The market, systematized
Aerondight Systems evaluates roughly 900 U.S. stocks each day using systematic equity models and a machine-learned market-regime classifier. The current regime is Bull Tech, with both models in agreement. Every signal is timestamped and published live—building a transparent record, one decision at a time.
Instruments
Three live systems, each with its own page: what the market is doing, what the strategy is buying, and where sector leadership is rotating.
Market Regime Classification
HMM + XGBoost system detecting four distinct market states to modulate equity scoring.
Today: BULL TECH · 94% conf Silver Alpha — LiveSilver Alpha Strategy
94 trades, 62% win rate, +325% return — full trade log with out-of-sample results from 2023–2026.
Latest signal: KEX · 2026-07-09 Sector Rotation — LiveSector Rotation
Weekly market risk status and sector rotation confirmations, updated after each Friday close.
This week: Transitional · 2/5 votesRecent signals
The two live feeds, side by side. Silver is the systematic strategy; Gold is the ML alpha model. Full history and outcomes live on the Silver Alpha page.
Research reports
Institutional-style single-ticker reports generated by Aerondight's Thesis engine — quantitative signals, fundamentals, and recent news in a structured seven-section analysis.
Latest research
Silver Alpha v2: Expanded 2020–2025 Out-of-Sample Validation
2026-05-04Widening the test window to include COVID, the 2022 bear market, and the AI recovery — 96 trades, +278% total return, +10% annualized excess over SPY.
Out-of-Sample Backtest: Silver Alpha Performance
2026-02-0194 trades, 62% win rate, +325% portfolio return — the numbers behind Silver Alpha's 2023–2026 out-of-sample track record.
AI-Powered Equity Research Reports
2025-12-31How Aerondight uses Claude to generate institutional-quality single-ticker research at ~$0.005 per report.
The Mastermind
My name is Ted. I build systematic equity research tools.
Aerondight Systems is my quantitative platform that scores ~900 US stocks daily using multi-factor models, machine learning regime classification, AI-generated sentiment models, and AI-generated research reports. The system runs two parallel scoring models across different time horizons, each informed by a market regime classifier that adapts to current conditions.
This site is where I publish methodology, backtest results, and sample research — building a public track record one signal at a time.