This paper provides a comprehensive empirical analysis of the economics and dynamics behind arbitrages between centralized and decentralized exchanges (CEX-DEX) on Ethereum. We refine heuristics to identify arbitrage transactions from on-chain data and introduce a robust empirical framework to estimate arbitrage revenue without knowing traders' actual behaviors on CEX. Leveraging an extensive dataset spanning 19 months from August 2023 to March 2025, we estimate a total of 233.8M USD extracted by 19 major CEX-DEX searchers from 7,203,560 identified CEX-DEX arbitrages. Our analysis reveals increasing centralization trends as three searchers captured three-quarters of both volume and extracted value. We also demonstrate that searchers' profitability is tied to their integration level with block builders and uncover exclusive searcher-builder relationships and their market impact. Finally, we correct the previously underestimated profitability of block builders who vertically integrate with a searcher. These insights illuminate the darkest corner of the MEV landscape and highlight the critical implications of CEX-DEX arbitrages for Ethereum's decentralization.

Byzantine fault-tolerant consensus protocols have provable safety and liveness properties for static validator sets. In practice, however, the validator set changes over time, potentially eroding the protocol's security guarantees. For example, systems with accountable safety may lose some of that accountability over time as adversarial validators exit. As a result, protocols must rate limit entry and exit so that the set changes slowly enough to ensure security. Here, the system designer faces a fundamental trade-off. Slower exits increase friction, making it less attractive to stake in the first place. Faster exits provide more utility to stakers but weaken the protocol's security. This paper provides the first systematic study of exit queues for Proof-of-Stake blockchains. Given a collection of validator-set consistency constraints imposed by the protocol, the social planner's goal is to provide a constrained-optimal mechanism that minimizes disutility for the participants. We introduce the MINSLACK mechanism, a dynamic capacity first-come-first-served queue in which the amount of stake that can exit in a period depends on the number of previous exits and the consistency constraints. We show that MINSLACK is optimal when stakers equally value the processing of their withdrawal. When stakers values are heterogeneous, the optimal mechanism resembles a priority queue with dynamic capacity. However, this mechanism must reserve exit capacity for the future in case a staker with a much higher need for liquidity arrives. We conclude with a survey of known consistency constraints and highlight the diversity of existing exit mechanisms.

Modern blockchains guarantee that submitted transactions will be included eventually; a property formally known as liveness. But financial activity requires transactions to be included in a timely manner. Classical liveness does not guarantee this, particularly in the presence of a motivated adversary who benefits from censoring transactions. We define censorship resistance as the amount it would cost the adversary to censor a transaction for a fixed interval of time as a function of the associated tip. This definition has two advantages, first it captures the fact that transactions with a higher miner tip can be more costly to censor, and therefore are more likely to swiftly make their way onto the chain. Second, it applies to a finite time window, so it can be used to assess whether a blockchain is capable of hosting financial activity that relies on timely inclusion. We apply this definition in the context of auctions. Auctions are a building block for many financial applications, and censoring competing bids offers an easy-to-model motivation for our adversary. Traditional proof-of-stake blockchains have poor enough censorship resistance that it is difficult to retain the integrity of an auction when bids can only be submitted in a single block. As the number of bidders n in a single block auction increases, the probability that the winner is not the adversary, and the economic efficiency of the auction, both decrease faster than 1/n. Running the auction over multiple blocks, each with a different proposer, alleviates the problem only if the number of blocks grows faster than the number of bidders. We argue that blockchains with more than one concurrent proposer can have strong censorship resistance. We achieve this by setting up a prisoner’s dilemma among the proposers using conditional tips.

