Expert Aggregation for Financial Forecasting
This work addresses portfolio optimization for financial investors by providing a method to enhance forecasting stability and returns, though it is incremental as it builds on existing online aggregation techniques.
The paper tackles the challenge of unstable algorithm performance in financial time series forecasting by applying Bernstein Online Aggregation (BOA) to combine forecasts from multiple machine learning models into long-short strategies, resulting in improved portfolio performance with higher Sharpe Ratio and lower shortfall.
Machine learning algorithms dedicated to financial time series forecasting have gained a lot of interest. But choosing between several algorithms can be challenging, as their estimation accuracy may be unstable over time. Online aggregation of experts combine the forecasts of a finite set of models in a single approach without making any assumption about the models. In this paper, a Bernstein Online Aggregation (BOA) procedure is applied to the construction of long-short strategies built from individual stock return forecasts coming from different machine learning models. The online mixture of experts leads to attractive portfolio performances even in environments characterised by non-stationarity. The aggregation outperforms individual algorithms, offering a higher portfolio Sharpe Ratio, lower shortfall, with a similar turnover. Extensions to expert and aggregation specialisations are also proposed to improve the overall mixture on a family of portfolio evaluation metrics.