Code and Capital: ICEF Students Learn Machine Learning for Investment Analysis
Learning practical skills alongside theoretical knowledge is of particular importance in majors such as finance, investment and economics. Given the rapid growth of technology, especially for data analysis and machine learning, the proficiency in modern programming languages such as Python is simply critical to finance career success.
This past March, ICEF delivered the online workshop “Financial Machine Learning: Linear models and clustering”, with Ivan Gudkov, Lead Economist at the RF Central Bank Risk Analysis Service, and Associate Professor Elena Dimova, who teaches Investment Portfolio Management of ICEF, as hosts.
For the audience of 75 students, this workshop proved an excellent opportunity to see real examples of how the ML tools can help predict stock prices and identify equivalents of stocks.
By demonstrating code-and-data work, Ivan Gudkov made it possible for the participants to gain a deeper insight into ML methods and their application.
To consolidate their skills, the participants were given tasks to be completed as part of their Investment Portfolio Management course. Those tasks involved using Python for data analysis and model building, enabling users to put their skills into practice.
Task 1: Predict stock prices. This task required learners to select 7 to 10 stocks, source their price data for the last five years, then calculate returns, build a Lasso linear regression model, benchmark the results and identify how pricing of the stocks under analysis is influenced by few selected factors. It also involved normality and correlation tests on model residuals.
Task 2: Identify equivalents of stocks. This task required learners to download the data on securities, group it using the k-means method, then interpret the results and calculate the number of parameter-based clusters. More advanced users were free to apply their own algorithms without using any built-in Python libraries.
The above tasks offer a good way to consolidate the knowledge gained and to hone the ML data analysis skills for more effective investment decision-making.
Financial Machine Learning, which uses linear models and clustering algorithms, can significantly enhance the efficiency of investment portfolio management. Linear models allow predicting asset prices based on historical data, serving as an effective tool to support investment decisions and asset trading strategies.
Grouping and clustering, in turn, allow identifying complementary assets, patterns of dependence, and alternative investment targets, helping to cushion risks and diversity portfolios. All this makes the ML methods a useful aide in market analysis, predicting trends and fine-tuning investment strategies.