The impact of aggregator composition on liquidity provision in over-the-counter markets

本项目的目的是研究在场外交易市场中，交易者和给定数量的流动性提供者之间的相互作用。场外交易市场的一个例子是外汇市场，这是世界上最大的金融市场，每日交易量达数万亿美元。与股票市场不同，股票市场是由具有透明限价订单簿的交易所组成的，外汇市场取决于与流动性提供者的双边交易。这通常是在一个聚合器，一种技术，巩固流动性从几个来源的形式的限价ordbook.This研究旨在解决一个空白的文献中关于特定的问题，以流动性提供在一个“盲拍卖”的柜台设置。到目前为止，这个话题受到的关注有限，与股票市场中的最优做市和最优执行问题形成鲜明对比。这主要是由于这两类市场之间的信息结构的差异。在“盲拍”场外交易中，流动性提供者不知道其他流动性提供者向交易者提供的价格。然而，在股票市场中，市场参与者通常知道限价订单簿的确切形状。这些数据也很容易获得，不像流动性提供者在场外市场提供的保密价格。由于场外市场的信息结构不对称，很难找到与微观结构一致的易于处理的模型。本研究有助于文献提供的理论结果，在设置时，信息共享的流动性提供者。虽然这是一个强有力的假设，但结果为与微观结构一致的实验提供了有用的基准。这些实验涉及流动性提供者必须在不确定的环境中妥协探索和利用（例如，运行多臂强盗算法）。总而言之，这项研究通过说明某些交易者行为如何导致对交易者和流动性提供者都不利的情况，应能提高场外市场流动性提供的质量。例如，交易者可能试图与尽可能多的流动性提供者进行交互，但实际上，人们可以争论聚合器中流动性提供者的最佳数量。交易者也可以通过混合不同风险管理风格的流动性提供者来创造一个囚徒困境。虽然这些问题在流动性提供者中是众所周知的，但是具有严格的数学以及广泛的模拟来支持传达给交易者的消息是有价值的。该项目属于EPSRC统计和应用概率研究领域的福尔斯。它是与德意志银行的固定收益和货币量化交易部门合作完成的。

The purpose of this project is to study the interaction between a trader and a given number of liquidity providers in an over-the-counter market. An example of an over-the-counter market is the FX market, which is the largest financial market in the world with a daily volume of several trillion dollars. Unlike the stock market, which is organised into exchanges with transparent limit order books, the FX market hinges on bi-lateral trading with liquidity providers. This is typically done in an aggregator, a technology that consolidates liquidity from several sources in the form of a limit order book.This research aims to address a void in the literature regarding issues specific to liquidity provision in a`blind auction' over-the-counter setting. This topic has received limited attention up to this point, in stark contrast with optimal market making and optimal execution problems in stock markets. This is mostly due to the differences in the information structures between these two types of markets. In a `blind auction 'over-the counter setting, a liquidity provider is unaware of the prices offered to a trader by other liquidity providers. In a stock market, however, market participants are typically aware of the exact shape of the limit order book. This data is also easily accessible, unlike the confidential prices offered by liquidity providers in an over-the-counter market.Due to the asymmetric information structure in over-the counter markets, it is challenging to find tractable models that are consistent with the microstructure. This research contributes to the literature by providing theoretical results in the setting when information is shared across the liquidity providers. While this is a strong assumption, the results provide a useful benchmark for experiments that are consistent with the microstructure. Such experiments involve liquidity providers having to compromise exploration and exploitation in an uncertain environment (for example, running multi-armed bandit algorithms). They shed light on practical aspects of over-the-counter market making such as algorithmic collusion.All in all, this research should improve the quality of liquidity provision in over-the-counter markets by illustrating how certain trader behaviours can lead to situations that are unfavorable for both the traders and the liquidity providers. For example, a trader may attempt to interact with as many liquidity providers as possible, but in fact one can argue about an optimal number of liquidity providers in an aggregator. A trader can also create a prisoner's dilemma setting by mixing liquidity providers with different risk management styles. While these problems are well known among liquidity providers, there is value in having rigorous mathematics as well as extensive simulations to support the messages conveyed to traders.This project falls within the EPSRC Statistics and applied probability research area. It is done in collaboration with the Fixed Income and Currencies Quantitative Trading department at Deutsche Bank AG.