Does this skill require a local MLflow server?

No, it can connect to any remote or local MLflow tracking server as long as the MLFLOW_TRACKING_URI and necessary credentials are provided in your environment configuration.

How does this differ from the previous mlflow-query skill?

Unlike the CLI-based predecessor, this skill uses the native Python API, allowing for write operations (logging), more complex DataFrame-based queries, and better handling of metric history.

Can I use this for non-trading experiments?

Absolutely. While it includes specialized features for trading backtests, it functions as a full-featured MLflow client for logging any machine learning parameters, metrics, and runs.

What trading metrics are calculated automatically?

By providing a returns CSV, the skill uses QuantStats to calculate over 70 metrics including Sharpe ratio, CAGR, Max Drawdown, Volatility, and Kelly Criterion.

MLflow Python Integration

Name: MLflow Python Integration
Author: terrylica

byterrylica

•

데이터 과학 및 ML

Automates experiment tracking and backtest logging using the MLflow Python API and QuantStats metrics.

소개

This skill provides a unified interface for Claude to interact with MLflow tracking servers, enabling seamless logging of experiment metrics, parameters, and artifacts directly from your development environment. It specializes in quantitative trading workflows by integrating QuantStats to automatically calculate over 70 performance metrics, such as Sharpe and Sortino ratios, from backtest returns. Whether you need to search previous runs using SQL-like filters, manage experiment metadata, or analyze metric time-series data, this skill streamlines the entire machine learning and strategy development lifecycle within Claude Code.

주요 기능

Automated calculation of 70+ trading metrics via QuantStats integration
Advanced run querying with SQL-like filtering and DataFrame operations
2 GitHub stars
Unified MLflow Python API integration for logging metrics, parameters, and artifacts
Complete experiment management including creation, searching, and metadata tracking
Native support for retrieving time-series metric history for performance analysis

사용 사례

Tracking and comparing quantitative trading strategy backtests with standardized metrics
Managing machine learning experiment parameters and results across distributed teams
Querying historical experiment runs to identify top-performing models or configurations

소개

주요 기능

Automated calculation of 70+ trading metrics via QuantStats integration
Advanced run querying with SQL-like filtering and DataFrame operations
2 GitHub stars
Unified MLflow Python API integration for logging metrics, parameters, and artifacts
Complete experiment management including creation, searching, and metadata tracking
Native support for retrieving time-series metric history for performance analysis

사용 사례

Tracking and comparing quantitative trading strategy backtests with standardized metrics
Managing machine learning experiment parameters and results across distributed teams
Querying historical experiment runs to identify top-performing models or configurations