A Simulation Model for Predicting Tramp Shipping Supply

Tramp shipping is a key part of the maritime industry which operates mainly in the spot market relying on voyage-by-voyage contracting, which forces them to reposition frequently in search of favourable cargoes. Market dynamics therefore emerge from how regional cargo demand aligns with the shifting distribution of available vessels. Forming multi-month forecasts of this evolving relationship between demand and supply is essential for market participants seeking to respond to rapidly changing market conditions. The supply side of the relationship remains relatively unexplored, particularly addressing vessels reposition and evolvement of regional availability over time. Bridging this gap, this paper introduces a simulation-based framework that models behaviour at the individual-vessel level and generates forward-looking forecasts of regional tramp-hipping supply over a 90-day horizon. The regional tramp shipping vessel supply prediction is presented through a mathematical formulation and an agent-based framework in which each vessel acts as an autonomous agent responding to market conditions is developed. To this end Neural network-based stochastic estimators of vessel behaviour are produced from historical data and used to simulate vessel-level decisions, yielding coherent forecasts of regional vessel supply. The framework is evaluated on the clean petroleum products market using datasets spanning the period 2020-01-01 to 2024-06-30. The results are compared with a regression benchmark relying on macro-economic variables, and the developed framework show to achieve higher supply prediction accuracy in 23 of 24 region-(vessel-segment) combinations, reducing average mean absolute percentage error from 13.01% to 4.79%.