stocktrainer 0.1.1

Stock environment for training machine learning agents

StockTrainer: Stock Environment for Human

StockTrainer: Stocks Made Easy

StockTrainer is high level API data generator for training python machine learning models on stock/cryptocurrency data and is capable of running with Keras, Tensorflow, sklearn, and many other machine learning APIs

Capabilities:

• Predict day to day stock prices
• Use multiple days to predict next stock price
• Predict succeeding stock prices over multiple days
• Train a reinforcement learning agent to simulate stock trades

Documentation available soon ;)

stocktrainer is compatible with: Python 3.6+

Getting Started

The core of algorithm is the model, here is a simple LSTM model to based on 5 days of stock data to predict the next day

import keras
import numpy as np
from keras.models import Sequential
from keras.layers import Dropout ,BatchNormalization, LSTM, Dense 


model = Sequential()
#input shape 5 days of data 
#each day has 6 data points (open, close, high , low volums, adj CLose)
model.add(BatchNormalization(input_shape=(5, 6)))#batchnorm bc high values
model.add(LSTM(512, return_sequences=True, activation='relu'))
model.add(Dropout(0.2))
model.add(Dense(512, activation='relu'))
model.add(Dense(128, activation='relu'))	
model.add(Dense(1, activation='relu'))

model.compile(loss='mse', optimizer='adam')

Next import stocktrainer and create your environment

from stocktrainer import Env
environment = Env("Standard", "AAPL")

Time to collect your data to train!!!

test_percent =.30
shuffle =True
start_date ='2003-01-01'
end_date='now'
agent_memory = 5
seed = 42
trainx,testx,trainy, testy = environment.train_test(
  test_percent= test_percent, shuffle = shuffle, start_date=start_date, 
  end_date=end_date, agent_memory=agent_memory, seed=seed)

Futher information on parameters in Documentation

That's it now train and test your model

#fit model
model.fit(trainx, trainy, epochs=10, batch_size=128, verbose=2)
model.save('model.h5')

#evaluate model
model.evaluate(testx,testy )
#use model to predict
model.predict(testx)

More examples on samples folder in github

Installation

Using pip

pip install stocktrainer

or download directly: https://pypi.org/project/StockTrainer/