Live results from the $100K AI trading experiment. Every metric public. No cherrypicking.
The AI agent is currently in its pre-validation trial period (May 16 – June 1). Performance tracking with full metrics begins when the official 365-day countdown starts.
This page will automatically populate with real-time data: monthly P&L, sector breakdown, benchmark comparison, drawdown charts, and trade-level attribution.
Daily mark-to-market including unrealized gains on open positions.
Percentage of winning trades and average expected value per trade.
Largest peak-to-trough decline. Hard risk limit: -15% portfolio drawdown.
Risk-adjusted return vs. risk-free rate. Updated rolling 30-day and since inception.
Daily comparison to buying $100K of SPY on Day 1 and holding for 365 days.
P&L broken down by sector to identify where edge exists (or doesn't).
Performance metrics update on every site build (daily at minimum). Closed trades appear 10 days after exit per the transparency policy.
Yes — once there are 30+ days of data, the Sharpe ratio will be calculated on an annualized basis using the risk-free rate (3-month Treasury yield).
SPY (SPDR S&P 500 ETF) bought on Day 1 of the experiment and held for 365 days. This is the simplest alternative — "just buy the index." If the AI agent can't beat this, the complexity isn't justified.
Trades are logged to the blotter with timestamps. The git history of the source repository is public. Trades publish after a 10-day delay, but the entry date is fixed at time of execution via the Alpaca API.