Key points are not available for this paper at this time.
Decision-making in the stock market is convoluted as it requires significant trading experience and knowledge. Faced with a huge range of stocks, investors in the stock market may be overwhelmed and daunted by the number of choices available on their plates. Financial institutions also face the problem of recommending stocks that cater to the investors’ trading strategies. In this paper, we develop and explore three different approaches that can be used to build a stock recommender system that takes into account the investors’ needs and interests. Furthermore, to ensure the profitability of the recommended portfolios and to support the investors in making risk-informed decisions, we evaluate their Expected Returns and Value-at-Risks. Hierarchical clustering was performed to better understand groups of traders that are similar in their needs and preferences and thus computational efficiency was improved. Three stock recommender systems, the K-Nearest Neighbour (kNN), Singular Value Decomposition (SVD) and Association Rule Mining (ARM) was explored and evaluated. Our short term, medium term and long term portfolios average rate of return was 4.15%, 10.24% and 23.17% respectively. Our study demonstrates promising recommendation results that not only catered to the user's profile but also contributed to the portfolio profitabilities with minimal financial loss. Stock recommendations were beneficial to users when their preferences were considered.
Gonzales et al. (Sat,) studied this question.