Adversarial blockchain queues and trading on a CFMM

TL;DR

This paper models blockchain queueing environments with adversarial schedulers, analyzing their impact on transaction ordering and price sensitivity in CFMM DEXs, including effects of MEV activity.

Contribution

It introduces a probabilistic model for adversarial blockchain queues and applies it to analyze order flow and price impact in CFMM DEXs.

Findings

Conditions for bulk-service queue behavior with priority discipline

Expressions for transaction positioning in the queue

Statistical models for price impact with and without MEV activity

Abstract

We describe a plausible probabilistic model for a blockchain queueing environment in which rational, profit-maximising schedulers impose adversarial disciplines on incoming messages containing a payload that encodes a state transition in a machine. The model can be specialised to apply to chains with fixed or variable block times, traditional priority queue disciplines with `honest' schedulers, or adversarial public mempools. We find conditions under which the model behaves as a bulk-service queue with priority discipline and derive practical expressions for the relative block and message number of a transaction. We study this setup in the context of orders to a CFMM DEX where the execution price a user receives may be quite sensitive to its positioning in the chain -- in particular, to a string of transactions scheduled for prior execution which is not knowable at the time of order creation. We derive statistical models for the price impact of this order flow both in the presence and absence of MEV extraction activity.