2026 Theses Doctoral

From Myerson to Automated Markets: Foundations of Decentralized Finance Microstructure

Milionis, Iason

Recent years have seen the development of a new paradigm of internet-native markets: decentralized systems. These systems have enabled decentralized finance (DeFi), namely, the proliferation of automated financial markets that operate without trusted intermediaries.

This thesis lays the foundations for decentralized finance and, in doing so, explores the design space of automated financial markets. Our starting point is automated market makers (AMMs)—the core primitive in DeFi—which implement continuously operating algorithmic markets that enable traders to exchange one asset for another programmatically. We introduce a general framework for exchange design wherein liquidity providers (LPs) who deposit the assets traded on these markets specify non-increasing demand curves representing the quantity of risky assets to be held at each price. This framework captures both the novel paradigm of AMMs and the traditional trading paradigm of limit order books (LOBs). We define an exchange complexity measure that formalizes the approximation–expressiveness trade-off such designs need to make.

We then use this setting to pose the problem of incentive-compatible (IC) exchange design for a monopolist LP in the spirit of Myerson's optimal auction theory. We characterize the profit-maximizing IC AMM via a generalization of Myerson's virtual values. Using these generalized virtual values, we interpret bid–ask spreads as the joint effect of adverse selection and monopoly pricing. We quantify LP welfare in AMMs under asymmetric information, identifying what we term "loss-versus-rebalancing" (LVR, pronounced "lever") as the central adverse selection cost resulting from stale prices on AMMs being exploited by informed traders (arbitrageurs). We validate our model empirically against observed LP returns and show how AMMs can be redesigned to reduce or eliminate LVR.

Finally, we extend this model by introducing exchange trading fees proportional to trading volume and stochastic, discrete arbitrageur arrivals, and obtain scaling laws for arbitrage profits. Together, these contributions place exchange market microstructure and mechanism design on common ground and provide a microstructural foundation for analyzing and engineering trading markets.