The current Proposer-Builder Separation (PBS) equilibrium has several builders with different backgrounds winning blocks consistently. This paper considers how that equilibrium will shift when transactions are sold privately via order flow auctions (OFAs) rather than forwarded directly to the public mempool. We discuss a novel model that highlights the augmented value of private order flow for integrated builder searchers. We show that private order flow is complementary to top-of-block opportunities, and therefore integrated builder-searchers are more likely to participate in OFAs and outbid non integrated builders. They will then parlay access to these private transactions into an advantage in the PBS auction, winning blocks more often and extracting higher profits than non-integrated builders. To validate our main assumptions, we construct a novel dataset pairing post-merge PBS outcomes with realized 12-second volatility on a leading CEX (Binance). Our results show that integrated builder-searchers are more likely to win in the PBS auction when realized volatility is high, suggesting that indeed such builders have an advantage in extracting top-of-block opportunities. Our findings suggest that modifying PBS to disentangle the intertwined dynamics between top-of-block extraction and private order flow would pave the way for a fairer and more decentralized Ethereum.

The classic wisdom-of-the-crowd problem asks how a principal can "aggregate" information about an unknown state of the world from agents without understanding the information structure among them. Such aggregation obviously has large social and private value, especially when it concerns important social or economic events. Therefore, it is important to understand the limits of such an exercise: without specific assumptions on the information agents have, how can we aggregate it? Classic results by Prelec et al. [2017] (henceforth PSM) and Arieli et al. [2017] show that that even when agents' signals are i.i.d. conditional on the state, knowing the first-order beliefs of even an infinite set of agents is generally not sufficient to learn the state. In the terminology of econometrics, there is an "identification problem."

When consumption of a good causes externalities, market outcomes may be inefficient. Economists have long recognized this and suggested as a remedy a "Pigouvian tax" equal to the monetary equivalent of the harm done to others. Despite their intuitive appeal, Pigouvian taxes are rarely seen in practice. Conversely, public discourse often involves regulations such as consumption caps or even prohibition of the activity, which economists consider "non-market" solutions. Can natural preferences of the planner justify this?

We present a general, efficient technique for providing contextual predictions that are "multivalid" in various senses, against an online sequence of adversarially chosen examples $(x,y)$. This means that the resulting estimates correctly predict various statistics of the labels $y$ not just marginally -- as averaged over the sequence of examples -- but also conditionally on $x \in G$ for any $G$ belonging to an arbitrary intersecting collection of groups $\mathcal{G}$. We provide three instantiations of this framework. The first is mean prediction, which corresponds to an online algorithm satisfying the notion of multicalibration from Hebert-Johnson et al. The second is variance and higher moment prediction, which corresponds to an online algorithm satisfying the notion of mean-conditioned moment multicalibration from Jung et al. Finally, we define a new notion of prediction interval multivalidity, and give an algorithm for finding prediction intervals which satisfy it. Because our algorithms handle adversarially chosen examples, they can equally well be used to predict statistics of the residuals of arbitrary point prediction methods, giving rise to very general techniques for quantifying the uncertainty of predictions of black box algorithms, even in an online adversarial setting. When instantiated for prediction intervals, this solves a similar problem as conformal prediction, but in an adversarial environment and with multivalidity guarantees stronger than simple marginal coverage guarantees.

We introduce a simple but general online learning framework in which a learner plays against an adversary in a vector-valued game that changes every round. Even though the learner's objective is not convex-concave (and so the minimax theorem does not apply), we give a simple algorithm that can compete with the setting in which the adversary must announce their action first, with optimally diminishing regret. We demonstrate the power of our framework by using it to (re)derive optimal bounds and efficient algorithms across a variety of domains, ranging from multicalibration to a large set of no-regret algorithms, to a variant of Blackwell's approachability theorem for polytopes with fast convergence rates. As a new application, we show how to ``(multi)calibeat'' an arbitrary collection of forecasters --- achieving an exponentially improved dependence on the number of models we are competing against, compared to prior work.

