Data Science & ML Claude 技能

Backtest Result Persistence

Persists trading backtest results to an SQLite database to enable historical performance comparison and model optimization.

Trading Drawdown Guardrails

Implements standardized risk management and drawdown protection patterns for algorithmic trading systems.

WRDS Research Data Access

Provides validated, high-integrity PostgreSQL query patterns and connection logic for Wharton Research Data Services (WRDS) datasets.

Sparse Expression Visualization

Optimizes the visualization of sparse single-cell gene expression data by implementing alternative plotting patterns that prevent boxplot collapse.

Selection Data Caching for Trading

Optimizes financial data retrieval by caching market symbol data to reduce API latency and avoid rate limits during trading bot development.

Multi-Agent Trading Integration

Integrates multi-agent Claude systems into algorithmic trading pipelines to optimize model training and manage live risk with automated oversight.

Network Architecture Sizing for Trading RL

Optimizes PPO neural network dimensions to balance trading model capacity, inference speed, and hardware memory usage.

Colab Unzip Path Manager

Corrects file path errors and API key configurations in Google Colab environments after repository extraction.

Deconvolution Intensity Scaling

Ensures quantitative accuracy in microscopy deconvolution by preserving original intensity relationships across image channels.

Trading Position Sizing & Action Space

Implements a 7-action space with integrated position sizing and small account simulation for reinforcement learning trading models.

Notebook-to-Module Refactoring

Refactors Jupyter notebook code into reusable Python modules while preserving critical variable definitions and configuration.

Jupyter Autoreload Awareness

Optimizes Jupyter notebook workflows by preventing unnecessary kernel restarts when using the IPython autoreload extension.

GPU Correlation Accelerator & Cache

Accelerates financial correlation matrix computations using GPU-vectorized PyTorch operations and persistent SQLite caching.

Persistent Market Data Cache & Gap-Filling

Optimizes historical market data retrieval by eliminating redundant downloads through persistent caching and incremental gap-filling.

Trading Gates & Pattern Filter

Implements a multi-stage validation system for algorithmic trading that filters signals through pattern-matching and risk-management gates.

Trading Dual-Folder Workflow

Manages separate development and production environments for private repository trading models and Colab training.

Notebook Checkpoint Pattern

Implements a structured checkpointing system for connecting multiple Jupyter notebooks into robust, memory-efficient data pipelines.

Dotted Symbol Exclusion for Trading Data

Optimizes financial data pipelines by automatically filtering non-standard ticker symbols that cause API errors during sector lookups.

Alpaca Trading Order Limitations

Enforces Alpaca broker-specific constraints by filtering unsupported crypto short orders while maintaining valid equity shorting capabilities.

Parallel Universe Selection Optimizer

Accelerates quantitative trading universe selection by 3-4x through parallel processing and optimized caching strategies.

Interactive Plotly FigureWidget Fix

Resolves KeyError: 'uid' errors when updating interactive Plotly FigureWidget shapes and sliders in VS Code.

Alpaca Algo Trader Plus Data Connector

Leverages the Alpaca Algo Trader Plus subscription to fetch extended historical market data for training robust trading models.

Lightsheet PSF Deconvolution

Eliminates horizontal banding artifacts in microscopy data by implementing true lightsheet Point Spread Function (PSF) calculations.

Per-Sector Trading Candidate Filter

Optimizes trading universe selection by applying sector-specific volume filters to ensure a diverse and manageable pool of candidates.

Trading Selection Diversity Validation

Enforces strict portfolio diversity constraints and correlation limits in trading systems using the Fail Loudly pattern.

PyMC Bayesian Modeling

Builds and validates Bayesian statistical models using PyMC's probabilistic programming framework.

LlamaFarm Text Classification

Classifies text data using zero-shot labeling or custom-trained machine learning models directly within your terminal.

US Equity Valuation Extremes Detector

Detects historical extremes in US equity valuations by normalizing metrics like CAPE and PE into percentile scores.

LLM Evaluation Suite

Implements comprehensive evaluation frameworks for LLM applications using automated metrics, human feedback, and LLM-as-judge patterns.

Matplotlib Scientific Plotting

Generates professional, publication-quality data visualizations and charts using Python's foundational plotting library.