symbol-selection-statistical
SKILL.md
Symbol Selection Statistical Analysis - Research Notes
Experiment Overview
| Item | Details |
|---|---|
| Date | 2025-12-12 |
| Goal | Create sophisticated symbol selection using Hurst exponent, half-life, GARCH quality, and regime persistence to find assets compatible with predator-prey Markov trading systems |
| Environment | Python 3.10+, numpy, pandas, scipy, arch, hmmlearn (optional) |
| Status | Success |
Context
Simple Sharpe/momentum/volatility screening is insufficient for selecting symbols compatible with advanced trading systems that use:
- Predator-prey Markov regime detection (needs clear regime separation)
- GARCH-based position sizing (needs good volatility model fit)
- Mean-reversion strategies (needs appropriate half-life for trading timeframe)
Literature review identified key statistical metrics that predict trading compatibility better than traditional metrics.