2 Commits

Include renames

Author	SHA1	Message	Date
mathiasandClaude Sonnet 4.6	3aded95ef1	feat(hpo): sweep results + update WINDOW default to 120 ... HPO sweep (18 configs, D_MODEL×DEPTH×WINDOW grid): Best: D_MODEL=128 DEPTH=2 WINDOW=120 → phase1_r2=0.3908 Worst: D_MODEL=64 (all configs) → max phase1_r2=0.3653 Key findings: - WINDOW=120 (5 days) > 240 > 480 — FX vol prediction is local, not regime-scale - DEPTH=4 doesn't improve over DEPTH=2 — 2 causal layers sufficient - D_MODEL=64 undercapacity; 128 and 256 comparable Updated WINDOW default: 240 → 120 (HPO winner). Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>	2026-06-26 12:42:50 +02:00
mathiasandClaude Sonnet 4.6	3445b6d267	experiment(phase0): NULL result — path B proxy gate (ref #5) ... Phase-0 SSL feasibility gate run on daily 2019-2023 EUR/USD (path B deviation: not hourly 2008-2022 + Go harness as specced in #5). Results: TS-JEPA silhouette mean=0.018 (need >0.20) — FAIL PCA baseline silhouette=0.136 — also below threshold sensitivity: 2000 ep + D=64 worsened to 0.004 (not a training-time issue) Root cause: 2 daily features (ret, realized_vol) carry minimal regime structure at this resolution. The JEPA objective with SIGReg pushes embeddings toward isotropic Gaussian — good for downstream probes (val_vol_r2>0) but may actively resist the clustering structure the silhouette gate measures. Null protocol: real gate requires #4 (Go harness) + #2 (hourly data, more features) before rerunning. HEPA (#14) noted as alternative backbone. Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>	2026-06-24 10:52:07 +02:00