Events

Date: 14 May 2026   Time: 13:00 - 14:00

Location: MB-503

Understanding price formation in markets with no central exchange—such as the informal currency market for the Cuban peso—has traditionally been challenging due to the lack of order books, tick rules, or centralized liquidity. In the first part of this talk, I will briefly present recent empirical findings on the Cuban informal market, showing that it is inefficient by standard metrics (runs test, variance ratio test), yet exhibits hidden structure recoverable via Empirical Mode Decomposition and Hidden Markov Models. We further show that inefficiencies create short-term predictability, which we exploit using GRU-type neural networks fine-tuned on social-media-derived Walrasian reference prices.

While these results characterize the symptoms of informal markets, they do not explain their microscopic dynamics nor how they differ from formal financial systems. This brings us to the central contribution of our talk: a unified mean-field framework connecting informal markets to formal Limit Order Book (LOB) dynamics.

We model both systems as interacting particle systems on a one-dimensional price lattice, governed by master equations for order entry, cancellation, and execution. The key parameter is a preferential interaction strength Ψ, which tunes the system from random, uncoordinated matching (Ψ=0, informal markets) to strictly optimal, mutually attractive trades (Ψ→∞, LOB regime). Using a grand-canonical ensemble, we identify thermodynamic analogues—interaction energy and price-dependent chemical potentials—that remain meaningful across both limits. Numerical simulations validate the transition, and we provide an analytical solution for the stationary symmetric case of the informal market.