results have limited reliability.
- Long-short direction is Q1 (low factor) minus Q5 (high factor); positive means low-factor group outperformed. Transaction costs not deducted.