What problem does it solve?
This Skill provides an end-to-end pipeline that pulls historical klines from the Aster DEX, transforms market data into numeric series, and produces zero-shot point forecasts with quantile prediction intervals via TimesFM to support trading signal generation and risk assessment.
Core Features & Use Cases
- Aster MCP ingestion: retrieve historical klines and optional funding rates at multiple intervals for any supported trading pair.
- TimesFM forecasting: run TimesFM 2.5 to generate point forecasts and full quantile prediction intervals for uncertainty-aware decisions.
- Batch multi-pair processing: forecast many pairs in a single batch for efficient signal generation and portfolio-level risk analysis.
- Risk-aware rules: interpret quantile spreads for position sizing, backtest forecast accuracy, and reduce exposure when uncertainty is high.
- Use Case: produce 24-hour hourly forecasts for BTCUSDT to inform trade entry/exit decisions and assess downside/upside risk via the 10th–90th percentiles.
Quick Start
Request a 24-hour hourly forecast for BTCUSDT with quantile intervals and return the point forecast plus the 10th and 90th percentile bounds.
