The Gateway to Algorithmic and Automated Trading

MTS: Continuity Begets Functionality

Published in Automated Trader Magazine Issue 19 Q4 2010

One of the questions Automated Trader readers most frequently ask is when they will be able to trade cash bonds on an automated basis. Interestingly, the majority of these enquiries come not from those interested in high frequency trading, but from those looking to automate existing spread or basis strategies at lower frequencies. The first step in this process is obviously an electronic bond trading platform, so we packed off Automated Trader’s founder, Andy Webb, to talk to Jack Jeffery, Chief Executive, and Fabrizio Testa, Head of Product Development at MTS to find out about their electronic evolution and to get their general views on possible future developments in automated bond trading.

Andy Webb:Jack, can we start with a bit of history? How did MTS come into existence and why?

Jack Jeffery:MTS was established in 1988 as it had become very apparent how critical the secondary market was to the effective redistribution of government debt, so the immediate objective was to encourage the improvement of secondary market liquidity in as transparent manner as possible. It was founded as a completely electronic trading utility from day one, through a collaboration between the Italian authorities and the banks.

Andy Webb:And then what happened?

Jack Jeffery:Over the years MTS evolved to where it is today - a truly commercial entity; 60% owned by the London Stock Exchange and 40% owned by the shareholder banks. That commercialisation theme already represents a significant shift from its original utility role but it is still very much ongoing.

However, in the early days the utility model was very valuable in terms of setting the right foundations. In the Italian market, the regulators played an important role in the evolution of an orderly market by ensuring that people provided the right sort of liquidity and support. That business model was effectively replicated across the rest of Europe, so MTS created the same utility structure in partnership with the banks in pretty much every country of consequence in Europe.

In addition to Euro-denominated bonds, in Denmark and Israel we also have two non-Euro countries represented on the platform. We have also listed Euro accession countries on the platform in the run up to their adoption of the common currency.

Fabrizio Testa & Jack Jeffery

Andy Webb:At present there seems to be growing electronic competition for MTS?

Jack Jeffery:Yes, we have seen the sovereign bond market in general expanding and opening up to multiple platforms across Europe. This additional electronic competition is obviously a challenge for us, but one that may in fact help us by growing the overall size of the liquidity pie - both organically but also in terms of volume taken from voice broking. We can already see a strong trend developing in Europe towards electronic trading of cash bonds. This is being driven by a variety of factors, including an increasing desire for transparency, cost reduction, the authorities wanting to improve liquidity and also the increased issuance of debt across continental Europe.

A further driver more recently has been counterparty risk. When conditions are as highly volatile as they have occasionally been during the last two years, the central counterparty clearing available on certain electronic platforms such as MTS obviously neutralises the risk of counterparty default.

Andy Webb:So what products does MTS offer?

Jack Jeffery:They fall into three main groups:

1. The interdealer market in cash bonds, which brings together more than 150 customers, thus allowing us to provide liquidity across most European countries' sovereign cash bonds.

2. We also have repo markets on the platform, which started with Italy, that we are now expanding across other European markets.

3. BondVision, our dealer to buyside customer platform. We see this as one of our greatest growth opportunities; it is certainly the largest growing area of activity from the banks' perspective in terms of fixed-income activity. Our strategy here is to focus mainly on the top 40 asset managers across Europe.

Mid and Striker prices



The MTS Cash MidPrice order book is a new order book available on MTS where the size and side of the quotes is not visible pre-trade. However, the executable price is always available, as the MidPrice is calculated from the primary MTS order book as follows: bid + ((ask-bid)*50%).

The reference bid/ask prices used for the MidPrice calculation are the current MTS Best Bid and MTS Best Offer. These reference prices both comply with anti-gaming rules, including:
• Minimum levels of depth in bid and in offer
• Maximum bid/offer spread of reference prices
• Best Bid and Best Offer prices can only be used in calculating the MidPrice if they have been in the MTS order book for more than a minimum time period.

