Finance
Part 4 of 4 — the results. Part 1: every safety net failed silently · Part 2: the free stack · Part 3: publishing signals
The first three parts were about how my portfolio monitor works and how its safeguards failed. This one is the question everybody actually asks: do the AI buy signals make money?
Read more “Six Months of AI Buy Signals: The Honest Numbers”
AI
Part 2 of 4 — the stack. Part 1: every safety net failed silently · Part 3: publishing signals · Part 4: what it returned
Part 1 was about the safeguards in my portfolio monitoring tool and the six ways they silently stopped running. This one is the boring, useful half: what it actually runs on, why every component is a free LLM API or free tier, and which of those free tiers turned out to be load-bearing in a way I hadn’t thought about.
Read more “Free LLM API Tiers: Running Three Models for $0/month”
Finance
Part 1 of 4 — the safeguards. Part 2: the free stack · Part 3: publishing signals · Part 4: what it returned
Three months ago I wrote up how Richfolio went from v1.0 to v1.6 — the guard pipeline, the two-stage Think/Plan prompt, the reasoning persistence, all of it borrowed from patterns I liked in OpenAlice. Richfolio is my AI portfolio monitoring system: a cron job that reads my holdings, asks two language models what to buy, and emails me a brief.
Read more “AI Portfolio Monitoring: Every Safety Net Failed Silently”