Innovative Technology for Trading Competitiveness

Jeff Hong, Capital Markets Industry, Sun Microsystems

Introduction

By one estimate, there are 40 plus pools of liquidity in the U.S. Industry pundits expect European markets to also embrace dark pools post-MiFID. With Regulation NMS, MiFID, and algorithmic trading further compounding liquidity fragmentation on both sides of the Atlantic, is it any wonder why algorithmic trading and liquidity discovery have become the hot topics for buy side and sell side firms globally? With these topics, a set of new challenges emerges ranging from the micro- to the macro-level; a few key challenges are highlighted for discussion in this article.

Impact of NEW Liquidity Pools and Market Regulations

Increased Fragmentation and Market Data Volume
There's no denying the impact of ATSs, ECNs - not to mention the regional stock exchanges post-demutualization – in the siphoning of liquidity away from the NYSE. In the U.S., there are approximately 40 or so liquidity pools. They range from very public pools to static pools like ITG Posit to dark pools. According to the Tabb Group, these pools handle ten percent (10%) of shares traded; by the end of 2007, the Tabb Group estimate exceeds one-half billion shares. The top five (5) pools handle perhaps 50% of the volume on a daily basis. One such pool, Sigma X, announced it had crossed 103 million shares at the end of May 2007.

In Europe, the role of alternative liquidity pools becomes cloudier. In conversations with European banks – recently at an algorithmic trading roundtable hosted by Sun, one question kept popping up. Will European markets become as fragmented as U.S. equity markets? What will be the impact of Project Turquoise? According to Equiduct, 50% of equity transactions are already handled away from market centers like FTSE, CAC40, etc. Fragmentation in the U.S. has been driven by algorithmic trading and now by Regulation NMS. In Europe, will algorithmic trading have the same impact?

The consensus of the roundtable participants was “Absolutely”.

Compounding the liquidity fragmentation - some will say is one of the leading causes of fragmentation - is Regulation NMS in the U.S. and MiFID in the EU. Mainstream firms must factor best execution/best price into their choice of execution venue. This requirement is pushing firms to access, post, and trade on more trading venues directly. Correspondingly, market data volume and messages rates in the market and within each firm have increased significantly. Instances of cancelled orders and of crossed and locked quotes have also increased significantly. It’s impossible and unprofitable for traders to manually eyeball, analyze, and act on each data point.

New and Faster Market Participants
New liquidity pool development and regulations are not the only changes roiling today’s markets. In the U.S., there’s been a rapid shift in the makeup of the top liquidity providers at the major exchanges like the NASDAQ. According to Brian Hyndman, SVP of NASDAQ Transaction Services,“Lime and rivals such as Automated Trading Desk … accounts for about 40 percent of the roughly 2.3 billion shares traded every day on NASDAQ … That's up from about 5 percent in 1999.”

In April 2007, several of the top liquidity providers are relatively new entrants such as Lime Brokerage. Along with firms like Biremis and Tradebot, these companies collectively provided 40% of the shares. How? These firms have ultra fast trading systems, and many of their key clients are highly active hedge funds. Lime for instance claims it processes 6,000 orders per second and 175 million shares per day2, the majority posted by hedge fund traders.

End Result: Harder to Fill Block Trades
As a result of fragmentation and increased competition with these next-generation brokerages, existing agency brokerages and buy-side firms have commented that it's getting very hard to put up a block trade at an exchange. With the purported average trade size of 400 shares on NYSE, it would take a minimum of 250 transactions to buy/sell 100,000 shares. So despite the potential conflict of interest and other misgivings associated with trading in liquidity pools owned by firms with prop trading desks, most participants whom I talked with agreed that alternative liquidity pools play a significant role in block trading.

How to Compete, What to Do

With all these changes, a head of program trading at one buy-side firm, expressed his wish to “slow” the market. Slowing a market appears counter to the current trading environment’s emphasis on speed. However this idea makes sense when the quality of the trade becomes a factor; in this case, being strategic is just as important as being fast. And in block and algorithmic trading, minimizing the market impact of and information leakage about your trade is about as strategic as you can get.

