As part of the University of Ghana Business School (UGBS) seminar series, the Department of Finance partnered with Tesah Capital for a seminar on the topic “What can unsupervised learning tell us about stocks listed on the Ghana Stock exchange”. The seminar was held at the UGBS Graduate School with Professor Elikplimi Komla Agbloyor, an Associate Professor of Finance, UGBS, as the speaker.
In his presentation, he explained unsupervised learning as a type of machine learning in which data is given to an algorithm and tries to make meaning from the data. He mentioned that a sample of 16 companies listed on the Ghana Stock Exchange was selected. The variables used were stock price, yield to date, volatility, Return on equity (ROE), Cash ratio, Net cash flow, Net Income, Earnings per share (EPS), Price Earnings (PE) ratio, number of years listed, value traded, dividend and market capitalisation he averred.
According to Prof. Elikplimi, the financial sector was the most significant sector among the 16 companies used in the study, with the highest dividend yield, profitability, and second-highest cash ratio. In contrast, the agricultural sector had the greatest price drop, highest cash ratio, and greatest volatility. The telecommunications industry had the best price-to-earnings ratio, liquidity, profitability, and market capitalization, as well as the lowest cash ratio and volatility. The Industrial sector also saw the second-highest price drop and the lowest dividend yield.
He noted that the k-means clustering, and hierarchical clustering approach of unsupervised learning were used to obtain four distinct clusters: Cluster 0, cluster 1, cluster 2, and cluster 3. Cluster 0 consisted of mature, stable, and dividend-paying firms; Cluster 1 of low profitability companies; Cluster 2 of the market giant and growth firm; and Cluster 3 of underperforming, high risk, and low dividend firms.
In his final statement, Prof. Elikplimi indicated the study aids in the understanding of the listed companies' traits or qualities. A point was also made that as the companies are divided into distinct groups, they show less correlation. As a result, this will aid in the construction of well-diversified portfolios by businesses and individuals making investment decisions.
copyright © 2016 University of Ghana Business School. All Rights Reserved