Orders in the MidPrice order book can only match at the MidPrice and there is no interaction with the primary MTS order book. MidPrice crosses occur continuously as soon as two Fill-and-Store orders (or a Fill-and-Store order and a Fill-or-Kill order) with opposite side and compatible size enter the MidPrice order book. Any orders or trades in the MidPrice order book are tagged as such on the system. MTS Cash MidPrice functionality is currently available to market makers in German bonds, but (subject to issuer and regulatory approval) will be available for other markets in Q4 2010.

striker price

Striker Price

The Striker Price (or striker yield) defines an additional parameter for single side quotes, so that they effectively function as a limit order linked to a single side proposal that will automatically hit/lift visible prices within its range. For example, a bid price of 104.5 with a 104.6 Striker price will trigger an order and lift a visible offer up to the limit of 104.6. The Striker price is only valid with single side quotes and can be defined as either an absolute value or a range.

Various other businesses spin off from these core areas. Our platform allows us to capture and distribute very accurate and granular data, both historically and in real time. This is likely to prove an important growth area in terms of both product range and sales. It obviously has future application to automated trading in the directional or basis/spread trading sense (as opposed to the market-making sense, which already exists on the platform).

Although currently in its infancy, another business area that is growing fast for us is indices. We recently launched the EuroMTS Investment Grade Broad Index family, composed of euro-denominated bonds issued by Eurozone sovereign issuers that have an investment grade credit rating from at least two major ratings agencies and expect to launch further indices over the coming year.

Andy Webb:Fabrizio - how has MTS's technology evolved to match the development of the business?

Fabrizio Testa:The technology milestones are very much in line with the evolution of the business. We have a dedicated technology team that works on new product developments based on client needs. Our CTO, Fabrizio Cazzulini has been with MTS since 2002 and has lived through all the key technology innovations we have brought to the market. The team also provides design specifications and guidance on the architecture of our systems and is supported by our long term technology partner SIA-SSB Group, which MTS has used since its early years. SIA-SSB are also present in areas like payment systems, payment card processing and network services in and outside Italy and used to be the provider of other secondary market initiatives such as the deposit market and the Italian stock exchange.

The privatisation of MTS in 1997 saw us ramp up technology considerably in order to ensure sufficient capacity for our expansion into European markets outside Italy. So in response to the evolution of the market and customer requirements, we launched a series of new products and functionalities. In 2001 we went live with a new technology platform to support the introduction of the BondVision market. At the same time we also introduced request for competitive quote capability along with other functionality, such as the ability to trade more than 1 leg at a time, e.g. switches and butterflies. In 2006 we revamped our hardware platform and introduced a new faster network protocol, which facilitated major progress in terms of overall system performance. Finally, we have recently released to our customers significant new market functionality in the form of Striker and MidPrice (see box "Mid and Striker prices").

Andy Webb:Can we talk a little about your technology infrastructure. Can you tell us a bit about your matching engine and network technology?

Fabrizio Testa:Our matching engines are located in Milan, with one platform servicing all three of the markets that Jack talked about earlier- interdealer, repo and BondVision. In the past we used Tandem technology, but now run the matching engine scaled across multiple physical machines, which are HP ProLiant DL 580 G6 servers running Red Hat Enterprise Linux.

We have an open network architecture, so participants can use an approved network provider of their choice to connect to Milan. On the network we have been using a native protocol since 2006 called SDP, which is an evolution of the previous OMI protocol. SDP has been continually upgraded and is used by the central matching system and all Italian clients; we have consistently found it to be appreciably faster than any other protocols we have considered to date.

For reasons of client convenience - especially buyside participants using portfolio and execution management systems (PMS and EMS) - we have now complemented SDP by also offering FIX protocol connectivity. This allows these participants to link their PMS or EMS to our front end from where they directly execute their business. The FIX link obviously provides them with straight through processing, including allocation to subaccounts when they execute block trades.

Andy Webb:When would you say that MTS became particularly aware of latency?

Fabrizio Testa:It's always been one of our priorities and has always been driven by client activity and appetite for speed on the platform. We have to be able to handle the high number of different bonds that each institution trades on the system, especially those global participants who price all Eurozone countries plus the non-Euro countries. They may have as many as 500 or more bonds priced at the same time on the system as two-way prices, so you need to have a pretty robust system to support that sort of activity.

