The Gateway to Algorithmic and Automated Trading

Algorithmic Trading - Fast Forward into the Multi Asset Future

Published in Automated Trader Magazine Issue 03 October 2006

A new evolution of synthetic trading is starting to generate interest multi asset class algorithmic trading. If there is a need to spread trading activity across a wider asset base, why not let the machine help finesse some of the component trades? David Rutter, CEO of Brokertec and deputy CEO of ICAP Electronics Broking examines the issues from a market provider's perspective

Integrating multiple assets visually on the trader's desktop has been a popular industry theme for some time. A variety of system vendors with an established pedigree in one asset class have extended their coverage and technology to accommodate other classes. Now however, we are seeing greater interest in this multi asset concept at a different level - the application programming interface (API). How many markets can you access in tandem through their APIs? Is there a single market that carries a broad range of assets and instruments accessible through a single API? Is there enough depth and variety in those instruments to throw up a worthwhile number of cross asset/synthetic trading opportunities? And are there algorithms available capable of finessing the placement of trades across those asset classes?
David Rutter, CEO of Brokertec and deputy CEO of ICAP Electronics Broking

David Rutter

At present, the answer to at least some of those questions is probably still "no", but that may not remain the case for much longer, with a number of markets already looking to expand their asset coverage and associated trading technology. However, markets and trading platforms wishing to do this have to lay some foundations first if they are to succeed in their expansion and capture a meaningful slice of cross asset activity.

API consistency

If a market place is offering electronic access for automated/algorithmic trading of multiple asset classes, it needs to do so in a consistent manner. For practical reasons, such as where two markets have merged, it may not initially be possible to offer all trading through a single API. If this is the case then at least support for those APIs must be as homogenous as possible; users will not want to ring three different support numbers for three different asset classes.
In the longer term, one API for all asset classes is the ideal, but there must be consistency within that API. Having completely different ways of performing the same function for different asset classes in unlikely to be popular with users.

Data, data, data

The platform that wants to coax automated and algorithmic traders aboard has to be able to offer them bullet proof data solutions. Obviously, given the historical testing such quantitative traders need to perform, this includes historical as well as real time data.
Data cleanliness is next to godliness, but additional depth and enrichment is also a competitive issue. Automated and algorithmic traders are increasingly looking to markets for additional information, such as historical depth of market data and machine readable news archives.
An important element when supplying historical data across multiple asset classes is consistency. If the asset classes were originally traded separately, has the data been gathered in the same manner? Every last tick of data might have been collected for one asset, while perhaps only one second samples for another. Ideally the provider needs to do some form of normalisation of such data, so that customers can have confidence in its synchronicity when modelling across multiple assets. Without this, they cannot have sufficient confidence in the reproducibility and accuracy of their historical simulations.

Back office

Offering a single API to multiple asset classes at the front end is only part of the story. An environment where a single transaction may be broken into perhaps even hundreds of trades across multiple assets has back office implications as well. Therefore whatever a market place offers at the front end it must be able to support at the back. Having a homogenous STP environment that integrates all asset classes and streamlines operations can therefore prove a crucial building block in attracting liquidity.

Orderly across assets

Traders prefer orderly markets, where they can have a degree of certainty as to their quality of their order fills. With multi asset trades, it becomes even more important that a market trades in an orderly manner across all assets.
It is of little use if Assets A and B are orderly, but Asset C is prone to manipulation or other dysfunction. Building single asset models to deal with such anomalies is hard enough, but in a multi asset trade the resultant uncertainties around legging risk and slippage make it a near impossibility, and will almost certainly render the original synthetic trading strategy non-viable. To stop this happening, the market provider has to develop and enforce a consistent set of rules for acceptable market practice across all assets.

"It is of little use if Assets A and B are orderly, but Asset C is prone to manipulation or other dysfunction."


Meeting all these competitive requirements will be expensive, which begs the question of how great the potential demand for multi asset algorithmic trading really is. There are encouraging signs. In the interdealer space we are already observing some interest in adding fixed income to the algorithmic mix.
At the same time, some dealers are noting nascent client demand for this type of trading and those dealers will need venues in which they can efficiently lay off the risks of making synthetic markets to those clients. It may not happen today or tomorrow, but the first glimmerings of a multi asset algorithmic future are definitely on the horizon.