Statistical Vulnerability of Mean Prices in Automated Constant Product Market Makers

Abstract

to work correctly, many financial services using the blockchain technology require independent manipulation-resistant price feed providers. One of the most common providers of such prices, which works completely on the blockchain technology itself, is an automated constant product market maker, which is a tool for price calculation based on the amount of two assets under its control. Financial application developers should consider all the possible vulnerabilities which can be introduced by using such price providers. This is the reason for the relevance of research into the vulnerabilities of automated constant product market makers prices. We studied the effects of the automated constant product market maker fees on the mean price based on the real trading data from the blockchain and Binance exchange. The results show that mean price deviation between automated market makers and the market average shows high autocorrelation making it possible to predict a future mean deviation of the prices between the exchanges. The simulation results show the predicted values for different prediction time frames. Based on the predictability of a future mean deviation of the prices, potential critical statistical vulnerabilities in the financial applications using the mean prices provided by constant product market makers are described, and vulnerability mitigation recommendations are given. The practical value to the blockchain application developers and smart contract security experts is that now they can prevent or mitigate potential critical statistical vulnerabilities in their applications.