In this paper, we develop a technique to calculate probability density of Amazon Spot Instance (SI) price. We consider both of the curve-fitting and historical similarities to compute the probability density of the SI price. Traditional SI bidding systems compute the bid price as a single value. That single value may not be suitable for all bidders. Different bidders have different urgency and preferences. The probability density of price may help users setting a bid-price considering both the urgency of the task and the condition of the market.