Sunteți pe pagina 1din 24

Automated Trading: Fueling Global FX Market Growth

2011 QuantHouse. All rights reserved. Reproduction of this report by any means is strictly prohibited.

Automated FX Trading

July 2011

TABLE OF CONTENTS
EXECUTIVE SUMMARY .................................................................................................................................... 3 INTRODUCTION .............................................................................................................................................. 5 MARKET TRENDS ............................................................................................................................................ 6 GLOBAL FX MARKET GROWTH .................................................................................................................. 6 DIVERSIFICATION OF MARKET PARTICIPANTS .......................................................................................... 6 MARKET FRAGMENTATION....................................................................................................................... 7 INCREASED ADOPTION IN ELECTRONIC TRADING .................................................................................... 8 INCREASED ADOPTION IN AUTOMATED TRADING ................................................................................... 9 INCREASED INTERNALIZATION BY SINGLE BANK PLATFORMS ................................................................ 10 NEW OPPORTUNITIES AND CHALLENGES ..................................................................................................... 11 OPPORTUNITIES IN PRICE AGGREGATION .............................................................................................. 11 CHALLENGES IN MANAGING MULTIPLE TRADING VENUES .................................................................... 11 CHANGING RELATIONSHIP BETWEEN BANKS AND AUTOMATED TRADING FIRMS ................................ 12 DECLINING SHELF-LIFE OF TRADING STRATEGIES ................................................................................... 12 KEY INGREDIENTS IN AUTOMATED TRADING ............................................................................................... 14 UNDERSTANDING THE STRATEGY DEVELOPMENT WORKFLOW ............................................................ 14 CHALLENGES IN FX FOR AUTOMATED TRADING .................................................................................... 15 AUTOMATED FX TRADING INFRASTRUCTURE ........................................................................................ 16 CASE STUDY .................................................................................................................................................. 19 CONCLUSION ................................................................................................................................................ 20 ABOUT QUANTHOUSE: ................................................................................................................................. 21 QUANTFACTORY AUTOMATED TRADING PLATFORM ......................................................................... 21 DATACENTER SERVER ........................................................................................................................ 22 DEVELOPMENT PLATFORM ............................................................................................................... 22 EXECUTION PLATFORM ..................................................................................................................... 23 QUANTFEED ULTRA LOW LATENCY MARKET DATA ............................................................................. 21 QUANTLINK ULTRA LOW LATENCY TRADING INFRASTRUCTURE ......................................................... 24

2011 QuantHouse. All rights reserved. Reproduction of this report by any means is strictly prohibited.

Automated FX Trading

July 2011

EXECUTIVE SUMMARY
Automated Trading: Fueling Global FX Market Growth commissioned by QuantHouse and produced by Aite Group, analyzes the rapidly evolving automated trading in the global FX market. The study examines the background behind the overall growth of automated trading in FX and highlights key technology components that are needed build an effective presence in the marketplace. The study concludes with a case study. Key takeaways from the study include the following: During the latter part of 2008 and well into 2009, customers faced a much different market from previous years, marked by wider spreads and declining liquidity. Consequently, average daily trade volume returned to earth in 2009, standing at approximately US$3.7 trillion. However, the global FX market bounced back quite nicely with a strong volume growth in 2010, reaching US$4.1 trillion. One of the most significant changes that the FX market is currently undergoing is the substantial increase in trading activities from non-bank financial institutions. In 1995, banks accounted for 64% of all trading, but that figure declined to below 40% by 2010. During the same time period, trading from non-bank financial institutions increased to 48% from 20% in 1995, thereby becoming the largest counterparty in the FX market. Electronic trading adoption continues to increase in the global FX market. Given that markets remain fragmented, the need to source multiple liquidity pools simultaneously has only strengthened the overall position of electronic trading. At the end of 2010, electronic trading accounted for 68% of all FX trading. In recent years, a new breed of actively trading market participants has driven significant innovation in trading technology. As traders navigate through growing complexity within the marketplace, development and implementation of automated trading strategies have become vital part of some of the more sophisticated FX traders. Automated trading in FX is poised to grow quite rapidly over the next few years, as the first-generation automated trading firms are joined by an influx of nextgeneration equity and futures automated trading firms looking to capture uncorrelated alpha in FX. At the end of 2010, automated trading accounted for 29% of overall trade volume. This figure is expected to hit more than 40% by the end of 2012 Challenges in automated FX trading include 1) latency; 2) decentralized, highly fragmented across numerous single bank and ECN execution venues; 3) largely unregulated market with lack of an industry benchmark and best execution obligation; 4) venue-specific market structure and communication standards; 5) lack of a comprehensive, consolidated view into the entire market; and 6) dispersed price discovery process. 3

