About

A forward test of an insider-signal-driven equity strategy. Started April 29, 2026 with $100,000 of paper money on Alpaca. Signals come from the OpenInsider MCP. Orders go through the Alpaca MCP. The strategy is mine. The rules are below. Every trade has a "why" entry.

This is not investment advice. It's me writing things down so a year from now I know which beliefs survived contact with reality.

Strategy rules

Simple on purpose. If it works simply, complicating it later is a real choice. If it doesn't work simply, complicating it doesn't fix it.

• Sources. OpenInsider MCP. cluster_buys, top_buys, officer_buys. Officer-driven (CEO / CFO / EVP) preferred over 10%-owner buys, since 10%-owners are often institutional rebalancers without an information edge.
• Filters. Market cap > $300M. No active S-3 or 424B5 (equity dilution) in 90 days. No NT-10K with accounting reason. No earnings inside the 60-day hold window.
• Sizing. 2-4% of book per name.
• Hold. 60-day minimum. The academic edge in officer purchases is on 1-6 month horizons, not days. Day-trading insider signals is a different and worse game.
• Stop. -20% hard stop, GTC, set immediately after fill.
• Benchmark. SPY weekly.
• Approval. Every order requires explicit approval before submission. There is no automation. Paper or live, the discipline is the same.

Stack

• Signals. OpenInsider MCP (Form 4, SEC EDGAR, FINRA short data, Yahoo quotes).
• Execution. Alpaca MCP in paper mode.
• Site. Astro, deployed to Railway from a private GitHub repo. Each trade is a markdown file with typed frontmatter.