Trader Cloud Backtest

功能

• Run heavy neural-trader jobs in the cloud
• Managed agent orchestration for backtesting, training, and parameter sweeps
• Cost estimation and optimization guidelines
• Detailed workflow with pre-flight checks and artifact handling

使用场景

• Performing multi-year walk-forward analysis for trading strategies
• Running large-scale Monte-Carlo simulations with thousands of paths
• Sweeping parameters across a grid for model optimization
• Training complex neural network models for financial forecasting

非目标

• Running quick, short backtests locally
• Replacing local development environments for simple tasks
• Providing a GUI for neural-trader configuration

工作流

1. Estimate job cost and resource needs
2. Provision or reuse a managed agent container with neural-trader installed
3. Perform a cheap pre-flight check for argument validity
4. Run the main trading job (backtest, train, sweep) with detailed metrics reporting
5. Pull necessary artifacts (equity curve, trade logs)
6. Ingest results locally (store metrics, pattern store)
7. Terminate the managed agent environment immediately

实践

• Cloud Execution
• Financial Modeling
• Cost Management

先决条件

• ANTHROPIC_API_KEY or CLAUDE_API_KEY
• Managed Agents beta access