Abstract: The newsboy problem has always been an important issue in inventory management.The exiting newsboy models describe and measure risk using Mean-Variance methodology and expected utility objective.These methods assume full knowledge of the demand probability distribution,however,in reality,it is often difficult to completely characterize the demand.This paper selects probability forecast to describe limited demand distribution,and constructs an online risk-reward model for the newsboy problem under probabilistic forecasts.Comparing with the existing studies,this model can help the new sboy choosing the minimal risk strategy with great flexibility,according to his own restrained reward and probabilistic forecast.

Key words: newsboy problem, online algorithm, risk, probabilistic forecast