Animals’ escaping behavior can be seen as an economic game between predation risk and foraging cost, which is measured by flight distance in the standard deterministic model. However, this fight distance should not be constant even in a stationary environment because it will also be affected by the predator’s parameters. In order to better understand animal’s escaping behavior, this paper investigates goldfish escaping strategies on the relevant position against a toy submarine in quantification, and further proposed a probabilistic model over the factors that influence the probability to escape. This research has determined goldfish’s hyperparameters in their escaping habits based on recording data and found that the flight distance is not only influenced by the risk and cost in previous research but also the approaching direction between goldfish and predator and approaching speed. The proposed probabilistic model shows that flight distance generally increases with the predator’s approaching speed and the extent of approaching direction against the vertical plane. The proposed model is finally cross-validated with the true video clip.