Julie Ros: From your vantage point on the buy side, how do you see the banking industry preparing for Dodd-Frank?
Mike Harris: There seem to be two camps: one includes those banks that are fully preparing, in terms of investment, technology and people, for what they think is coming, and the other consists of those that are taking a gamble – staying on top of things, but not building the systems or allocating human resources until it is absolutely clear what is coming.
JR: If Dodd-Frank were to get watered down, or isn’t what people thought it would be, do you think those first mover banks will be disadvantaged because they devoted time and resources into building something they don’t need, or will it become a competitive advantage because now their technology is that much better?
MH: The banks that are not actively preparing argue that it is a waste of time and money, but I think you can make a case that the gap between the top tier prime brokers may spread even more, because it’s the bigger moneycentre banks that have already begun the central clearing build-out.
JR: How would you characterise the prime brokerage landscape today versus 10 years ago?
MH: In the past, clients were rarely turned away, because banks were hungry to build their businesses so it became a land-grab for any and all clients. Today, banks better understand what their limitations are and are comfortable raising rates for those customers that cause a big strain on their infrastructures.
JR: The cost of prime brokerage has dropped a lot in recent years, while Dodd-Frank is clearly going to cost the banks more. Do you expect them to pass on that increase?
MH: Instead of an outright increase, I think we are more likely to see something in the form of a central counterparty clearing fee.
JR: What’s your view on prime brokers actively courting the high frequency sector?
MH: I wonder if having a lot of high frequency clients doing 20 yards a day, with an average ticket size of $1 million is a good selling point, or if I should be scared to death of partnering with somebody that primes for a lot of HF trading firms. My concern is that, while we are paying the bills by doing a good amount of volume, we certainly aren’t putting up HFT volume numbers. Does this mean that I am further down on the pecking order? Are they the marquee clients now instead of traditional macro hedge funds and CTAs? Do I really want to partner with a PB that is servicing that kind of business? It’s a hard decision because you can make a case from a technical perspective, that because we’re using algorithms that may break a parent order into hundreds, if not thousands, of slices; we are printing a fairly large number of tickets ourselves. In doing that, I need a PB that can handle high ticket volume from a processing perspective. So it’s difficult, because I’m attracted to partners that can handle that community, but also nervous if they have too many of those clients, as we could fall by the wayside.
JR: Do you use multiple PBs?
MH: We had three at one point, but were reduced to two after Royal Bank of Scotland acquired ABN Amro. If our assets grow over time, we will probably look for a third again to further diversify our counterparty risk. We currently use RBS and UBS.
JR: How do you differentiate between PBs?
MH: Prime brokers, although they all try to look the same, have very distinct skill sets. Today almost all of the PBs are plugged into Traiana, but this wasn’t the case a couple of years ago. Banks either decided to partner with Gil Mandelzis and his team or they tried to build out the plumbing on their own. That said, even though everyone seems to have embraced the Traiana model, some banks are still much further along the connectivity curve than others. Now that pricing has become fairly standard, we spend most of our time focusing on the level of relationship management and technology that a PB can provide. Everyone has spent money on various elements of technology – some more than others. Some have focussed on high frequency or high ticket processing; others have focussed on credit line management. You have to look at what’s best for you. In the end, the one thing we learned from our experience with ABN in the early days was that, while they did not have the best technology on the Street, they were one of the best at client relationship management. Tony Dalton, who is one of the guys that helped pioneer the modern Prime Brokerage model, overcame a lack of technology with exceptional customer service.
JR: Are you afraid that relationship management might be overlooked in this technology race?
MH: People forget that in the end, when the technology breaks, someone has to have the skills to quickly assess the scenario, mitigate the risk, fix the problem, do a post-mortem and effectively communicate everything that transpired to the client. Recently, we brought UBS on as a new PB. They had a solid infrastructure, although there were banks with newer technology, but when I looked at UBS, I saw people like Brian McDermott and Ed Pla. These are industry veterans who have been at the bank for years with little to no turnover on their teams. They understand the business, they know their clients well, and are great at customer service. Chances are that five years from now, these guys are still going to be at UBS and Campbell is still going to get the same level of service to which we have always been accustomed.
