Fusing Sell-Side Analyst Bidirectional Forecasts Using Machine Learning

Sell-side analysts’ recommendations are primarily targeted at institutional investors mandated to invest across many companies within client-mandated equity benchmarks, such as the FTSE/JSE All-Share index. Given numerous sell-side for a single stock, making unbiased investment decisions is not often straightforward portfolio managers. This study explores use of historical create an fusion analyst forecasts that bidirectional accuracy optimised using random forest, extreme gradient boosting, deep neural networks, and logistic regression. We introduced 12-month rolling features generated from standard recommendations, coverage, point directional accuracy, while avoiding forward-looking biases. introduce novel “AI analyst” by fusing forecast analysts machine learning algorithms. observed added benefits these more than one systematically generating incrementally better prediction publicly available with Random forest algorithm showing highest relative performance. In highly volatile sectors, like resources, algorithms perform in low volatility suggesting importance bi-directional presence high volatility. Using feature importance, we observe incremental contribution features, relationships between volatility, accuracy. Furthermore, parameters regression identify and, initial target price some essential when modelling predictions.