One full trading cycle on a fresh $10,000 demo account, run by Claude following the same routine as the scheduled tasks. It researched the market, placed three orders, and the risk layer let one through.
How much of each hard limit the account is using after the session. A full bar means the risk layer will clamp or block further buys.
trade.py begin. First cycle, so no notes yet.Each order shows Claude's confidence against the 0.6 minimum (the dark tick), and the risk layer's ruling.
Long-term uptrend intact: price is above the 50-day (80.7k) and 200-day (71.9k) averages after a 3% weekly pullback. Starter position at 10% of the account, leaving room under the 25% crypto cap.
Risk layer: approved at full size
Down 7.5% this week and below its 20- and 50-day averages. Momentum is weak, so this was a low-conviction probe.
Risk layer: confidence 0.45 is below the minimum of 0.6
Core index position: SPY is near its 52-week high and above its 50- and 200-day averages.
Risk layer: US market is closed, so stocks and ETFs can't be traded now
The ETH and SPY orders were sent on purpose to show the risk limits at work. A normal run wouldn't submit orders it expects to be rejected.
Daily closes with the 50- and 200-day moving averages Claude compared against. Hover a chart to read exact values.
The data behind the decisions, from Yahoo Finance at the time of the session.
| Asset | Price | 5-day | vs 50-day avg | vs 200-day avg |
|---|---|---|---|---|
| Bitcoin | $82,977 | −3.3% | +2.8% | +15.5% |
| Ethereum | $2,508 | −7.5% | −1.8% | +17.7% |
| Solana | $110.41 | −8.6% | +2.0% | +27.5% |
Claude's only memory between runs. The next session starts by reading this.