Publications Repository - Gdańsk University of Technology

Pair trading strategy is a well-known profitable strategy in stock, forex, and commodity markets. As most of the world stock markets declined during COVID-19 period, therefore this study is going to observe whether this strategy is still profitable after COVID-19 pandemic. One of the powerful algorithms of DBSCAN under the umbrella of unsupervised machine learning is applied and three clusters were formed by using market and accounting data. The formation of these three clusters was based on book value per share, earning per share, classification of sector, market capitalisation and with other factors formed from PCA on the returns of daily data of six months of the 80 sample firms for year 2019–2020. An average of 0:32% average excess monthly return with Sharpe ratio of 0:0012 and Treynor ratio of 0:0231 is to be observed in COVID-19 pandemic period. However, the result of risk-adjusted performance under Jensen's alpha is observed to be insignificant. The policy implication of this study, for different portfolios and fund managers is suggested to use machine learning approach to get positive and higher returns for their clients

Authors

• Muhammad Khalid Sohail,
• Abdul Raheman,
• Javid Iqbal,
• Muzammal Ilyas Sindhu,
• Abdul Staar,
• Muhammad Mushafiq link open in new tab ,
• Humaira Afzal