Applying Probabilistic Model Checking to Financial Production Risk Evaluation and Control: A Case Study of Alibaba's Yu'e Bao

The core challenge of financial companies is to maximize business profits and minimize capital risks in enterprise operations management, mainly by considering their liquidity risk and liquidity surplus control. Thus, an effective approach to financial production risk evaluation and control must be found to positively determine the optimal cash reserve ratio. In this paper, we were motivated to analyze Ali Pay data sets, published by Alibaba's Yu'e Bao, to demonstrate that purchase amounts and redemptions strongly influence user behaviors. To this end, first, we employ a probabilistic model to verify the uncertainty of user behaviors by computing the probabilities for financial production risk evaluation and control. Second, investors' behaviors are formalized into a discrete-time Markov chain model (DTMC) that can factually describe the probability profiles of investors' purchases and redemptions. Third, we use probabilistic computation tree logic (PCTL) to determine the probability that users will exhibit purchasing or redemption behaviors. Furthermore, the probabilistic model-checking tool PRISM, which takes the formal model and properties as input and outputs quantitative results, is employed to perform automatic verification. Fourth, based on the verification results, a strategy evaluation model that considers profits and risks is proposed to measure the capital reserve ratio. Finally, we employ a real-world test data set that includes 2.8 million transaction log records published by Ant Financial Services. These data are used to conduct experiments to demonstrate the effectiveness of our proposed method.