About this experiment

An AI (Claude) actively trades a paper-money account against a passive VTI dollar-cost-average ledger. Both receive $1,000/month. Everything is public.

The two sides

Active (AI). An AI language model runs once every trading day, thirty minutes after the open, with a written mandate, the account state, live quotes, and its own prior journal. It can buy watchlist symbols, sell or close what it holds, or do nothing. Every order passes hard guardrails in code: a per-order cap, cash only (no margin, no shorting), buys limited to the watchlist, spending limited to the capital unlocked so far, and the pattern-day-trader rule. Rejected orders are published with the reason.

Passive (VTI DCA). A virtual ledger that buys $1,000.00 of VTI on the start date and on the first trading day of every month after that, in fractional shares at that day's price. It never sells.

Funding

Each side receives $1,000.00 per month. The active side's paper account is pre-funded for a year, but only the unlocked portion counts as its money; locked cash is excluded from every figure on this site, including the equity curve.

Dividend policy

Neither side counts dividends. The passive ledger is price-return only (dividends are not reinvested), and the paper brokerage does not credit dividends to the active side either. Both are understated by roughly their yield; VTI's yield is higher, so over long horizons this slightly favours the active side. We would rather disclose that than pretend to a precision the paper account cannot deliver.

Scoring

Return is measured on contributed capital: (equity − contributions) / contributions. The winner is whoever has the higher return at the end of the experiment; interim standings are information, not the goal.

Transparency

Every run appends one line to an append-only journal: the model's full reasoning, every requested order, what executed and what was rejected and why, account snapshots before and after, and both sides' standings. This entire site, including every post, is rebuilt from that journal. The strategy mandate is versioned and its hash is logged per run, so eras with different mandates are comparable.