That said, if they didn’t have the systems in place to facilitate all the algo trading that we’re doing, we couldn’t have partnered with them. So you truly have to have both, you can’t lean too much on one or the other, there’s a balance you have to achieve. We picked UBS because they had the perfect mix of technology and relationship management.
JR: If your key PB relationship managers left, would you follow them? Is that how important the relationship is?
MH: It’s hard because setting up a new PB is not a one day process. Between new legal documentation, credit lines and reconfiguring your trade processing workflow, it can take several months to make the transition. This is exactly why we have multiple PBs. In the event that counterparty risk concerns forced us to quickly leave one PB, we already have an existing partner to accept our positions and facilitate our trading. If our key relationship managers left, we would certainly have to review the relationship, but would also keep in mind that onboarding a brand new PB involves a lot of resources and can’t happen overnight.
JR: Campbell was a leader in getting the buy side better access to the multibank platforms – namely EBS and Reuters. Can you give some background on the environment then, and how it has changed?
MH: Knocking on doors in search of interbank liquidity access – specifically EBS and Reuters – was not an easy task. I give our former head of trading, Doug York, all the credit for breaking down those barriers and apparently Profit & Loss does too since they inducted him into their Hall of Fame a few years ago. Back then, there were real fears of letting the buy side into those pools of bank-only liquidity. Many thought that it would disrupt the market, but we stood by our argument that it would only improve liquidity and tighten spreads. Many interbank participants told us they weren’t afraid of letting us on, the problem was that when they opened that door, other players were going to get access as well, maybe people that didn’t have the integrity that Campbell has – and it was those people that truly scared them.
JR: What’s your view on HFT’s impact on the FX market?
MH: It has been a long time since we talked at a conference about giving the buy side access, as many of our banks joke that we now have more liquidity than they do. Now the conversation has shifted to whether the high frequency community should have unlimited access to our trading destinations. Many of the people I have spoken to at the conference are surprised at the current level of FX liquidity given the current crisis. EBS volumes are off the charts, and other platforms are having their biggest months in several years. But this time around, with the exception of Swissie, we haven’t seen spreads blow out like they did in 2008. I feel that’s the impact of high frequency trading and other buy side liquidity providers making markets at many FX dealing venues. During the Lehman crisis, several key market making banks widened their spreads to “protect themselves” and the rest followed, but did they go too far? I can only speculate that if the banks start to widen their spreads out this time, they will see their trading volume decline significantly as these new liquidity providers stay tight and effectively step in front of their prices to capture the lion’s share of active spread crossing in the e-FX market. For the buy-side, this is a welcome change in the FX market structure as trading costs are a key component of any FX Alpha strategy.
JR: So do you think we’ve reached a turning point?
MH: The market micro structure seems to have completely changed, and I don’t think this is a short term thing – I think this is a real significant change.
JR: What’s your view on the May 2010 flash crash – do you think that could occur in the FX market?
MH: What worries me – and I don’t blame high frequency – is that market makers aren’t there to lose money, so when their models begin to haemorrhage in a volatile market, they reduce liquidity, widen spreads or pull the stream completely. When that occurs, it opens you up to some fairly large volatility and big price swings. I think the FX market is different from the equity markets in lot of ways, so it’s not a direct comparison. Someone made the point on one of the earlier panels that if the regulators do clamp down on high frequency trading, it will probably happen in equities first. If that occurs, HF trading firms, many of whom are already trading multiple asset classes, will begin to shift more of their market making business into FX as it is more difficult to regulate cross border.
JR: At our June conference in New York, Congressman Joe Crowley asked the audience if they thought a flash crash could happen in FX. The overwhelming answer was no. What’s your opinion?
MH: Flash crashes happen in FX all the time, we just don’t call them flash crashes – we call them big moves or volatile days. Think about the move we just saw in euro/Swiss. Imagine if that kind of move had happened in equities, people would have been up in arms! Stocks are a very different asset class because the majority of investors are buy and hold. The vast majority of the equity market has long beta exposure which means that most people lose money when the market drops violently. The media only makes the situation worse by talking about how much value was “erased” from the market on those big days. The FX market by contrast is a zero-sum game with some participants long and others short, so large moves means that somebody wins and somebody loses. For everyone that had a really bad day when the S&P downgraded the US, there were people that had an incredible day. So multi big-figure moves have always been something that was priced in and widely accepted in our markets.
