ALGOTRADING_Research

To-Do:

ML Model :

• try an three-gate approach using clustering or simple cutoffs to define strong sell / sell / hold / buy/ strong buy, then run a ML multi-class classification on top of it.
  • find a way to fix the class imbalance problem with time series for the ML training
• possibly use fear and greed index for sentiment analysis
• save 1st models to JSON

REST-API:

• find the easiest way to create a REST API connecting to Alpaca

Plan:

• use a web framework like Flask or FastAPI to build a REST API that connects to the Alpaca API. This will allow to get the new prices and send buy/sell orders.

• use the Alpaca API to retrieve the latest OHLCV data for ETHUSD every hour. get_last_trade function from the alpaca_trade_api module to get the latest trade data.

• load pre-trained CatBoost model from the JSON file and use it to predict the buy/sell signal based on the latest OHLCV data.

• use the submit_order function from the alpaca_trade_api module to place a buy or sell order on Alpaca papertrading based on the predicted signal.

  • set the order parameters such as quantity, limit price
  • stop loss price as needed.

• use the papertrading results to calculate the Sortino and Sharpe ratios to measure the performance of the trading strategy.