You are here

Talk Abstract: Mahsa Moosavi and Jeremy Clark

Trading on-chain: how feasible is regulators’ worst-case scenario?

The Landscape of Cryptocurrency Exchanges. First, we present an evaluation framework in which we compare five major trade execution systems for blockchain-based assets: (1) central exchanges (CEX) (e.g., Binance, Bitfinex),(2) on-chain dealers (e.g., Uniswap), (3) hybrid designs (e.g., EtherDelta, 0x, IDEX), and (4) on-chain order books. Using the evaluation framework, we argue that the fully on-chain exchanges have a superior threat model yet they rarely exist in practice because they tend to (i) be slow and (ii) are difficult to regulate. We show how slow in the rest of the talk. Blockchain Limitations and Solution. Here, we run through blockchain limitations that make it infeasible to drop a continuous-time order book onto a blockchain when designing a fully on-chain order-driven exchange. These limitations include (i) block intervals are slow and not continuous, (ii) there is no support for accurate time-stamping, (iii) transactions can be dropped or reordered by miners, and (iv) fast traders can react to submitted orders/ cancellations when broadcast to network but not in a block and have their orders appear first (front-running). We then introduce an alternative data structure to continuous-time order book; a call market, and describe how it works. Call Market Design. We give an overview of our proof-of-concept system, Lissy, and its primary operations. We discuss why we choose a priority queue (PQ) as the core data structure for the call market and what priority queue implementation (among 5 variants we implemented and deployed) best optimizes the performance. We do this by presenting the results of a variety of performance experimentations and optimizations we performed. Measurements. The main research question is how many orders can be processed in a single Ethereum transaction when closing the call market, using Ethereum today. To address this question, we illustrate the results from a variety of tests and optimizations, we performed to benchmark the performance, including testing the full call market on an Ethereum Layer-2 scaling solution; Arbitrum.

Discussion. We discuss the key design decisions and extensions of Lissy. We briefly run through (1) front-running and price manipulation resistance, token divisibility and ties, (2) order cancellations, (3) market clearing price, (4) scheduling events (openmarket()/closeMarket() invocations and paying the cost), and (5) collateralization options.

Relevance to a Canadian Audience: This research work is a practical example of using blockchain technologies for public interest in Canada. The outcomes of this research, which will be discussed in the talk, will help theAMF (a Canadian regulatory agency) and other public sectors in Canada (e.g., Financial Transactions and Reports Analysis Centre of Canada (FINTRAC) and Canada Revenue Agency (CRA)) forecast how trading can be impacted by blockchain technologies and offer a more sophisticated understanding of how these technologies can be used in financial use cases.