David Easley

David Easley, Yoav Kolumbus, Eva Tardos

We analyze the performance of heterogeneous learning agents in asset markets with stochastic payoffs. Our main focus is on comparing Bayesian learners and no-regret learners who compete in markets and identifying the conditions under which each approach is more effective. Surprisingly, we find that low regret is not sufficient for survival: an agent can have regret as low as $O(\log T)$ but still vanish when competing against a Bayesian with a finite prior and any positive prior probability on the correct model. On the other hand, we show that Bayesian learning is fragile, while no-regret learning requires less knowledge of the environment and is therefore more robust. Motivated by the strengths and weaknesses of both approaches, we propose a balanced strategy for utilizing Bayesian updates that improves robustness and adaptability to distribution shifts, providing a step toward a best-of-both-worlds learning approach. The method is general, efficient, and easy to implement. Finally, we formally establish the relationship between the notions of survival and market dominance studied in economics and the framework of regret minimization, thus bridging these theories. More broadly, our work contributes to the understanding of dynamics with heterogeneous types of learning agents and their impact on markets.

Soumya Basu, David Easley, Maureen O'Hara et al.

Blockchain-based cryptocurrencies prioritize transactions based on their fees, creating a unique kind of fee market. Empirically, this market has failed to yield stable equilibria with predictable prices for desired levels of service. We argue that this is due to the absence of a dominant strategy equilibrium in the current fee mechanism. We propose an alternative fee setting mechanism that is inspired by generalized second price auctions. The design of such a mechanism is challenging because miners can use any criteria for including transactions and can manipulate the results of the auction after seeing the proposed fees. Nonetheless, we show that our proposed protocol is free from manipulation as the number of users increases. We further show that, for a large number of users and miners, the gain from manipulation is small for all parties. This results in users proposing fees that represent their true utility and lower variance of revenue for miners. Historical analysis shows that Bitcoin users could have saved $272,528,000 USD in transaction fees while miners could have reduced the variance of fee income by an average factor of 7.4 times.