backtrader结合大语言模型chatGPT

点此获取扫地僧backtrader技术教程

========

现在大语言模型chatGPT非常热，很多人用它来辅助编写程序。那么，能不能利用大语言模型来辅助编写backtrader策略程序呢？

github上还真有人做了一个项目backtrader_copilot，可以通过LangChain使用Chatgpt来辅助backtrader编程。

特性如下：

Features

• Loading backtrader code
• Generating natural language descriptions of implemented strategy rules
• Generating a graph visualisation of the implemented backtest (required graphviz installed)
• Generating feedback for implemented trading strategy
• Generating backtrader backtest code based on natural language description for
  • data
  • strategies
  • analysers
  • (...) more to come

• Autopilot that guides user through the process of setting up a backtest in backtrader by asking a series of questions.
• Save generated code

Quick start guide

• see main_example.py

Limitations

• A lot of limitations, this is v0.1 with the aim of providing functioning basic infrastructure.
• This version can only provide very limited and simple but functioning examples, this version is not production ready.
• Simple prompt templates are implemented, these have a lot of potential for improvement.
• No web-based UI yest (planned).
• No custom indicators yet (planned).
• Only a single shot implementation of the coding agent. More complex code implementation and testing strategies will be required (planned).
• Autopilot is static and does not ask dynamic questions yet (planned)
• No logging implemented (planned).