ML Trading Strategy Version 2

This repository is dedicated to present a machine learning approach to predict stock movements.

Results

Model 1

Description -

Input variables - eps_actual, month, eps_surprise, eps_growth, volume, normalized_value, 20 day logistic regression, 10 day logistic regression, 3 day logistic regression and 5 day logistic regression

Output variables - probability of the data point being a buy or a sell

Using past eps, eps surprise & eps growth has little effect on the output.

Model 2

Description -

Input variables - volume, month, normalized_value, 20 day logistic regression, 10 day logistic regression, 3 day logistic regression and 5 day logistic regression

Output variables - probability of the data point being a buy or a sell

Sample prediction on Visa (V) stock.

Model 2 - Rolling average probability in the last 3 days