The Study of Using Machine Learning Algorithms to Construct Portfolio Formation

Abstract

The earning of the investors in the stock market is related to their investment choice and the behavior of the stocks. The key to investors making a more intelligent investment decision is to predict the price and volatility of the stocks accurately. In addition, building a portfolio formation is also an effective way to reduce investment risk. However, the stock price is beyond the predictive ability of any individual, and the portfolio formations are hard to build due to the stock market volatiles incessantly. In such a condition, this project attempts to propose a methodology for designing portfolio formations by using machine learning methods. The ultimate aim of this project is to predict the profit and volatility of stocks in the American stock market and use these predictions to construct portfolio formations. In this paper, the Support Vector Machine (SVM), Light Gradient Boosting Machine (LGBM), and Long Short-Term Memory model (LSTM) are used to predict the tendency of stocks. The experiment result shows that the LGBM model has the best performance on stocks price prediction. Based on the prediction of price, investing profit is easy to predict. The financial model called exponential weighted moving average (EWMA) is used to calculate the volatility of the stocks. After getting the prediction of profit and volatility, using the spectral clustering algorithm can get the portfolio formation. The result shows that the portfolio formation given by the proposed model has high profit and low volatility.

A Study on CNN Feature Extraction for Stock Price Prediction

“…(2) Exponential Moving Average An Exponential Moving Average (EMA) is a moving average (MA) that gives greater weight and importance to the nearest data point [7]. For the series {𝑥𝑥 𝑛𝑛 } define its exponential moving average 𝐸𝐸𝑀𝑀𝑀𝑀 𝑁𝑁 (𝑥𝑥 𝑛𝑛 ) with period N up to the nth term as:…”