strategy-compose

Orchestrates signal selection, portfolio weighting, and backtesting into traceable strategy research reports.

Updated Jun 27, 2026
One-click install
npx skills add https://github.com/HKUST-QUANT-SOCIETY/quantcode --skill strategy-compose-hkust-quant-society
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: strategy-compose
Source: https://github.com/HKUST-QUANT-SOCIETY/quantcode/tree/main/.opencode/groups/strategy/skills/strategy-compose
Command: npx skills add https://github.com/HKUST-QUANT-SOCIETY/quantcode --skill strategy-compose-hkust-quant-society

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve? Quantitative strategy research often produces untraceable results because signal selection, portfolio construction, and backtesting are run ad hoc without recorded provenance. This Skill coordinates those steps through published capability adapters and authoritative components, producing a verifiable StrategyReport with recorded contracts, sources, and environments. ## Core Features & Use Cases - Capability Discovery: Queries the published capability catalog and group Memory to verify the real status of DataAccess, Modeling, Riskfolio-QS, and VectorBT-QS components before use. - Reuse Recording: Logs coverage decisions via organization_reuse and freezes goals, acceptance criteria, and file scope through organization_solution. - Ordered Strategy Pipeline: Runs signal selection, portfolio weighting, and backtesting through published adapters, ending in a StrategyReport validated against real contracts and native task events. - Use Case: A strategy group member needs to combine existing momentum signals into a portfolio and backtest it; the Skill checks which adapters are actually connected, records reuse decisions, runs the pipeline, and produces an auditable report. ## Quick Start Start a strategy task that checks the capability catalog and group Memory, then runs signal selection, portfolio weighting, and backtesting to produce a traceable StrategyReport.

Frequently Asked Questions about strategy-compose

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I run a strategy backtest with signal selection and portfolio weighting?

Query the capability catalog and group Memory first to confirm which adapters like select_signals, combine_signals, and run_strategy_backtest are actually connected. Then run the business sequence of signal selection, portfolio weighting, and backtesting, ending with a StrategyReport.

How do I verify which quantitative components are available before a strategy task?

Check the published capability catalog and group Memory to confirm the real status of DataAccess, Modeling, Riskfolio-QS, and VectorBT-QS. Adapter names in the catalog do not guarantee the components are connected.

Does this Skill reimplement portfolio optimization or backtesting algorithms?

No. Portfolio optimization, fill replay, costs, and corporate action conventions come from authoritative components like Riskfolio-QS and VectorBT-QS. QuantCode does not rebuild these algorithms, and unconnected or staging results stay in their original state.

Can a strategy task deploy results directly to production?

No. Production deployment goes only through the independent Admin management plane and production service accounts. Strategy tasks can prepare candidate artifacts but cannot gain production shell access from role names or a single ordinary approval.

What determines whether a strategy task is accepted as complete?

Acceptance is based on a real StrategyReport, evidence from authoritative components, and native task events. It is not judged by a fixed number of tool calls or example weights, and budget exhaustion stops the task rather than expanding budget through HumanGate.