We show how to achieve the notion of "multicalibration" from Hebert-Johnson et al. (2018) not just for means, but also for variances and other higher moments. Informally, this means that we can find regression functions which, given a data point, can make point predictions not just for the expectation of its label, but for higher moments of its label distribution as well—and those predictions match the true distribution quantities when averaged not just over the population as a whole, but also when averaged over an enormous number of finely defined subgroups. It yields a principled way to estimate the uncertainty of predictions on many different subgroups—and to diagnose potential sources of unfairness in the predictive power of features across subgroups. As an application, we show that our moment estimates can be used to derive marginal prediction intervals that are simultaneously valid as averaged over all of the (sufficiently large) subgroups for which moment multicalibration has been obtained.

There is great interest in whether machine learning algorithms deployed in consequential domains (e.g. in criminal justice) treat different demographic groups "fairly." However, there are several proposed notions of fairness, typically mutually incompatible. Using criminal justice as an example, we study a model in which society chooses an incarceration rule. Agents of different demographic groups differ in their outside options (e.g. opportunity for legal employment) and decide whether to commit crimes. We show that equalizing type I and type II errors across groups is consistent with the goal of minimizing the overall crime rate; other popular notions of fairness are not.

We consider settings in which we wish to incentivize myopic agents (such as Airbnb landlords, who may emphasize short-term profits and property safety) to treat arriving clients fairly, in order to prevent overall discrimination against individuals or groups. We model such settings in both classical and contextual bandit models in which the myopic agents maximize rewards according to current empirical averages, but are also amenable to exogenous payments that may cause them to alter their choices. Our notion of fairness asks that more qualified individuals are never (probabilistically) preferred over less qualifie ones [8].

We study how privacy technologies affect user and advertiser behavior in a simple economic model of targeted advertising. In our model, a consumer first decides whether or not to buy a good, and then an advertiser chooses an advertisement to show the consumer. The consumer's value for the good is correlated with her type, which determines which ad the advertiser would prefer to show to her---and hence, the advertiser would like to use information about the consumer's purchase decision to target the ad that he shows.

We study the problem of implementing equilibria of complete information games in settings of incomplete information, and address this problem using "recommender mechanisms." A recommender mechanism is one that does not have the power to enforce outcomes or to force participation, rather it only has the power to suggestion outcomes on the basis of voluntary participation. We show that despite these restrictions, recommender mechanisms can implement equilibria of complete information games in settings of incomplete information under the condition that the game is large---i.e. that there are a large number of players, and any player's action affects any other's payoff by at most a small amount.

We consider the design of the revenue maximizing mechanism for a seller with a fixed capacity of C units selling over T periods to buyers who arrive over time. The buyers have single unit demand and multi-dimensional private information– both their value for the object and the deadline by which they must make a purchase are unknown to the seller. This contrasts with previous work where buyers have single dimensional private information– deadlines are publicly observed and only values are private. Here, the optimal mechanism can be computed by running a dynamic stochastic knapsack algorithm. However, these mechanisms are only optimal with private deadlines when the calculated allocation rule is monotone– buyers with higher values and later deadlines should be allocated with higher probability. Such monotonicity only arises in very special cases. By contrast, in the classic static environment of Myerson [7] monotonicity is only violated for ‘irregular’ value distributions. Myerson characterizes the optimal mechanism by a procedure he calls ‘ironing.’ We characterize the optimal mechanism in our general dynamic environment by providing the dynamic counterpart of ironing. We show that only a subset of the monotonicity constraints can bind in a solution of the seller's dynamic programming problem. The optimal mechanism can be characterized by ‘relaxing’ these constraints with their appropriate dual multiplier. Further, the optimal mechanism can be implemented by a series of posted prices followed by a ‘biased’ auction in the final period where buyers have the auction biased in their favor depending on their arrival time. Our theoretical characterization complements the existing computational approaches for ironing in these settings (e.g. Parkes et al. [10]).

Inspired by online advertisement exchange systems, we study a setting where potential buyers of a unique, indivisible good attempt to purchase from a central seller via a set of intermediaries. Each intermediary has captive buyers, and runs an auction for a 'contingent' good. Based on the outcome, the intermediary bids in a subsequent upstream auction run by the seller. In this paper, we study the equilibria and incentives of intermediaries and the central seller.