Hermann Noubissie Noussa
Technological trends in the portfolio management industry
One of the biggest challenges in investing is getting the maximum return on an investment for a given level of risk. There are several alternatives to try to resolve the question, between making investments in banks, investing in a fund or investing in the stock market. In other words, the problem intrinsically comes down to the choice of asset class (stocks, bonds, fixed income products, etc.) in which one should invest. The portfolio management approach explores this question by orienting the question differently: instead of isolating each asset or asset class, the idea is to review the coexistence of a set of assets or asset classes. in a wallet. Portfolio management is then based on the selection and diversification of assets to meet the financial objectives and risk tolerance of an individual or an institution. The portfolio management process can be broken down into three stages:
- The planning stage: which consists of understanding the client’s needs (objectives and constraints) and translating into the issuance of an investment policy (IPS).
- The execution stage: this includes asset allocation, securities analysis and portfolio construction. This step allows you to build a portfolio based on the client’s IPS.
- Feedback stage: this last stage consists of monitoring the performance of the portfolio and rebalancing the contributions of the securities if any changes arise (market conditions or circumstances with the client).
Technologies in the portfolio management industry are primarily involved in the execution stage which is the heart of the value creation process.
The exponential growth and digitalization of data in the global arena is driving the creation of additional data to explore in the portfolio management industry. In 2013, IBM estimated that 90% of data was recorded in the previous two years. The portfolio managers now have the task of determining the significant data and efficient technologies for processing such data:
- Big Data: The term “Big Data” refers to the use of big data (structured or unstructured) across different technologies to generate value. The data can come from traditional sources (stock exchanges, companies or governments) or from non-traditional sources such as media networks (Twitter, Facebook, etc.), image data or others: this data is called alternative. Analysis of social media data serves as the basis for obtaining key indicators of market sentiment and trends for certain products and services. Image data such as those from satellites make it possible to visualize economic conditions. Alternative data is mostly traded in realtime;therefore, portfolio managers must use advanced statistical techniques to analyze it.
- Advanced and analytical tools: Artificial Intelligence and machine learning. With the advent of big data, the technologies used are those around artificial intelligence, that is to say techniques developing techniques capable of simulating tasks traditionally associated with human intelligence (reasoning, learning, aptitude decision making, …). These technologies are now essential. Machine learning or “Machine Learning” brings together all the techniques that extract knowledge from big data without making assumptions about the probability distribution of the data. Several algorithms are used such as neural networks, decision trees, logistic regression, the k nearestneighbors’method. For example, decision trees can be used to automatically choose the investment strategy according to IPS guidelines.
PolyFinances has carried out several applications of technologies in the portfolio management sector:
- Machine learning algorithm based on the decision tree technique: The algorithm builds a portfolio of stocks that are likely to beat the S&P 500 Index based on random forest regression on a monthly basis. The portfolio is completely rebuilt every month. Transaction costs and dividends have not been recognized. The algorithm takes as input the financial ratios of all the constituents of the S&P 500. By building each month, a portfolio with the 10 best stocks with the best chances of beating the index, the average arithmetic return in December 2018 of the stocks hadaexcess return of + 1.48%, while the compound return over 15 years with excess return of + 1160%. Data are from December 2003 to December 2018 imported on Bloomberg.
- Automation of fundamental analysis: a statistical data visualization tool necessary for fundamental analysis.
- TSX Filter – Market Sentiment Analysis from Twitter: Software to determine which TSX stocks have the highest market sentiment through the Twitter platform. Based solely on the results of the software, a portfolio was built with the top 10 stocks and renewed quarterly. No transaction costs have been taken into account. Over a 10-year horizon (July 2010 – March 2020), the portfolio’s valuedoubled,and the portfolio’s average return was four times that of the TSX.
Today with the rise of big data, portfolio managers are turning to different technologies just for the sake of delivering better returns to the client. These technologies as explored by PolyFinances seem to provide comfortable results.
CFA Institute. (2020). Alternative investments and portfolio management. Charlottesville: Wiley Global Finance.
Investopedia. (2020, Aout 01). Portfolio Management. Récupéré sur Investopedia: https://www.investopedia.com/terms/p/portfoliomanagement.asp