From 2001, when we started adding secondary markets to the same platform, people became far more aware of latency and were keen to be able to handle automated quoting of a large number of instruments as quickly as possible. We also realised that as we rolled out to new markets this would further increase the quotation traffic, which had obvious implications for networking and the central system round trip time.

This was one of the factors behind our switch from Tandem in 2006, a change which has contributed to our reduction of round trip times from 15 milliseconds then to under 1 millisecond today (see Figure 1). In addition, the recent addition of Striker and MidPrice involved significant changes to the central system, including the creation of a new order book. Therefore a key objective for us was to ensure that the additional load imposed by these changes did not prejudice the latency reductions we had already made - an objective that I'm happy to say we achieved.

figure 1

Figure 1

Andy Webb:Any other items in the pipeline relating to latency?

Fabrizio Testa:Colocation is one item we are working on. Clients connecting from London incur a latency penalty of somewhere between 25 and 30 milliseconds, so for those non-local clients wishing to take full advantage of the 1 millisecond central system matching time, colocation is the obvious solution.

We will be working with specialist colocation providers to offer this. The new colocation facilities will either be in the same facility as the central MTS system or in an adjacent building, so they offer the shortest possible hop to the matching engine.

Andy Webb:It's relatively common to see trading venues dealing with multiple technology providers over time, yet MTS has stuck with SIA-SSB since it was created. What were the reasons for that?

Fabrizio Testa:A number of factors are involved. SIA-SSB has made a large investment in research and development in order to retain us as a client. They have also attained a high level of expertise in this field and have achieved a very good relationship with our clients, with whom they have direct contact when providing them with the API and associated services.

However, their re-appointment is by no means automatic. Whenever we have renewed their contract we have also taken the opportunity to check whether other providers could have matched or exceeded what they offer. On every occasion we have come to the conclusion that retaining the existing provider and ensuring they invested was more efficient than transferring the business to a new provider.

Andy Webb:Presumably a long term vendor relationship also has advantages in terms of time to market for new enhancements?

Fabrizio Testa:The long-term relationship certainly gives you an advantage in terms of time to market, because you don't have to wait for the new provider to get up to speed on your technology and business. The existing provider can probably do several system upgrades in the time a new provider would take to fully understand how everything worked. This was an important consideration for us, as time to market is definitely a competitive issue; you simply can't afford to lose a couple of years in the process of changing provider.

The complexity of the functionality now supported by the system is another consideration. You may find a provider who is very competent as regards generic matching engine technology, but each market that we operate is extremely complex in terms of pre-trade, trade and post-trade functionality.

As mentioned earlier we also benefit from having an in house technology team that has an established skill base with the existing technology. When it comes to future planning, they are pivotal in maintaining the close relationship with the technology partner. The cash bond market is moving very quickly at present as new electronic venues appear, so the need to innovate and do so in a timely manner is pressing. We have just released a number of new technologies but it would be very disappointing if we looked at our roadmap for technology development in two or three months time and it was empty. In the longer term one also needs to be thinking about the way the market is developing. At present, some of the more obvious developments for us to be aware of are the introduction of APIs and the extension of automated trading. In both these cases we feel that we can tick the boxes in terms of preparation.

However, notwithstanding the need to look ahead and innovate, your innovation absolutely cannot be allowed to jeopardise the stability of your existing offering. For example, MidPrice and Striker were important innovations for the market but they also involved major changes to our central systems. Those types of change have to be thought through extremely carefully, otherwise you can end up being innovative but unreliable. If you have a long-established relationship with a vendor who has a deep understanding of your technology and business, then that helps to minimise the chances of falling into this trap. On that note, we were gratified to see that in a recent customer satisfaction survey technology and platform stability were the areas where we scored most highly.

Andy Webb:How do you measure the 'quality' of the market?

Fabrizio Testa:The bid offer spread and the depth of the market are the two variables that we track on a continuous basis and they provide a straightforward way for checking the quality of prices. The underlying prices are of course monitored according to an acceptance range and most of the time the quality of the prices is close to 100% (see Figure 2). However, events like the Greek crisis will for a limited period have an impact on triple-A bonds, while for other less highly rated bonds the impact on price quality will be more prolonged and widespread. For example, Greece obviously had a knock-on effect on Ireland and Portugal and to some extent Spain, which in turn had an impact on the bid/offer spread available on the system.