Slowing the Market
One difference between successful and less successful trades means acting on truly significant events and not on noise. Firms adopting algorithmic trading are relying more on complex event processing (CEP) engines to filter and piece together seemingly random events into significant events. In this way, CEP software can help “slow” the market, by reducing the volume of data that reaches the trader or that is processed. The data that does reach the trader is enriched and/or significant, simplifying the trader’s ability to put up a trade.   An example of enriched data or a significant event is options Greek calculations or an earnings surprise respectively. Firms are building these algorithms based on CEP software from vendors like Aleri and on low latency market data systems such as Reuters RMDS or Wombat MDS optimized for Solaris 10 and Sun’s AMD Opteron and Intel x64 servers3.

Addressing Toxic Liquidity
As the use of alternative liquidity pools grows for block trading, one increasing use of CEP software is for the detection of gaming or toxic liquidity2 in these pools. At a London Stock Exchange symposium where I presented on dark pools, toxic liquidity was a concern expressed by some buy-side firms. This concern centers about the potential for abuse by a bank’s prop trader to post orders in its liquidity pool for the purpose of analyzing and then profiting on clients’ trades. In response, alternative liquidity pools are going the extra step to address the trust issue by installing policies and technologies to detect gaming, to block toxic liquidity from impacting price and execution quality, and to detect unethical trading practices such as front-running.

No one can estimate with certainty or at all the opportunity and actual costs incurred by toxic liquidity at this time.   Especially in the case of dark pools, their future success may well depend on addressing this issue satisfactorily for both regulators and institutions. CEP-based software can potentially help analyze complex events for patterns of gaming and for “real time” surveillance.

Minimizing Information Leakage
Firms are buying or building algorithms to access, navigate and plumb the depth of exchanges and dark pools for liquidity and price. These 3rd party algorithms promise super fast liquidity/price discovery and order execution of block orders with minimal information leakage. As in the case of the lifespan of information asymmetry among traders, the lifespan of trading strategies and algorithms is shrinking. CEP software can analyze price and spreads, bid/offer volume, et. al. to gauge the market impact of a strategy. Next generation algorithms will monitor and react to real-time market conditions.

Innovative Approaches to Trading Competitiveness

In reality, message rates will likely not slow down, but rather will become more volatile and continue to grow at an exponential rate. To this end, the CEP software and the entire trading eco-system from the data feeds to the operating system to server platform need to scale up, be highly available, function predictably, and be fast.

Focus on Latency and Throughput
As a first step, most firms are focused on reducing latency and increasing throughput in their pre-trade and execution systems including:

Adopting direct market access
Automating trading processes including using algorithmic trading
Upgrading to Reuters Market Data System (RMDS) 6 and Wombat MDS systems
Refreshing servers, OS, and network technologies e.g. Sun’s x64 systems, Solaris 10, and Infiniband respectively
Streamlining and tuning in-house applications

The above approaches are ones that most if not all firms are looking at today. Firms should look for out-of-the-box approaches to latency reduction that will give them a competitive edge; for those traders who do care about the fastest ping approach, the competitive edge is measured in micro-seconds. For all others, a competitive edge may mean “real time” surveillance, or may mean increasing performance without the data center sprawl.

Example: Market Data Systems Consolidation with Solaris Containers
An example of a more innovative approach to latency reduction involves consolidating multiple market data systems onto one server. A typical small RMDS deployment requires eight (8) dual-socket systems. Each RMDS module runs on its own system with other modules. While RMDS is essentially a single-threaded application, using Solaris Containers (a feature of Solaris 10), Sun was able to consolidate a small RMDS deployment onto a single Sun server running 8 Solaris Containers. With the reduction of servers, this Solaris Containers consolidation approach saved space, lowered administrative costs (heating too!), AND reduced latency significantly without impacting throughput. One could take the next step with Solaris Containers by consolidating other key components of the pre-trade eco-system such as the CEP engine and trading application for further latency reduction.

Summary

No one can predict what the market will look like five (5) years much less ten (10) years from today. Industry regulations, investor confidence, the emergence of the BRIC middle class and eco-Green movement, the intertwined economies and politics, cross-border mergers, et. al. will all contribute to the growth or decline of the market systems. What is certain over the next several years, that for institutional trading firms to compete successfully and more efficiently, they need to improve operating efficiency by automating their trading processes and to utilize new approaches and technologies such as CEP software and Solaris Containers.