Developing Pattern for Determining Trading Strategies, with an approach based on Future study, Fundamental Analysis, Feature Engineering and Machine Learning Algorithms.

Investors in stock market, always seeking for novel and efficient methods to predict stocks price and make the appropriate trading strategies. This paper propose a pattern for implementing appropriate strategies with a model composed of future studies, Fundamental analysis, expert trading rules and machine learning algorithms. First with use of feature study and expert’s opinion, stock market scenarios designed and a portfolio consist of 6 fundamental stocks is built. In next step for each selected stocks a model for stock price movement prediction is developed by means of related stocks data from 1393 to 1398 and 7 machine learning algorithms. Model inputs includes, technical indicators, technical trading rules, stock sign reading rules and stocks trading data. Results show that implementing proposed composed model for investment in stock market led to greater performance compared with Tehran stock market index and implementing short term trading strategies based on the model trained with light gradient boosting machine (LGBM) algorithm cause better performance in comparison with Buy& Hold and Technical trading Strategies.