Market Surveillance, Liquidity Shocks and Regulation
Thu, 02 Dec 2010 17:38:00 GMT
An Interview with Natan Tiefenbrun
In this interview for the High Frequency Trading Review, Mike O’Hara talks to Natan Tiefenbrun, Commercial Director of Turquoise, the pan-European MTF.
Natan has extensive expertise in the business and technology of electronic trading platforms and algorithmic trading, having spent 12 years with Instinet, managing the firm’s algorithmic trading, dark pool equity crossing, portfolio trading, execution management and transaction analytics products and services. He joined Baikal at the London Stock Exchange Group in January 2009, and was appointed Commercial Director at Turquoise after its transaction with LSEG closed in February 2010.
HFTR: Natan, maybe we can start with you telling us about your own background.
Natan: Certainly. I have been in the electronic trading arena since 1993. I spent most of my career at Instinet, an agency electronic equity broker, which was also the first ECN (Electronic Communications Network) in the United States. I spent time with Instinet both in London and in New York running TCA (Transaction Cost Analysis) and portfolio trading.
I was also responsible for quantitative customers globally, who ran from passive indexing at one end of the spectrum through to active quant through to statistical arbitrage through to electronic market making and other high frequency trading (HFT) strategies.
In doing that I developed an EMS (Execution Management System) for Instinet, which was used both by the firm and by the institutional hedge fund customers for single stock and portfolio trading. Then we extended that platform to provide algorithmic trading.
After Instinet I spent a couple of years doing other things in the telecoms industry and IP Telephony, before returning to the financial markets to work within the London Stock Exchange on their dark pool business. Ultimately the LSE acquired a majority stake in Turquoise and so when that transaction happened I took on the role of Commercial Director, responsible for product, sales, marketing, etcetera.
HFTR: So, now you are an MTF operator, let me ask you about one of the issues that seems to come up again and again at industry conferences and events (and I think you’ve addressed the topic on your blog too). How much does an exchange or an MTF operator need to know about its members’ end-client activity, how much does it actually know, is it enough and is there a need to have a greater transparency in that area?
Natan: As an MTF, our responsibility is to run an orderly market and from a surveillance standpoint that means spotting any behaviour which we believe, either in the context of our own market or more broadly across markets, is suspect. It could be a form of manipulation or it could be contravening any of our own rules or any of the regulator’s rules about market conduct.
In terms of what we know, we know the member firm that is submitting orders and we know down to the session level (the FIX session or the native session), where those orders are submitted from. Some of the time that session is tied to a specific end client or a specific trading strategy. For example if we have co-located DMA, often we know that for a certain strategy, a certain end customer is using a particular session. Or if we know that within a firm there’s a mixture of agency and proprietary business, we might know that a particular user relates to a proprietary trading group within the bank.
But for the client business, excluding sponsored access type arrangements, we typically have no way of differentiating between the different end customers of our members or indeed often between the different trading strategies of our members.
HFTR: Doesn’t that make market surveillance difficult?
Natan: I guess it can make some aspects of surveillance difficult. For example, if we see a series of orders that our surveillance system has highlighted in one of its alerts, we wouldn’t necessarily know if those underlying orders are attributable to the same end customer, which means our surveillance system can generate false positives.
A lot of our contact with member firms on the surveillance side therefore is asking them for additional data, to establish whether something that the surveillance system has identified as a potential problem is all attributable to a single underlying customer (and therefore quite possibly problematic) or whether it’s just a confluence of different end customers.
More broadly I think the question the regulators are increasingly asking is how to perform surveillance on not just one venue, but across venues?
HFTR: That was going to be my next question, so I’ll let you address that.
If you are not a registered user, click here to register for free