Our current finding: not yet — and we published that verdict about our own system. On September 6, 2026, after sixty-plus live trading days, we marked the current strategy NOISE on the record. Then we kept the experiment running, because the question deserves a real answer, not a hopeful one.
Why this question matters
If an AI agent can compound money legally, transparently, and repeatably, the implications go far beyond trading — think about what it would mean for how ordinary families save. That upside is exactly why this corner of the internet is drowning in grifters selling "AI trading" courses and black-box bots. Our sister station Signal & Noise tears those schemes down weekly, fine print first. This page is the other half of the deal: if we're going to stamp other people's schemes, we publish our own books.
The Daily 2 experiment
- The system: a rules-based engine that builds a market-wide watchlist every morning, selects a small book of names, manages it through the day with hard risk gates, and settles to cash by the close. Fully autonomous; humans watch, agents trade.
- The contract: day one, non-negotiable — never lose more than a small fixed percentage of the book in a single day. The guardrails that enforce it have held, including on the days that hurt.
- The evidence standard: live results outrank backtests. We distrusted our own backtesting lane after it produced flawed results, and now judge every change on what the engine actually does in the market.
- The verdict so far: the guardrails work; the edge doesn't. The strategy shows no realistic path to its goal at current expectancy — that's the NOISE stamp. The experiment continues on paper while the search for a real edge goes on.
What's next
Daily 2 Production — a nightly documentary of the experiment: what the engine saw, what it did, what it learned, published every trading day as a show. The production is in design now and will take over daily2.dbnr.ai when it airs.