水准点(测量)
概率逻辑
采购
提前期
采购
计量经济学
点(几何)
经济
运筹学
计算机科学
运营管理
工程类
数学
几何学
管理
大地测量学
人工智能
地理
作者
Burakhan Sel,Stefan Minner
标识
DOI:10.1016/j.tre.2022.102881
摘要
Rate volatilities in seaborne freight markets lead charterers and ship owners to use financial agreements such as Forward Freight Agreements (FFA) for fixing freight rates in advance. The use of FFAs requires effective hedging policies since fixing freight rates in advance might cause both benefits and opportunity costs depending on future rate changes. We propose a data-driven hedging policy prescribing purchasing decisions for FFAs based on comparisons of FFA rates with future spot rate forecasts. The proposed approach is based on probabilistic forecasts instead of point forecasts because the most accurate forecasts in terms of predictive errors do not necessarily lead to the best decisions. We adjust spot rate forecasts by selecting percentiles that result in minimum prescriptive errors (i.e., cost) in cross-validation. Experiments on synthetic data show that the probabilistic forecast-based hedging policy outperforms the point forecast-based policies and benchmark policies, including data-driven policies from the literature. Experiments on Baltic Exchange data from 15 dry bulk and tanker routes confirm the performance of the proposed policy. Compared to two different point forecast-based policies defined in this study, the proposed approach achieves on average 3.31% and 3.02% total procurement cost reduction per route in 15 routes for four years testing period. It results in from 0.67% to 4.79% cost reductions against the benchmark policies.
科研通智能强力驱动
Strongly Powered by AbleSci AI