gentrade 0.0.20

GenAI based Trading Package

GenTrade

The python GenTrade package provide the core functions and agentic trading workflow to support GenAI based algorithms trading for crypto and stock markets.

It can help algorithms traders smoothly transfering from service API programming to prompt based agentic programming.

• Install the python package

  pip install gentrade
  

  if you want to try the package from source code you can

  set PYTHONPATH=<root of gentrade>/src/
  

• Then try:

  • Demos: command line based app to show market OHLCV data
  • Tests: test cases for the API and utilities
  • Apps: the application or microservices based on gentrade.

1. Demo

NOTE: please set environment variable like BINANCE_API_KEY, BINANCE_API_SECRET,
OPENAI_API_KEY before running below demos.

1.1 Use traditional API approach for crypto currency or US stock:

1. Draw the latest 100 hours btcoin's prices (OHLCV)

   cd demo/crypto-cli
   python run_matplot.py -a btc -t 1h -l 100
   

   Output:

   

2. Use SMA (Simple Moving Average) to analyze the ETH's prices

   cd demo/crypto-cli
   python run_sma.py -a eth -t 1d -l 200 -g
   

   Output:

   

   It also support US stock like applying SMA on TESLA (TSLA)'s price:

   cd demo/stock-us-cli
   python run_multiple.py -a TSLA -t 1d -l 200 -g
   

   

3. If want try other strategy like RSI

   cd demo/crypto-cli
   python run_multiple.py -s rsi
   

   

   NOTE: of course you can try more strategies like macd, bb, wma etc

1.2 Use Agentic LLM approaches:

1. If want ask LLM to select a strategy via a simple prompt like

   • Prompt: Please get past 400 days price for bitcoin, then different strategy to do back testing, and figure out what strategy is the best according to final portfolio value
   • Prompt: 请获取过去300天的以太坊的价格，并使用不同的策略进行回测，最后选出最佳的策略

   cd demo/agent
   python run_auto_strategy.py
   

   

2. If want ask LLM to generate strategy according to your idea and do back test, for example:

   • Prompt: "请获取过去300天的以太坊的价格，并使用简单平均移动策略来进行回测，在这个策略中，请设置慢线为9，请设置快线为26"
   

2. App Services

2.1 OHLCV Data Service

• Start Server

  # Pull image
  docker pull registry.cn-hangzhou.aliyuncs.com/kenplusplus/gentrade_data_serv
  
  # Create .env file from .env_template
  
  # Run OHLCV datahub service
  docker run -p 8000:8000 \
      --env-file=.env -v <data folder>:/app/cache \
      registry.cn-hangzhou.aliyuncs.com/kenplusplus/gentrade_data_serv
  

• Client Test

  # Get all supported markets
  curl -X 'GET' \
    'http://127.0.0.1:8000/markets/?market_type=all' \
    -H 'accept: application/json'
  
  # Get all available assets from a specific market
  curl -X 'GET' \
    'http://127.0.0.1:8000/assets/?market_id=b13a4902-ad9d-11ef-a239-00155d3ba217&start=0&max_count=1000' \
    -H 'accept: application/json'
  
  # Get OHLCV for a specific asset
  curl -X 'GET' \
    'http://127.0.0.1:8000/asset/get_ohlcv?market_id=b13a4902-ad9d-11ef-a239-00155d3ba217&asset=BTC_USDT&timeframe=1m&since=-1&limit=10' \
    -H 'accept: application/json'
  
  # Start OHLCV collector threading in the background
  curl -X 'POST' \
    'http://127.0.0.1:8000/asset/start_collect?market_id=b13a4902-ad9d-11ef-a239-00155d3ba217&asset=DOGE_USDT&timeframe=1h&since=1732809600' \
    -H 'accept: application/json' \
    -d ''
  

The cached data can be found at this directory