The Gateway to Algorithmic and Automated Trading

Sellside Round Table

Published in Automated Trader Magazine Issue 22 Q3 2011

After Q2’s fascinating discussion of the distinct characteristics of FX algo usage, evaluating the critical differences between FX and other asset classes in this respect, our sellsiders move on to discuss FX algo development and the prospects for FX algo functionality that will accommodate multi-asset strategies.

Joining us at the

Cameron Mouat - Head of foreign exchange algorithmic execution, Deutsche Bank

Gary Stone - Chief Strategy Officer, Bloomberg Tradebook

James Dalton - Director of FX Algorithmic Execution, Citi

Jonathan Wykes - Head of AES FX Sales EMEA, Credit Suisse

Which are the most critical trading venues to which FX execution tools must be connected to be competitive?

Cameron Mouat: There is a core set and they are fairly well known. In addition Deutsche Bank has a large pool of clean liquidity which is independent from other venues.

Gary Stone: We have created our relationships to grow our FX market place's ecosystem. As a result we have created a deep liquid pool and this enables us to be partners with the liquidity providers and liquidity takers so that both sides win. Therefore, we know exactly what we can do in our market place. However, we realise that not everyone in the FX market has the technological ability to operate this way so we have developed technology to include others.

James Dalton: There is a primary liquidity pool for each currency pair. Obviously Commonwealth and EM currencies are traded mainly on Reuters, and Euro Dollar and Dollar Yen are EBS centric. This may not remain the case for ever, as firms like the CME and Currenex are continually eyeballing these enviable brokerage businesses while investing millions in technology to try and build the functionality that will attract clients away. The ground is shifting fast and when the market is busy it is not always the primary markets that see the most volume or the first flurry of activity. We are also yet to clearly understand how pending regulatory decisions may impact where we trade.å Again I would mention that the top tier of electronic market making banks and their internalisation capabilities make them a vital cog in the liquidity landscape.

Jonathan Wykes: It's important to have all the standard primary venues (inter-bank), and ECNs. If you are trading in the majors then you do need to have all of those set up.

One thing that we have done, that has been key, is the use of Cross Finder which is very big in the equity space. It's essentially an internal matching system for all our client activity. We have been actively developing that, taking in liquidity from more local sources and crossing more and more customer to customer flow. This is important because when you start trading illiquid currency pairs there is only so much that you can do in the market. You can obviously get good executions but if you can trade without going to the market as often then you are going to have less impact and less impact means that you can give a much better execution to your client. The key is to have a very diverse pool of liquidity rather than lots of liquidity that really can end up being magnified.

Cameron Mouat

Cameron Mouat

How do the various buyside segments vary in their demand for and use of FX execution tools?

Cameron Mouat: Real money, corporates and larger hedge funds are predominantly interested in anonymity of execution, minimising market impact as well as execution performance. Smaller hedge fund prop desks use algos more for the functionality and execution performance.

Gary Stone: If you take the buyside you have the hedge funds, the asset managers, the corporates and they are all looking for something a little bit different. Hedge funds, as we discussed before, seem to be technical in nature. They tend to use algos that help them build positions or set up entry and exit, trailing stops and setting up event-driven trades that key off different market data. We are seeing more and more asset managers start to dabble in some of the scheduled TWAP or wave algorithms to manage the FX exposure arising from their foreign equity activities more efficiently. A few have started to use average price algorithms which seek a limit which is an average price - a soft limit rather than a hard price limit. Corporates like to do outrights, so we developed technology that enables them to seek best execution on their spot transactions, using similar tools to the asset managers, but automatically roll it forward.

James Dalton: In retail it's a minimum. The order sizes tend to be typically smaller than we have to service in the corporate or the institutional market. That may change as the segment continues to mature. For Real Money Investors & Hedge Funds the take-up depends on the nature of their execution goals.

There has been a notable increase in scrutiny around the way all financial institutions operate in terms of where costs are incurred and how markets behave. The regulatory changes being mooted at present reflect this. Pressure will also come to bear on the buy side, with transaction costs coming under a similar level of focus to what exists in equities. It is only a matter of time before the use of algorithmic strategies becomes the norm, due to the transparency and efficiency they can offer. A directional trader would probably still lean towards using risk pricing but if it's relatively passive flow then the take up in terms of utilisation of algorithms is on the increase.

Jonathan Wykes: I think you can categorise it into two sections. Firstly you have your traditional real money firms that are typically predisposed to using scheduling algorithms like VWAP and TWAP. Their primary concern is impact and they like to work things over a slightly longer period of time. You could also say that about the corporate section, which is one area we have seen quite a lot of demand for that type of algo of late. That is largely due to people trying to improve on best execution and they see VWAP and TWAP as a means of doing that.

The macro and prop guys tend to trade in a very similar fashion and use the more opportunistic strategies. They will be aggressive when they see an opportunity but also they are quite happy to be passive if the market dictates. Strategies that allow them to simultaneously buy on the bid and pay offers are very popular.

Gary Stone

Gary Stone

Which buyside segments are primarily driving the pace of sellside FX execution algo development?

