HkEquityStrategies

Chinese README

Investing involves risk. This project does not provide investment advice and is for education, research, and engineering review only.

Research maintenance status

New Hong Kong strategy research and selection are on hold while the project focuses on US equities. get_runtime_selectable_profiles() and get_runtime_enabled_profiles() now return an empty set: these package selection interfaces offer no profiles for new runtime selection. Catalog metadata, strategy code, evidence and historical interfaces remain available for existing baseline resolution; their historical labels do not authorize new selection.

This change does not stop existing schedules, update deployed or pinned consumers, or change existing targets or risk controls. Resumption requires an explicit owner decision and the existing evidence and platform approval checks.

What this repository is

HkEquityStrategies is the Hong Kong equity strategy package for QuantStrategyLab. It contains reusable strategy implementations, manifests, catalog metadata, runtime adapters, and live-enablement checks shared by HK-capable platform repositories.

This repository is a strategy layer, not a broker or deployment layer. It does not store broker credentials, submit orders by itself, publish snapshot artifacts, or decide whether a profile is safe for live trading without external evidence.

QSL architecture role

• Layer: strategy-library.
• Responsibility: Hong Kong equity strategy implementation package.
• Owns: HK strategy code, manifests, runtime adapters, live-readiness checks.
• Consumes: QuantPlatformKit and validated HK snapshot artifacts.
• Must not: own broker credentials or deployment switches.

Current runtime surface

Canonical profile names were simplified in this branch. The old names are not kept as aliases, so platform repositories should reference the canonical names below through STRATEGY_PROFILE or RUNTIME_TARGET_JSON.

Direct runtime strategies

These profiles use platform-provided market_history and do not require a separate snapshot artifact before the strategy entrypoint can produce target weights.

The combo wrapper hk_equity_combo is retained for research/orchestration only and should not be treated as a runtime-enabled profile.

Profile	Name	Input	Benchmark	Current role
hk_global_etf_tactical_rotation	HK Global ETF Tactical Rotation	market_history	02800	Retained HK ETF runtime profile with broader ETF exposure and heavier product checks.

Snapshot-backed runtime strategy

The strategy package exposes one promoted snapshot-backed runtime entrypoint. HkEquitySnapshotPipelines owns the artifact contract, production data lineage, and promotion evidence.

Profile	Name	Input	Benchmark	Current role
hk_low_vol_dividend_quality_snapshot	HK Low-Vol Dividend Quality Snapshot	feature_snapshot + manifest	02800	Retained HK snapshot-backed runtime candidate; production point-in-time evidence is still required.

Removed research profiles

Rejected prototypes are no longer part of the public catalog, manifests, entrypoints, runtime adapters, or snapshot contracts. Platform repositories should only expose profiles returned by get_runtime_enabled_profiles().

See docs/research/hk-strategy-selection-20260603.md for retained and rejected strategy notes.

Performance and evidence boundary

The research numbers in this repository are review evidence, not return promises. Before enabling or changing any live profile, rerun the relevant research/readiness commands and review short, medium, and long windows where applicable:

• return and benchmark-relative return
• maximum drawdown and drawdown stability
• turnover, costs, lot-size, slippage, suspension, VCM, and CAS behavior
• data freshness and artifact version
• broker dry-run order preview, bilingual notification logs, rollout controls, and operator approval

If evidence is stale, incomplete, or the profile is not returned by get_runtime_enabled_profiles(), keep it out of live runtime settings.

Quick start

python -m pip install -e '.[test]'
python -m pytest -q

Inspect runtime readiness

These commands are read-only unless you explicitly pass an evidence file to a validator:

python scripts/print_hk_live_enablement_matrix.py --json
python scripts/print_hk_runtime_readiness.py --profile hk_global_etf_tactical_rotation --platform longbridge --json
python scripts/print_hk_runtime_readiness.py --profile hk_low_vol_dividend_quality_snapshot --platform longbridge --json
python scripts/validate_hk_runtime_live_enablement.py --print-template --profile hk_global_etf_tactical_rotation --platform longbridge --json

For local smoke coverage of the ordinary ETF rotation path:

python scripts/smoke_hk_global_etf_tactical_rotation_dry_run.py --json

How this connects to execution

Platform repositories consume this package through strategy loaders and runtime metadata. They own broker credentials, market-data access, account state, dry-run/live switches, order submission, notifications, deployment settings, and rollback controls.

Supported HK runtime platforms currently include:

• InteractiveBrokersPlatform
• LongBridgePlatform

Deploy safely

1. Keep broker credentials and account identifiers outside Git.
2. Use platform repositories for dry-run, paper, or live execution switches.
3. Confirm strategy evidence and platform dry-run output before enabling scheduled execution.
4. Review generated orders, notifications, artifact URIs, and rollback settings.
5. Start with small staged exposure and keep kill-switch procedures documented in the platform repository.

Repository layout

• src/: strategy implementations, manifests, catalog metadata, runtime adapters, and readiness policies.
• tests/: unit, contract, and regression tests.
• docs/: integration notes, strategy research, and live-enablement evidence guides.
• scripts/: local research, smoke, readiness, and evidence helpers.

Useful docs

• docs/platform-integration.md
• docs/research/hk-strategy-selection-20260603.md
• docs/research/hk-global-etf-tactical-rotation.md
• docs/hk-low-vol-dividend-quality-snapshot-live-enablement.md

Safety and contribution notes

• AI assistants and maintainers follow AGENTS.md.

• Do not commit secrets, tokens, cookies, broker credentials, account identifiers, or private order data.

• Keep behavior changes small and include tests or reproducible evidence commands.

• Do not promote a research profile into live runtime settings without the documented evidence gates.

Community and security

• See CONTRIBUTING.md for pull request scope, local verification, and documentation expectations.
• Follow CODE_OF_CONDUCT.md for maintainer and contributor conduct.
• Report credential, automation, broker, exchange, or cloud-resource vulnerabilities through SECURITY.md; do not open public issues for secrets or live-execution risk.

License

See LICENSE.

Manual AI verification

Run Drift Check with verification_only=true to exercise the existing input and AI-review path without publishing drift issues. The review receives measured dimensions, thresholds and baseline lineage. Missing evidence or a non-approving advisory still parks the result; verification never grants trading authority.