The Gateway to Algorithmic and Automated Trading

Taming the Tiger

Published in Automated Trader Magazine Issue 07 October 2007

Michael Donahue, managing director of TransMarket Groupís Singapore operations, tells AT how the firm is meeting the challenges of automated trading in the diverse and fast-changing Asian financial markets.

How did TransMarket Group develop into the global trading operation it is today?

We started out in 1981 as a market-making and broking operation at the CBOT (Chicago Board of Trade), but within a few years we started operations in London to trade on LIFFE. As well as our long-term presence in New York and Sydney, we've more recently opened offices in Madrid (2003), Mumbai (2004) and Singapore (2006) and now have just shy of 200 traders. We were originally a floor-based trading firm with futures exchange memberships in Chicago, London and Sydney, but when the markets started going fully electronic in 1998-9, we expanded trading to Eurex, Intercontinental Exchange (ICE), Singapore and others. Our overall trading strategy is what I would describe as 'loose-form arbitrage'. We're not interested in buying a COMEX-listed gold contract and selling CBOT-listed gold against it, for example, because those kinds of trades tend to be fully arbed out. There's very little opportunity to capture any kind of incremental profits, although it's a different matter when these contracts are traded in Asia overnight and volumes are thinner. Instead, we tend to focus on cross-market spreads where there's demonstrable correlation.

Why and how did you establish trading operations in Singapore?

We see a lot of potential in Asia and chose Singapore over Hong Kong or Tokyo for a variety of reasons. First, Singapore is a lot more orientated toward listed futures and options than the other locations, but also cost of business is lower and finding highly qualified IT professionals is relatively easy. A key factor was the decision by CME (Chicago Mercantile Exchange), CBOT, Eurex and LIFFE to locate their Asian hubs here. That means we have direct gateways to the big four as well as ICE and the Singapore and Sydney futures exchanges, which already all have hubs here, as well as connections to Brokertec and eSpeed. We also have good access to many other Asian exchanges because our clearer, FIMAT, has located its infrastructure for Asian futures and options trading in Singapore. We plug directly into that infrastructure to trade with seven exchanges across Asia. Via the CME's Globex platform we also have access to NYMEX (New York Mercantile Exchange) and COMEX (New York Commodities Exchange). Within the next year hopefully the Chinese markets will open up and we'll have access to the Shanghai and Dalian futures exchanges.

We're always interested in being among the first participants as any new market opens up, so we're using Singapore to identify opportunities in the Asian markets. Singapore is also a good location because it is more or less equidistant between Mumbai, where we have a team of 50, and the Chinese markets.

"We're always interested in being among the first participants as any new market opens up, so we're using Singapore to identify opportunities in the Asian markets."

How important is speed, both your own and that of the exchanges you trade on?

The natural latency we suffer from sitting in Singapore, Sydney or Mumbai puts us at a significant disadvantage when trading on the big four exchanges. Because we cannot make an electron go any faster than 300,000 kilometres a second, we tend not to do bog-standard arbitrage trades that someone located in London or Chicago would engage in. Even when we have low natural latency within Asia, the exchanges' order processing times comes into play. Latency considerations increase with the number of participants doing the same kind of trade, because it soon becomes a matter of who has the quickest machine. Locating your servers close to the host helps, but co-location services haven't taken off in Asia yet. Automated trading is in its infancy on Asian futures exchanges compared with Asian equities exchanges, which have begun to adopt the kind of latency standards you'd expect in a US or European exchange.

How do Asian trading costs compare to other markets?

As high-volume traders, we're very price sensitive and Asia's still an expensive place to do business with the exception of the Singapore exchange.

The two main costs anywhere are broker and exchange fees. The majority of Asian markets are still heavily voice-broked and this adds an extra layer of cost. The role of the broker may have diminished elsewhere, but in Asia calling up your broker is often the only way to do business. Exchange fees can be high also, with some charging almost a multiple of the broker's fees. The Tokyo Commodity Exchange (TOCOM) is an interesting example; clearing costs and exchange fees are relatively low, but high margins charged by brokers and clearers are stifling growth. To attract retail customers, TOCOM halved tick sizes on a number of contracts, but did not make a commensurate cut in fees or initial margins - nor did the brokers - so trading costs actually doubled. This knocked a lot arbitrage activity out of the market and volumes dried up. We've been lobbying the exchange to reduce fees or return contract sizes to their notional value prior to 2003. The one contract they didn't change was gold and that's where we're most active. For speculators it's not that important that transaction costs are low, but for arbitrage firms like us pennies per contract make a big difference.

How are Asian futures exchanges attracting automated traders?

The most active exchange is the Singapore Exchange (SGX). Historically, a large part of its business has been based on listing other exchange's contracts, e.g. trading eurodollar contracts when the CME was closed. Now exchanges are open all ours, the SGX is introducing contracts that allow them to compete in the Asian market - such as index products - with the aim of becoming the pre-eminent futures exchange in the region. Moreover, SGX have very low transaction costs and seem to understand the needs of the algo trader better than most exchanges.

