factorminer
minihellboy/factorminer
FactorMiner is a governed research framework for discovering, evaluating, and combining interpretable alpha factors using typed domain-specific languages and LLM-guided workflows. It enables quantitative researchers to backtest portfolios, generate tearsheets, and manage factor libraries under strict transaction costs.
99 28MITUpdated 2026-08-03
84
GreatOverview
FactorMiner is a comprehensive research engine designed for quantitative financial alpha discovery and backtesting. It combines a typed formula DSL, LLM-guided search loops (Ralph and Helix), structured policy memory, and runtime recomputation to help users build robust composite signals. The platform provides tools for quintile backtesting, transaction-cost modeling, monotonicity checks, and automated reporting while maintaining a strict research-only boundary that leaves execution and risk limits to human operators.
Capabilities
- ▸Typed formula DSL and operator registry over OHLCV data
- ▸Automated factor generation, debate, and canonicalization loops
- ▸Portfolio backtesting with transaction cost and turnover modeling
- ▸Quintile analysis, IC tracking, and visual tearsheet generation
- ▸Model Context Protocol (MCP) server integration for agent workflows
- ▸Evidence packaging with content hashes and integrity verification
Best for
Discovering and evaluating alpha factors using LLM-guided search loops, Combining multiple factor libraries into composite signals with transaction cost adjustments, Running quintile backtests and generating visualization tearsheets for quantitative strategies, Validating financial market datasets and performing ablation studies or benchmark tests
Works with
Claude CodeCursorGitHub CopilotOpenCode
Evoa Score breakdown= Σ (score × weight)
84Task usefulness20%100 → +20.0
実タスクにどれだけ役立つか（機能の豊富さ・用途の明確さ）。 — AIによるcapabilities/use-cases解析
Code quality15%90 → +13.5
実装・指示の品質。 — AIによるSKILL.md/README解析
Maintenance15%100 → +15.0
リポジトリがどれだけ活発に保守されているか。 — GitHub 最終push日時の新しさ
Documentation12%50 → +6.0
ドキュメントの充実度・分かりやすさ。 — README/独自要約の情報量
Security15%80 → +12.0
危険・不審な挙動が無いか。 — AIによるセキュリティレビュー
Originality10%85 → +8.5
ありふれたラッパーではない独自性。 — AIによる独自性判定
Popularity8%55 → +4.4
コミュニティの採用度。 — GitHub Stars/Forks（対数スケール）
Compatibility5%100 → +5.0
対応AIエージェントの広さ。 — AIによる対応エージェント判定
Weighted total84.4 / 100
ライセンス不明/制限あり（Red）のSkillは総合スコアに0.85倍の補正を適用します。 ランキングはこのScoreのみで決まり、広告で変わりません。 算出方法の詳細 →
Security considerations
FactorMiner is restricted to a research-only boundary. It operates locally, processes user-supplied market datasets, and interacts with LLM providers through standard connectors. It explicitly lacks order-routing and live trading capabilities to prevent autonomous financial risk.
Categories
Summary and analysis are original content generated by AI Skills Rank. The skill's source text is not reproduced here — view it on the linked repository.