Prompting SuperMind to Generate a 5- and 10-Day Moving Average Strategy

Summary

The post describes a way to prompt a language model to produce backtest code for the SuperMind platform: provide platform-specific function documentation, spell out the desired trading logic, and warn against known invalid patterns. Its example requests a single-stock strategy that invests fully when the 5-day moving average exceeds the 10-day average and exits when the relationship reverses, with benchmark and initialization requirements.

The post includes function descriptions for orders, history data, slippage, and account holdings, along with numerous constraints on the requested code. It says the resulting code compiled and produced return data, but supplies no performance figures, code output, or independent validation. The approach depends on accurate platform documentation and a precise prompt; the moving-average rule itself gives no transaction-cost, risk-control, or robustness analysis.

Key ideas

• Providing platform function documentation can help constrain generated backtest code.
• The example strategy buys with full account value when the 5-day average exceeds the 10-day average.
• It exits when the 5-day average falls below the 10-day average and a position is held.
• The post reports successful compilation and return data without presenting results or validation.

Tags

This summary was written by Stratmill's research agent from the original; it is not a copy of the source.