Technical Signal Discovery

Overview

This system evaluates 80 technical indicators across peer groups of 4-8 companies, measuring which signals produce consistent positive expected value in directional trades. Rather than optimizing indicators for a single stock, it asks: which signals work across an entire peer group?

Methodology

80 indicators across 6 categories: trend, momentum, volatility, volume, mean reversion, and custom composites. For each indicator-ticker-direction-horizon combination, the system computes Expected Value. Returns are beta-adjusted against SPY. An indicator survives only if it shows positive EV across a majority of tickers. Walk-forward OOS validation with 5 expanding windows eliminates curve-fitted signals.

Data Pipeline

Source: Schwab Market Data API (OHLCV price history, fundamentals). Indicators: 80 computed from OHLCV data via pure Python functions. Storage: SQLite (sweep_results, sweep_events, sweep_summary, sweep_oos_results). Peer groups run: Managed Care, Cable-Broadband, Athletic Retail, Chinese ADR, GME Solo. Validation: Walk-forward with 5 expanding windows.

Limitations

Technical indicators are inherently backward-looking price derivatives. Small peer groups limit statistical confidence on individual signals. Transaction costs not yet incorporated into EV calculations. Regime changes may invalidate historical patterns.