How to train an uniform model for all symbols
ohfighting opened this issue · comments
Hi, edtechre
I was deeply attracted by pyb recently.
I want to train and backtest model using "Walkforward", but the framework of pyb is based on one symbol one model.
Mining global rule is a native need for ML, e.g. technology stocks(goog, msft) may have same latent law.
So, In pybroker.model(...,fn(symbol, train_data, test_data),...), "symbol" may can be a group of symbols, as well as train_data and test_data?
Hi @ohfighting,
Global models are not yet supported. This will be something I add eventually.
You can use a pretrained model to do this however. #46 has more information.