The other major difference is they don’t bust trades in FX. There were a lot of people that made great trades on the day of the flash crash, only to have them busted. In many cases, they got really hurt because they bought it on the dip, then sold it when it rallied only to find out two hours later that, that buy at the bottom was busted, so now they’re short with no hedge and the market had rallied. I think this is one of the biggest themes we hear at conferences like this one. Participants cannot understand why the FX market needs additional regulation when it routinely deals with extreme market volatility and made it through the Lehman crisis without issue.
JR: How important is the FX portion to Campbell’s product mix?
MH: Based on positioning, FX could be as low as 10% or, if currencies are trending and profitable, as much as 40% of our portfolio risk.
JR: The FX market has become very fragmented, how do you determine where to execute?
MH: We still believe, as most of the major banks do, that the portals aren’t all the same. We see a primary and a secondary market in FX. We believe EBS and Reuters are still the primary market. Whenever we’re aggressing, whether DMA or via algo, it’s easy: it’s who has the best price and that includes transaction cost. So if one destination is charging $5 per million to trade and another is charging $2 per million and showing the same offer, obviously we’re going to smart order route to the ECN with the lower fee. But within that, we also write rules in our smart order router to look at whether it’s a primary or secondary platform, because from a market impact standpoint, if I hit a $1m bid at Reuters or EBS, there’s a good chance that it’s a spot desk trader trying to clear $100m and has $99m more to do, so giving him that one buck is not a big deal. We’re always worried that if we hit someone at a secondary platform on that $1 and it’s a market maker with a high frequency algo strategy, first off, we don’t even get a buck, we get $200,000, and magically the market evaporates and goes down 10 ticks. We still see that to this day, so even if we pay a bit more to trade at those primary liquidity sources, from a market impact standpoint, we typically will route to them if they have an equal or better price.
JR: What do you think about the fifth decimal place pricing on EBS?
MH: It was significant because we’ve actually seen our volumes go up on EBS as a result of the fact that the same market makers on some of the secondary platforms were also market making on EBS, but they couldn’t show that fifth place price. Now that they can, we don’t see the secondary ECNs always jumping in front of EBS.
JR: Are you happy with the level of smart order routing available in FX?
MH: Going back to the FX versus equities comparison, I think FX is behind on this important functionality and needs to be expanded over time. For example, we use Morgan Stanley as our equity PB because we feel that they have a strong smart order router. When we deliver an order to them, we don’t tell them where to bid when we need to buy Microsoft. They use real-time trade flow metrics so that they always know the best place for me to bid in that name. I don’t have that same level of confidence in FX, so we work with our banks to help them improve their smart order routing process. We currently have FX algo partnerships with both Credit Suisse and Citi, and are working to onboard JP Morgan. That said, we routinely engage them on why a trade was routed to a particular destination. We want to hear what their rationale was and make sure that it was well thought out – not that an ECN was the first one on the list or that the bank had a lower fee to trade there.
In the absence of real-time data analysis for truly smart order routing, many banks simply embrace the model that we began with years ago. Begin by passively showing your interest at a primary market (EBS or Reuters), and if you have to aggress, you route to the destination with the best price. That said, you have to also consider what to do if the primary market is showing the same price as a secondary ECN, you always route to EBS or Reuters because you know historically you have less market impact there. The latest development seems to be a hybrid model whereby you also introduce the algo-providing bank’s internalised stream into the liquidity pool as a kind of FX dark pool. That way, if the algo goes to cross the spread, it makes more sense to deal with this stream at the same price or better as there should be limited to no market impact and no ECN fees.
JR: What’s your view on the rumoured bank-only platform being considered?
MH: Liquidity in e-FX would certainly suffer if the banks moved to a Pure FX platform, but I believe that the market would still function since there are so many non-bank liquidity providers out there. It would just mean that we would have to trade more with high frequency and other buy-side market makers. The question that remains unanswered for me is, if I am using a bank-provided algo, will they be able to execute my order in that bank-only liquidity pool?
JR: Campbell was one of the first to aggregate back when aggregation was considered a dirty word – how has reception changed?
