What problem does it solve?
Provides structured machine learning patterns, indicators, and workflows to convert raw market and financial data into predictive signals, risk metrics, and backtestable trading strategies so teams can move from exploration to disciplined execution.
Core Features & Use Cases
- Technical Indicators & Feature Engineering: implementations for SMA, EMA, RSI, MACD, Bollinger Bands, ATR and automated feature pipelines for returns, volatility, volume, and position features.
- Model Patterns & Selection: LSTM sequence models, XGBoost and Random Forest feature-based patterns, transformer guidance, and statistical models (ARIMA/GARCH) for volatility forecasting.
- Backtesting & Validation: walk-forward backtesting template, common pitfall mitigations (look-ahead, survivorship, transaction costs), and example backtest orchestration.
- Risk & Portfolio Tools: Sharpe/Sortino/Calmar/VaR utilities, Kelly position sizing, and mean-variance portfolio optimization with numerical solvers.
- Exchange Integration: workflow notes for integrating ML signals with Aster DEX including data fetch, risk checks, mandatory user confirmation, and execution monitoring.
- Use Case: train an LSTM on BTC-USD, generate hourly signals, run walk-forward backtests including realistic fees, compute position sizes, and produce a risk-adjusted portfolio allocation.
Quick Start
Use the finance-ml skill to compute features from recent BTC-USD price data, train a predictive model, perform a walk-forward backtest with transaction costs, and summarize signals and risk metrics for review.
