Execution in an aggregator

Abstract

An aggregator is a technology that consolidates liquidity—in the form of bid and ask prices and amounts—from multiple sources into a single unified order book to facilitate ‘best-price’ execution. It is widely used by traders in financial markets, particularly those in the globally fragmented spot currency market. In this paper, I study the properties of execution in an aggregator where multiple liquidity providers (LPs) compete for a trader’s uninformed flow. There are two main contributions. Firstly, I formulate a model for the liquidity dynamics and contract formation process, and use this to characterize key trading metrics such as the observed inside spread in the aggregator, the reject rate due to the so-called ‘last-look’ trade acceptance process, the effective spread that the trader pays, as well as the market share and gross revenues of the LPs. An important observation here is that aggregation induces adverse selection where the LP that receives the trader’s deal request will suffer from the ‘Winner’s curse’, and this effect grows stronger when the trader increases the number of participants in the aggregator. To defend against this, the model allows LPs to adjust the nominal spread they charge or alter the trade acceptance criteria. This interplay is a key determinant of transaction costs. Secondly, I analyse the properties of different execution styles. I show that when the trader splits her order across multiple LPs, a single provider that has quick market access and for whom it is relatively expensive to internalize risk can effectively force all other providers to join her in externalizing the trader’s flow thereby maximizing the market impact and aggregate hedging costs. It is therefore not only the number, but also the type of LP and execution style adopted by the trader that determines transaction costs.

Keywords:

• Aggregation
• Liquidity access
• Transaction costs
• Adverse selection
• Prisoner’s dilemma
• Market impact

Acknowledgements

Oomen would like to thank two anonymous referees, Natalia Fabra, Alex Gerko, Søren Johansen, Anthony Neuberger, Mark Podolskij, colleagues at Deutsche Bank, and the seminar participants at the London School of Economics, Erasmus University Rotterdam, and the ‘Microstructure of Foreign Exchange Markets’ conference at the Cambridge-INET Institute for helpful comments.

Disclaimer

The views and opinions rendered in this paper reflect the author’s personal views about the subject and do not necessarily represent the views of Deutsche Bank AG or any part thereof. This article is necessarily general and is not intended to be comprehensive, nor does it constitute legal or financial advice in relation to any particular situation.

Disclosure statement

Roel Oomen is employed as the global co-head of electronic FX spot trading at Deutsche Bank AG.

Notes

1 In independent and concurrent work, Cartea and Jaimungal (2015) study how venue specific last look requirements influence the choice of where latency arbitrageurs operate.

2 The illustrations throughout the paper require a choice of specific model parameter values. Because statistical inference is beyond the scope of this paper, I normalize on and set on the basis that the high-frequency data literature estimates the so-called ‘noise ratio’, i.e. in the set-up here, to be around 0.5 for a range of liquid currencies and US equities (see, e.g. Christensen et al. 2014, table 3).In standard market microstructure models, the spread s typically compensates the market maker for providing immediacy in a risky asset (with risk measured by ) and/or to protect against adverse selection due to information asymmetry or mis-pricing (magnitude of this is measured by ). It therefore seems reasonable to set the spread to a (small) multiple of the or . For the parameters and , there is little guidance available so I set them to ad hoc values: in a range (0.5, 0.75) and in a range (0.5, 0.9) depending on the specific illustration.

3 As an aside, note that the chart also includes the pre-deal mid-price evolution (adjusted for trade direction) which shows an increasing aggression of the price in the run-up to winning a deal request. This pattern is specific to the set-up here, and will look very different if for instance the trader’s buy or sell decisions are triggered by the (true) price reaching specific levels, e.g. stop-loss or take-profit orders.

4 In practice, one may find the sensitivity of the reject rate to changes in to be limited because (i) empirically and s tend to move in tandem and (ii) whilst in the model here, a larger s does affords the LP with more room to loosen the tolerance level . Alternatively, it is of course possible to specify a trade acceptance rule where is an increasing function of or s.

5 For example, see http://www.iextrading.com/insight/stats/, for monthly statistics on the fill ratios of the IEX SOR. For July 2015, it ranges from 69% for orders routed to CHI-X, to 88% for NYSE, to 99% for BATS.

6 The use of internalization as a risk management methodology is highlighted in Bank of England et al. (2014, p. 59):

This has led to an increase in ‘internalization’ in the spot FX markets where banks are able to match off client orders internally without having to go to the inter-dealer market to hedge their risk. Market participants have indicated that some dealers with large enough market share can now internalize up to 90% of their client orders in major currency pairs.

In contrast, an example of a business model centred around externalization is Virtu’s: ‘Our strategies are also designed to lock in returns through precise and nearly instantaneous hedging, as we seek to eliminate the price risk in any positions held’. (see Virtu Financial, Inc. 2014, p. 2). It is important to point out that the internalization/externalization classification is not simply one of bank vs. non-bank LPs because there are banks that externalize significant portions of their flow and funds that actively internalize. In practice, the hedging approach adopted by any LP will lie somewhere along the spectrum from pure externalizer to pure internalizer, and may also vary by—for instance—market conditions and flow characteristics.

7 Externalization—as defined here—involves instantaneous hedging. Of course it is also possible to externalize via gradual hedging in a way that minimizes market impact and makes it observationally indistinguishable from internalization. Consequently, LPs that follow such a strategy should be considered internalizers in the context of this paper.