| Indicator | Type | Layer | Status | Key Finding | Operational Rule |
|---|---|---|---|---|---|
| FAI | Direction Signal | 1 | ✅ Live | Ranging SR=3.19 · IC(t+1)=−0.133 · autocorr=0.941 | Activate in ranging; suppress in bull (IC≈0) |
| PCI | Direction Candidate | 1 | ⚠ Pending Audit | Bull IC=−0.218 (p=0.021) · lookahead gap +0.210 unresolved | Paper-tracking only — timestamp audit required |
| DSI | Risk Overlay | 2 | ⚗ Experiment | SR 3.09→4.42, MDD −19.8%→−9.4% · only 7 bear days tested | Deploy when bear OOS ≥ 30 days |
| EFI | Warning Signal | 2 | ✅ Live | Vol-IC=+0.322 · 1.99× vol mult · extreme ret_t1=+0.008 | Close-auction alert only · NOT a short signal |
| PRI | Binary Filter | 2 | ✅ Live | Vol-IC=+0.296 · 1.73× vol mult · binary gate confirmed | Block entries at p75+; continuous scaling prohibited |
| FGI | Composite Gauge | 3 | ◈ Monitor | Composite of 5 indicators · IC(t+1) not significant | Regime context label only · not directional |
| CSI_z | Crowding Alert | 3 | ◈ Monitor | Vol-IC=+0.386 · positioning extreme ≠ direction | Alert only at |z|>2 — not a standalone signal |
| RLC_z | Medium-Horizon | 4 | ⚗ Experiment | Ranging IC(t+10)=+0.378 (p=0.005) · bull counterintuitive | Valid t+5~t+10 in ranging only · t+1 prohibited |
| Hypothesis | Verdict | Key Statistics | Interpretation |
|---|---|---|---|
| H1 — Information Advantage Foreign futures positioning leads next-day returns |
Partial | IC(t+1) = −0.133, p = 0.095 Direction Acc. = 56.2% (p = 0.114) |
Direction & sign consistent with theory. Borderline significance due to n=161 (8 months). Requires n≥250 to confirm at 5%. |
| H2 — Time-Varying Predictability Information advantage varies by market regime |
✅ Supported | Bull IC = +0.009 · Ranging IC = −0.237 FAI autocorr(1) = 0.941 |
Bull vs. Ranging IC gap is sharp. High autocorr confirms time-varying structure. This is the core novelty of the FAI design. |
| H3 — Gradual Price Discovery IC peaks at t+3~t+5, not t+1 (Kyle 1985) |
❌ Not Supported | IC(t+1)=−0.133 · IC(t+5)=−0.102 Fisher z p = 0.789 |
IC weakens t+1→t+5 (reversal vs. prior n=175 run). Bull-heavy sample (66%) masks medium-horizon signal. Re-test when ranging n≥50. |
| H4 — Contrarian Profits sign(−FAI) strategy generates excess return |
Conditional | Full SR = 1.07 (t = 0.84, p = 0.40) Ranging SR = 3.19 (n = 48) |
Full-sample statistical significance weak. Ranging-regime economic significance strong. Supports regime-conditional deployment. |