Time charter rate forecasting by Parsimonious Intelligent Support Vector regression Search Engine

Reliable and accurate forecasts of time charter rates are crucial for shipowners navigating the highly volatile global shipping market. This paper introduces a novel framework, the Parsimonious Intelligent Support Vector Regression Search Engine (SVRIMSE), designed to enhance the forecasting of time charter rates. Unlike traditional models that rely on subjective assumptions regarding causality, time lags, and training set sizes, which can compromise accuracy and pose challenges for non-experts, SVRIMSE offers a more robust solution. It not only delivers precise forecasts but also autonomously identifies significant explanatory variables without requiring prior knowledge or assumptions about the shipping market. Our comparative analysis demonstrates that SVRIMSE outperforms several baseline models across three error metrics. Notably, the results highlight Fleet Development (Number) and Scrap Value as the most influential variables in forecasting time charter rates. These findings provide both accurate forecasts and valuable insights into the factors impacting the freight market, offering a significant contribution to the field.