MH: The market-making banks would get upset if we would touch them in three or four different venues at once. Our response was that since the ECNs were anonymous, we didn’t know who we were ultimately trading with, so we worked with our ECNs to limit duplicate liquidity streams where possible. In addition, the banks were not doing a very good job of limiting their market making risk back then. If their risk tolerance was $50 million, they should have put $10m out at five different venues. Instead, many were putting $50m out at five different ECNs and were angry when someone hit them for $250 million at once.
If anything, turning off some of those streams to show more interest-based dealing might be a better reflection of where the true market is instead of all this phantom liquidity. There’s no doubt that the liquidity mirage is still out there.
JR: Why do you think people still haven’t honed this?
MH: Everyone is playing that arms race: ‘How can I get my price out to as many places as possible and yet still be able to pull it as quickly as I can?’ I think that’s really where high frequency forced the banks to up the ante on their technology, because they were being picked off. So they either had to cut back on what they were putting out there or get a lot faster.
JR: How important is co-location to your business?
MH: We are currently speaking with Portware about potentially co-locating our FX aggregation environment. It isn’t because we’re getting into the high frequency game, it’s because there are times we see prices and go to trade on them via DMA and they’re gone. In many cases it’s simply because there isn’t a fast enough refresh of market data or the order is slow to get to market. I now spend a lot of time with our IT team learning the mechanics of data centres, which is not something in my wildest dreams I would have thought I would have to understand, learn about, or even get excited about.
The other thing that it really cuts down on is the spider web of managed lines. When you have 12 points of liquidity you’re connecting to, that’s $1500-2000/month per ECN to be able to connect to them, only then to have to deal with the latency issue. If I co-locate a box inside the NY4 data centre, and I’ve got eight of 10 major ECNs with either a point of presence (POP) or their matching engine itself there, I get rid of the latency game and now I pay $100 for a cross connect because I can simply run a 100 foot cable from one box to another instead of running that managed line from New York to Baltimore. Sure, there are some expenses you have to go through to get co-located, but as more and more people are doing it and data centres are getting bigger, the costs are certainly coming down.
JR: Your office is in Maryland, close to that August earthquake in Virginia. Did it impact your business?
MH: That earthquake was a game changer for everyone on the East Coast. I think it caused a lot of people to rethink their backup facilities. Many of us have disaster recovery sites within close proximity of our primary locations so that we can get there quickly in the event of a problem. So if the earthquake had been bad enough, it could have rendered both facilities inoperable. Now, because of cloud computing and proximity hosting, many industry participants want to have a critical trading infrastructure sitting in another city so that all they have to do is connect to it and continue trading.
Another interesting thing about data centres is that a lot of the banks, particularly those facilitating algo trading, are finding they have to support three separate data centres in New York, London and Asia. FX is a continuously traded product. If I place a trade with you in Asia, what happens when it closes? Does my trade get passed from one box to another? I think a lot of people are realising the need for one central data centre with servers to support 24-hour trading, with backups in other centres. I think it’s something everybody’s dealing with across asset classes, but FX is leading the way because it truly is a 24-hour market. As futures, equities and other asset classes become more and more global, people will ask: ‘If the boxes are self-sustaining, why not just leave them on so that we can capture more volume during the overnight session?’
JR: A number of banks have rolled out algos, what is your preference between the agency and principal model?
MH: There is certainly a big divide amongst the banks on this topic. In the agency model, banks will charge you a fee to facilitate your trades across all the ECNs, while potentially improving your fill and impact by also executing with its own internal flow and other PB algo clients. In the principal model, you don’t pay a fee upfront, but it doesn’t enable you to execute passively and you have to exclusively trade against the bank’s stream as there is no access to ECNs in this model. Jeff Becks and I have worked for years to try and keep us as anonymous in the markets as possible, so we were not about to use an algo that would merely provide a fully transparent axe for the bank’s spot and prop desks throughout the day. To me, that takes the best execution model and chucks it out the window. Who’s to say that the bank’s price doesn’t uptick or widen out just as the algo goes to cross the spread. Some clients have a hard time stomaching the upfront costs of the agency model, but I think they are missing the bigger picture by not looking at the total cost of execution. Consider an order to buy in a 20-22 market, if an agency algo gets you done on the bid at 20 and charges you a pip for use of the algo, your all-in is still better at 21 then if you had simply crossed the spread and paid 22.
