A wavelet and neural network model for the prediction of dry bulk shipping indices

Shipping markets are typically volatile in nature, as manifest in the dynamics of freight rates. According to shipowners’ risk propensity, volatility-related decisions vary from what kind of contract (time/voyage charter) to engage in to whether to enter/exit the business. After a sharp drop in freight rates at the end of 2008, a discussion about appropriate risk management concepts and statistical tools is needed. Owing to the magnitude of investment required in shipping, any additional information regarding the future direction of market volatility is of the utmost importance. The ambition of this article is exactly the same: to study fluctuations in the freight rates of the Baltic Panamax route 2A and the Baltic Panamax route 3A, using a tool of analysis that is new to shipping economics: a hybrid model of wavelets and neural networks. The wavelet multiscale decomposition of time series reveals volatility dynamics across different time frequencies and will uncover patterns that will be used by neural networks for prediction.