jamiejoin / btc_codebase

A code base for testing btc trade algorithm

btc_codebase

A code base for testing btc trade algorithm

---

train mode:

train with holding_amount_data and btc price, model selected from LinearRegression, SVR, KNeighborsRegressor

python btc_model.py \
	--grid_search False \
	--read_path './data' \
	--save_path './output' \
	--train

eval mode:

eval model without training, assuming there is model existed in output directory

python btc_model.py \
	-- save_path './output' \
	--train False

Todo:

• Eval Mode: test trained model, return prediction with given data
• Plotting: draw price and prediction in real time
• More auxiliary data: e.g. mood data
• Model's Parameter Search: random search, bayesian
• Advanced Model: RNN coming