Lecture by Prof. Robert Ślepaczuk and Krzysztof Płachta – 03 November

Krzysztof Płachta, affiliated with Citi, and Professor Robert Ślepaczuk will present a study entitled „Machine Learning for Daily Return Direction Forecasting: A Comparative Study with Explainable AI Insights”. The lecture will take place on 3 November at 18:30 in Room B002 at the Faculty of Economic Sciences.

There is also the possibility to attend the seminar online. Those interested in remote participation are kindly asked to contact the event’s organizers at: b/@vc_8nRw'ea=%?$2]#[Tp/^zW%T[mznGywHqy or the Communication Section at: b/@vc_8nRw'ea=%?$2x]h}]#[S|+dSC}PioqKj0yHfqnfYj.

On behalf of the organizers, we kindly ask all participants to arrive or log in to the meeting at least 5 minutes in advance. The lecture will last approximately 45 minutes, followed by a discussion session, to which all participants are warmly invited.

---

This study addresses two interrelated objectives. First, we conduct a comparative evaluation of six supervised learning models: Lasso, Random Forest, LightGBM, LSTM, and two feedforward neural networks, for forecasting the next-day direction of SPY returns. Using a long-horizon of 25 years, expanding-window backtesting framework, we generate strictly out-of-sample forecasts and assess model performance based on risk-adjusted metrics, with the Sortino ratio as the primary criterion. Second, to address the opacity of complex machine learning models, we apply a model-agnostic explainable AI approach to the best-performing model. This involves systematically removing individual features and predefined feature groups, followed by full retraining with hyperparameters optimization, to quantify each variable’s contribution to overall performance. The results were mixed. Lasso, LightGBM, and most notably Random Forest – which achieved Sortino ratio of 0.61, over 30% higher than the benchmark, and exhibited a substantially lower maximum drawdown – outperformed the buy-and-hold strategy. On the other hand, contrary to expectations, all neural network-based models significantly underperformed. In the second part, feature importance analysis for the Random Forest model suggests that signals based on technical indicators, foreign exchange, and commodity prices consistently contributed to improved performance, while certain interest rate and equity index features were detrimental. These findings emphasize the importance of thoughtful model design and feature selection in the development of machine learning strategies for financial forecasting.