LQ1 full-input ElasticNet many factor rank v1: Higher‑ranked stocks led in the 10‑day window and lower‑ranked stocks led in the 60‑day window.
- Now
- The ranking relationship reverses between the 10-day and 60-day windows.
- Model
- The 10‑day ICIRRolling ICIRThe mean information coefficient divided by its standard deviation over a trailing window, usually annualized. It measures the persistence of ranking skill across observations in the window.Open glossary entry → of +4.99 means higher‑ranked stocks led; the 60‑day ICIR of –3.34 shows lower‑ranked stocks led, meaning the rank direction reversed between short and long horizons.
- Interpretation
- Walk-forward elasticnet forecast. Scores ASOF D from prior-date factor inputs; training excludes labels that would not have been known by D-1.
- What to watch
- Compare short- and long-horizon results before changing the factor weight.
Evidence
| Measure | Reading |
|---|---|
| ICIRRolling ICIRThe mean information coefficient divided by its standard deviation over a trailing window, usually annualized. It measures the persistence of ranking skill across observations in the window.Open glossary entry → · 10D | +4.99 |
| ICIR · 20D | +3.25 |
| ICIR · 60D | -3.34 |
| Latest daily ICInformation coefficient (IC)The cross-sectional correlation between a signal score and a subsequent return. Rank IC uses ranked values and measures whether the signal orders securities correctly.Open glossary entry → | +0.0089 |
| IC historical percentile | 59.5% |
| IC robust z | +0.24 |
| IC hit rateIC hit rateThe share of completed evaluation dates on which a factor’s date-level information coefficient has the registered positive sign.Open glossary entry → | 42.4% |
| Spread Sharpe | -2.56 |
| Evaluation dates | 85 |
| Peer percentile · 10D | 87.9% |
| Measured peers | 356 |
| Nearest-peer IC correlation | +0.91 |
Factor
#LQ1 full-input ElasticNet many factor rank v1