ai trading bot reality checker

The problem

AI-generated trading bots often rely on flawed logic, look-ahead bias, and unrealistic slippage assumptions. Developers risk deploying code that lacks critical risk controls, error handling, or statistical validity, leading to catastrophic capital loss.

What it does

• Identifies hidden premises like market stationarity or infinite liquidity.
• Scans for execution risks including race conditions, integer rounding errors, and missing kill-switches.
• Audits backtesting methodology for survivorship bias and overfitting.
• Traces order placement logic to ensure idempotent handling and price sanity checks.
• Generates a structured Reality Check Report with a pass, conditional pass, or fail verdict.

Why this beats prompting it yourself

Standard LLMs tend to be overly optimistic about the code they generate. This skill enforces a systematic adversarial methodology that specifically hunts for failure modes, black-swan vulnerabilities, and statistical noise that generic prompts miss.

Use cases

• Auditing a new strategy proposal before committing capital to a live environment.
• Validating a backtest report to ensure results are net of realistic fees.
• Stress testing bot execution logic against exchange downtime or API failures.
• Reviewing AI-generated trading rules for hidden look-ahead bias in indicators.

Known limitations

This skill performs static analysis and logic audits. It cannot execute live trades or verify real-time latency in a production network environment.