Andy Webb:Will the buyside only ever have access to BondVision and not the dealer platform?

Jack Jeffery:Yes they will only have access to BondVision, but in the future BondVision may take various forms in terms of the markets we provide. We have to make sure we don't close any doors and always remain aware of the potential for that model to change. Whether we are the ones driving that change is another matter; things may just change of their own accord, as they have in other markets.

Figure 2: EuroMTS AAA Government 11:00 CET Quotation Level (%)

Figure 2: EuroMTS AAA Government 11:00 CET Quotation Level (%)

We will obviously develop markets for the buyside that we hope will fit their needs. However, the reality one has to bear in mind is that fixed income markets in Europe are very different from, for example, the US. The US Treasury market is far simpler in structure than the hugely diverse universe of bonds in Europe. By the same token, it has entirely different liquidity distribution; at present no European bond can match the liquidity available in on the run US Treasuries. But that may change over time when we start to see European markets unlock and as more trading activity is conducted electronically.

Andy Webb:What about automated trading?

Jack Jeffery:A lot of the work we have been doing around latency and data granularity/quality would obviously support that, but frankly right now the much bigger priority for us is to move as much cash bond liquidity as possible into the electronic environment and away from voice. While the market is definitely moving in that direction anyway, there are still wide disparities between markets in terms of electronic adoption. While countries such as Italy have embraced electronic trading of cash bonds, one of our major objectives is to make that level of electronic adoption more consistent across Europe. That's still very much a work in progress; while we have seen our volumes increase by above or near 100% year on year in France, Germany, Spain and the Netherlands, in many cases that has been from a relatively low base.

It's only once you have a critical electronic mass of highly liquid sovereign cash bond trading across Europe that I think the environment will really be suitable for automated trading. And though many people seem to see the terms as synonymous, by "automated trading" I most definitely don't mean high frequency trading. In the short to medium term, strategies such as high frequency inter-market arbitrage simply won't be feasible because of the lack of alternative electronic venues with sufficient liquidity in the same issues. The cash bond market is very different from the highly condensed markets (such as major FX pairs and certain futures contracts) where high frequency trading is currently commonplace,. By contrast, cash bonds are far more disparate in nature.

Andy Webb:Which category of asset manager are you currently trying to engage with BondVision?

Jack Jeffery:Our primary focus is currently on the top traditional asset managers, but we certainly aren't turning customers away from other segments. We want to be able to provide the top traditional asset managers with a specialised niche service that fulfils their needs.

That has a beneficial knock on effect because as these buyside clients become engaged, the sellside observes the quality of the liquidity they generate. The average size of our deals involving buyside participants has doubled in the last year, which is obviously attractive to the sellside in terms of the information and size of transaction, which they will readily compete for.

Some alternative asset managers have expressed serious interest in MTS, but at present the BondVision platform is only appropriate for a subset of them. For example, BondVision uses a request for quote model, which wouldn't fit well with many quant funds. However, we have received enquiries from relative value hedge funds who take curve and inter-country positions for whom it is a suitable trading mechanism.

The MTS Milan technology team

The MTS Milan technology team

Andy Webb:But surely the infrastructure enhancements you have made will have appeal for other types of hedge funds as well?

Jack Jeffery:That's true, but in fact a lot of the improvements around data, trading tools and reducing latency have been driven by interest from existing bank customers, who are increasingly inclined towards trading strategies that benefit from those capabilities.

Certainly some existing customers are very interested in having enhanced automated trading functionality, so we are working on that, which would obviously also assist us in dealing with future market evolution. However, to some extent, our future direction in this area depends upon how client banks deploy the functionality with their own clients.

Andy Webb:for what segment of the fixed income market do you currently see strongest demand at present?

Fabrizio Testa:Sovereign bonds are definitely very active, particularly the short end of the curve; demand for shorter dated instruments often increases when conditions are volatile and central banks are taking decisions on the future of monetary policy.