A Beginner Workflow for Coding and Testing Stock Strategies on BigQuant

Summary

This tutorial introduces two strategy-building approaches on BigQuant: an AI-driven workflow that trains a model from selected factors and filters, and a user-coded approach with explicit selection rules. It focuses on the latter, outlining how to clone a sample, define custom factors, combine them into stock-selection conditions, and backtest a new rule. It recommends reviewing daily holdings and returns, examining both strong and weak individual picks, and adjusting the rules before testing them over a different date range.

The article says the shared example holds positions for two days and concentrates heavily in one stock, but it does not include the strategy source in the text provided. Its stated annual and cumulative returns and drawdown are claims without supporting test details such as costs, benchmark comparison, or out-of-sample validation. The workflow is useful as a basic iteration outline, but continued historical performance across another period is not sufficient by itself to establish robustness or live-trading suitability.

Key ideas

• The tutorial contrasts model-trained AI strategies with strategies built from explicit coded rules.
• Its workflow covers factor creation, rule definition, backtesting, and review of individual holdings.
• It recommends testing adjusted rules over a different date range to check whether results persist.
• The described sample holds positions for two days and concentrates capital in one stock.
• The reported performance claims lack supporting methodology and do not establish live-trading robustness.

Tags

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