We’ve embraced the agency model because it mimics what we’ve always done in the futures and equity markets. From our standpoint, it’s yet another way that we can effectively outsource trading technology so that our developers can focus on our Alpha strategies and not just maintaining our own FX algo code. Another benefit of this model is the economies of scale at work to improve the bank’s algos. The bank gets my feedback on their algos, as well as input from 50 other clients, so the product is always getting better. The key is to get in early enough to truly form a partnership with the banks. They expect candid feedback on their performance and with our prior experience building algos, we give lots of it.
JR: That model has worked for you for a long time – going all the way back to embracing OTC FX before many of your competitors, helping to build out PB, being one of the first to aggregate and now, co-location. What else is on your radar?
MH: We are partnering with Traiana to help develop a buy side version of their bank credit line monitoring product. We were one of the first clients from the buy side to Beta test the product and give our two cents. The buy side is just as interested in line usage as the prime broker who is monitoring our trading activity. I love their concept of a “kill switch”, because in the past I would have had to make 12 calls to get liquidity turned off at all of our ECNs, so the idea of flicking a switch at Traiana and instantly killing my credit to all these venues simultaneously is a strong step towards better risk management.
JR: What keeps you up at night?
MH: My primary goal is always to make money for our clients and that is no easy feat in these challenging markets. Beyond that, my two largest concerns are counterparty risk and regulatory risk. What banks are safe to trade with? Which firms have Greek exposure? I joke with clients that sometimes I feel like I watch my counterparties just as much as the markets. Everyone wants to know who the next Lehman Brothers or Bear Stearns will be, so you can make sure you are not executing with a bank that might potentially be in trouble. We actively monitor equity price, credit default swaps, news headlines, government support, credit ratings and market chatter. That said, we don’t ever want to make a situation worse for a particular bank, so if we have made a decision to temporarily suspend trading with one of our counterparties because of a potential risk factor in the short term, we don’t communicate that outside the firm to anyone – not even the counterparty itself.
With regards to regulatory risk, we are closely monitoring all new developments when it comes to the impact of the US Dodd-Frank Act on the FX markets. One of the things the market isn’t fully pricing in is that, even though we have had a recommendation that spot and forwards should not be included in Dodd-Frank, there is nothing in writing at this point that guarantees they will be 100% excluded. The buy side is very much in wait-and-see mode. Everyone is watching each other and the sell side to get a feel for what people think the regulatory environment is going to look like.
Since the FX market has been largely voice-driven up until the past few years, many FX trading systems are designed around the flexibility that relationship-based trading has afforded us. Regulation in any form, particularly if you’re talking about doing business through a swap execution facility (SEF), is going to mean people’s strategies are going to have to change. Doing an electronic request-for-quote at a SEF will not be as customisable as talking to a salesperson at a bank, so market participants will all have to conform to the same market standards. I think many people are evaluating what it is going to take to deliver a trade to a SEF, execute it, receive trade details back into your order-management system, get it to the central counterparty for clearing and to notify the trade repository of the deal. That in itself, is enough to keep you up at night!
Mike Harris is director of trading at Baltimore, MD-based Campbell & Company, Inc. Harris joined the firm in 2000 after working as a futures and options broker for Refco in New York. Prior to that, he worked in the sales and product development groups at Morgan Stanley Managed Futures. He began his career at Campbell executing futures and currency orders during its European shift for two years before moving to the North American shift as a specialist in foreign exchange.
In 2004, he became deputy manager of trading and was charged with overseeing the day-to-day operations of the Asia, Europe and North American trading shifts. In June 2006, he was named director of trading and currently manages global execution for all of the asset classes that Campbell & Company trades. He sits on the firm’s Executive, Investment, Risk and Best Execution Committees.
He attended Gettysburg College in Pennsylvania and received a B.A. with honours in Economics and Japanese Studies. He also spent time studying abroad at Kansai Gaidai University in Osaka, Japan.
Founded in 1972, Campbell & Company is one of the oldest and largest alternative investment management firms in the world. Campbell and its affiliates manage global futures, currencies and equities portfolios using systematic absolute return investment strategies.