James Dalton: We find that systematic accounts with medium frequency trading requirements have the most active interest in understanding how the different algorithms can be utilised. For many funds the challenge begins when their performance is solid and AUM increases; they find that transaction costs incurred by the increase in size can begin to detract from performance. This is the customer group we expect to see the bulk of customisation requests from.

Gary Stone: More and more asset managers and "real-money" investors are now looking at FX. They are realising that the way they are dealing with their FX exposure is suboptimal and results in severe slippage in performance from their base strategies (which causes the need for an FX transaction in the first place). We have seen some third party transaction cost analytics providers estimate that FX accounts for as much as 90bps of transaction cost - resulting in a direct hit to portfolio performance. Looking at equities, the development of the market structure, increase in electronic trading and use of algorithms resulted in transactions costs declining from a high of 40bps in 2001 to around 17bps.

We are seeing more and more market participants across the spectrum starting to examine whether the management of FX exposure should be moved from the current operations/settlements staff to a trading role and managed from the trading desk. As this evolution occurs, we expect the use of execution algorithms to grow significantly and transactions costs to decline.

Jonathan Wykes: All of them do. When we first started doing this our algorithms were far more basic than they are now. I think two things have happened. The way we look at the market and the data we collect really dictates the types of strategies that we can use, as you get more data you can do more sophisticated things. Instead of just deciding that you are going to peg yourself to the market you can peg yourself to the market and then read the order book. The volume will then dictate how you go back into the market and how you interact with it. You couldn't do that two or three years ago because that data wasn't available. We've been doing it in equities for many years but now you can do it in FX. It wasn't because the technology wasn't there it was just we didn't have the data to enable us to.

I would say that the corporate and real money clients are very much into TCA and benchmarking whereas I think the macro and leverage community just want greater control of their own executions. They are happy with taking risk on board. You tend to find that real money investors are not so familiar with taking risk but have started to realise quite recently that having a bit more control over things actually can give them a better execution.

James Dalton

James Dalton

So it's less of the case of retrospective risk management and more turning it into real time alpha capture opportunity at the same time?

Gary Stone: It is both depending upon why you need to enter into the FX transaction. For some hedge fund investment strategies it would be direct alpha. For market participants (including hedge funds) where FX is a by-product of another investment strategy it is a risk management function that contributes directly to portfolio performance. These participants are now asking themselves whether it is more efficient to leave the FX transaction to their custodian or prime broker to do at the end of the day (where they may not be 100% certain what price they are actually hedging at) or to take greater control over the transaction and trade it as if it was a position with an associated P&L.

How strong is the demand for execution tools that include FX functionality alongside that for other markets to facilitate cross asset trading? Is that demand growing?

James Dalton: Many FX related initiatives in the cross asset space have been based around equity and cross border arbitrage opportunities where the differential, in terms of the net price in your base currency, was locked in via accessing a streaming price. So having an API, generally from a bank, plugged into an EMS that could complete the relevant transactions simultaneously was key.

A lot of people talk about cross asset trading, but I think it's quite hard for some of the larger firms to organise themselves (with some exceptions of course) so they can implement software that meets the requirements of all internal groups. I can see the market moving towards providing wrapped services across equities, FI and Currency in the future but there is a lot of investment that needs to be completed before it becomes a generic or stock offering across the street. It will come, but perhaps not for another couple of years. Demand will continue to be driven by CTA's and Hedge Funds in the interim.

Jonathan Wykes

Jonathan Wykes

Gary Stone: We are seeing tremendous growth in clients leveraging our cross-asset trading (pairs) execution algorithms. For example, if they are trading Vodafone versus AT&T, they will have an FX component that requires hedging out. Although we have the capability to sweep FX intraday when the pair creates "X" amount of FX exposure, we find that many clients are leaving the FX component hedging to the end of the day.

Jonathan Wykes: It's growing but not at a huge pace. We offer the ability to trade FX alongside our equities algorithms. It's very common for clients to argue tooth and nail about the slippage they get in equities but then leave dollars on the table when it comes to their FX. I think people are starting to realise that actually FX is very important especially with increased volatility. You could be paying away quite a lot of money if you are not looking at the cost of your FX hedging. I think that the demand for that will increase but not necessarily linking it into the same ticket; although we can do that.

In terms of cross asset trading it will be more akin to multi asset trading. In other words someone doing their equity and their FX rather than the two linked into the same strategy. We can do that, and we do, but when I go to see real money guys you will have one guy on the desk trading the equities and another just taking care of the FX risk for him, as opposed to the same person using one ticket to do it. I think growth has been slow because people don't necessarily understand how they are going to settle it or how they are going to cope with just one ticket. People are so used to trading in silos that when you try and combine all these things together often they can't handle it operationally.

Execution algos for FX options - already a reality?

James Dalton: While many banks provide their customers with an electronic interface for pricing and dealing options, this does not necessarily lend itself to provision of algorithmic execution capabilities. While standard strikes and tenors may be relatively liquid in the interbank market and banks can already auto-hedge into the broker liquidity pools, the customised nature of the client side of the options business would limit the appeal of automated execution product.