"The role of the broker may have diminished elsewhere, but in Asia calling up your broker is often the only way to do business."

The Japanese exchanges are also waking up to this. We've had meetings with TOCOM, and the Tokyo and Osaka exchanges, about how they can reinvigorate volumes. Clearly, infrastructure improvements are crucial and they're all in the process of doing that. But to attract firms like ours, you have to be competitive globally on a transaction cost basis. That's hard when Eurex is trading around four million lots a day across all contracts and have such economies of scale that they can turn a profit on incremental margins. Asian exchanges are really about a tenth of the size of the big exchanges by volume, but they don't cost a tenth to operate. We anticipate a major consolidation in the futures and options space and it wouldn't surprise me if within five years there were just a handful major exchanges globally, with a wider range of contracts incorporated onto the existing platforms of those major exchanges, for example, CME the listing Korea Composite Stock Price Index futures.

How does TransMarket Group develop trading models for Asia?

We've undergone something of a transformation over the last six months or so. Historically, automated trading projects were developed at the local level. So if we wanted to do cross-market spreading in Europe, for example, our London developers would automate those trades. But recently we've consolidated our resources by employing experts with global experience at our Chicago HQ that can do the development work on behalf of the local offices. In future, the financial engineers and programmers in Chicago will explore the viability of automating trades as requested by local offices, and begin writing the code, but they would then work closely with the local team to deploy the algorithm. However, we will still need senior programmers and financial engineers to help us develop our automated trading operations at the local level.

It's a sign of the times that we've increased the number of programmers and project managers while downsizing our number of traders. This reflects where the business is heading. In Europe and the US, we've already seen the demise of the point-and-click trader over the last three to four years. Our role is not to pick the direction of the market consistently - very few can do that - but we do know what spreads are doing within any particular market. What we're focused on is relative value. The explosion of automated trading in the last couple of years is such that old-fashioned point-and-click people are leaving the market on a weekly basis.

To what extent does TransMarket Group buy or build its technology platform?

For the big four exchanges, we'll still do our own programming, but for the smaller exchanges around the world we work very closely with third-party vendors to develop their functionality so our trading can be done on an automated basis. As more exchanges become accessible and competition increases between the big exchanges, trading firms must be able to reach as many markets as possible, especially if, like us, you're looking to get access to the less liquid markets where there still is 'edge'. Because technology is changing so rapidly and exchange systems are being constantly upgraded, it becomes prohibitively expensive. If every upgrade costs $100-200,000, two upgrades per exchange per year becomes very expensive very quickly if you're connected to 10-20 exchanges. We would rather leave that in the hands of experts who have experience with the exchanges' systems and can do the upgrades in weeks not months.

"As more exchanges become accessible and competition increases between the big exchanges, trading firms must be able to reach as many markets as possible."

Because there's not any one system on the market that provides the combination of functionality and connectivity that we need in Asia, we use a combination of systems. We use Trading Technologies to access the big four markets, but we rely on PATSysytems to connect to the eight Asian exchanges we currently trade on and we're also a long-term customer of RTS. The combination of platforms is perhaps not ideal so we cajole certain providers to improve coverage and others to improve functionality.

What obstacles are there to growth of volumes in Asia?

Our volumes are still relatively low in Asia, but we've only been operational in Singapore since July 2006. Across our three Asian offices, we trade in excess of 2.5-3m contracts per month, but 'local' volumes are perhaps 1.5m per month, with the majority traded through Sydney. It took us longer than we expected to get our Asian connectivity solution in place and there are still hiccups. Anyone who's trying to put together a full-blown Asian connectivity solution - whether with vendors or writing their own - will come up against seemingly insurmountable problems on a regular basis. That's a function of the infrastructure in place at some of the Asian exchanges, but also the sheer size of the region. If you're hired by a western firm to 'run Asia' you spend a good deal of your time in the air between Tokyo, Mumbai, Sydney, Shanghai etc. From a trading perspective, there's no all-encompassing solution that connects the region together effectively.

"From a trading perspective, there's no all-encompassing solution that connects the region together effectively."

And we can't grow our Asian volumes on our own. As a market maker, we provide prices but need counterparties with which to trade. We anticipate that as the local market becomes more sophisticated and gets access to a wider range of exchanges, volumes will increase significantly. At present, a lot of the local exchanges rely heavily on non-domestic users to engage the market, with the exception of Korea, where around 80 per cent of trading in major contracts is done by locals. But when brokers are comfortable trading the Nikkei in Osaka, it's difficult for the SGX to capture that business. Brokers shouldn't care where they go to get business done, but they are creatures of habit and won't necessarily go to where they can get a better price or lower transactions costs. We expect European and US demand to drive change, and hopefully that will then spur on more activity by Asian exchange clientele.

We believe more and more investors will gain access to Asian markets and this should drive volumes in the long term. This is already happening. There's a lot of activity focused on attracting Chinese institutional participation and I would think it's pretty safe to assume that the big four exchanges will eventually locate hubs in China once regulatory barriers come down.