Summary

Adds a separate global_etf_confidence_vol_gate runtime profile for testing the Global ETF confidence plus relative-volatility gate without changing the existing global_etf_rotation default behavior.

Context

Recent research showed that simply concentrating Global ETF rotation into Top1 improved CAGR but widened drawdown. The more useful signal was a guarded version: only shift from equal-weight Top2 to 75/25 Top1/Top2 when Top1 has a high momentum z-gap and its trailing volatility is not materially higher than Top2.

Changes

• Extends the shared Global ETF signal engine with optional confidence weighting.
• Keeps global_etf_rotation default confidence weighting disabled.
• Adds global_etf_confidence_vol_gate with:
  • sma_period=250
  • confidence_threshold=1.0
  • confidence_top1_weight=0.75
  • confidence_volatility_window=126
  • confidence_volatility_max_ratio=1.3
• Registers the new profile in catalog, manifest, runtime adapter, account sizing, docs, and tests.
• Adds a research note documenting the production-like backtest snapshot and rollout caveat.
• Hardens direct compute_signals calls by normalizing ranking and canary inputs before ticker-list construction.

Validation

• python -m ruff check src/us_equity_strategies/strategies/global_etf_rotation.py src/us_equity_strategies/catalog.py src/us_equity_strategies/manifests/__init__.py src/us_equity_strategies/entrypoints/__init__.py src/us_equity_strategies/runtime_adapters.py src/us_equity_strategies/account_sizing.py tests/test_global_etf_rotation.py tests/test_catalog.py tests/test_entrypoints.py
• python -m pytest tests/test_global_etf_rotation.py tests/test_catalog.py tests/test_entrypoints.py
• python -m pytest

Rollout

This PR does not alter the existing global_etf_rotation profile. The new profile should be treated as an experimental runtime-enabled candidate for paper or small allocation first.