2011 QuantHouse. All rights reserved. Reproduction of this report by any means is strictly prohibited.

Automated FX Trading

July 2011

Despite existing challenges, firms have been implementing automated FX trading for many years and developing and maintaining a robust automated FX trading infrastructure has been vital to their continuing success. Key components of automated FX trading infrastructure include the following: Access to real-time and historical data Robust strategy development framework Trading engine Low latency connectivity Trade analytics

Automated trading in FX is not driven entirely by independent low latency proprietary shops or hedge funds. In fact, most of the global FX banks have either acquired or developed robust low latency FX prop desks to compete in the marketplace and account for a significant percentage of the overall market share. In recent months, several traders from major banks have left to start their own firms, taking with them not only their quantitative skills, but also their market structure knowledge, which could potentially pave the way for the next phase in automated trading, in which non-bank trading firms play a larger role in liquidity provision. The first round may be over in FX automated trading, but many more remain before the real winners can be determined in this rapidly evolving marketplace.

2011 QuantHouse. All rights reserved. Reproduction of this report by any means is strictly prohibited.

Automated FX Trading

July 2011

INTRODUCTION
After robust market growth in 2007 and 2008, the global FX market experienced lower volatility, lower volume, and in most cases, wider spreads in 2009. The global FX market fared much better in 2010, resuming its overall growth and seeing volume levels skyrocket, returning to close to its record 2008 level. In the meantime, large FX banks have pressed on, maintaining the global FX markets overall market dominance, increasingly relying on internalization to manage their P&L. The global FX market remains highly fragmented with single bank, multi-bank, and interbank execution venues vying for precious market share. On the exchange-traded market, the CME Group has developed a formidable FX futures franchise that currently has a virtual monopoly. With the client-to-dealer market outpacing the growth of inter-dealer market in terms of overall trading volume, much focus has been placed on the dynamic changes within the different client segments. In recent years, a new breed of actively trading market participants has driven significant innovation in trading technology. As traders navigate through growing complexity within the marketplace, development and implementation of automated trading strategies have become vital part of some of the more sophisticated FX traders. Another important thing to note is that automated trading in FX is not driven entirely by independent low latency proprietary shops or hedge funds. In fact, most of the global FX banks have either acquired or developed robust low latency FX prop desks to compete in the marketplace and account for a significant percentage of the overall market share. Under this competitive environment, this study analyzes the rapidly evolving automated trading in the global FX market. The study examines the background behind the overall growth of automated trading in FX and highlights key technology components that are needed build an effective presence in the marketplace.

2011 QuantHouse. All rights reserved. Reproduction of this report by any means is strictly prohibited.

Automated FX Trading

July 2011

MARKET TRENDS
G LO BA L FX M A R K E T G ROW T H
Thanks in large part to high volatility, 2008 yielded historical highs in terms of overall trading volume, followed by an inevitable decline in 2009. The industry averaged approximately US$4.3 trillion in daily trading volume in 2008 compared to about US$4 trillion in 2007. During the latter part of 2008 and well into 2009, customers faced a much different market from previous years, marked by wider spreads and declining liquidity. Consequently, average daily trade volume returned to earth in 2009, standing at approximately US$3.7 trillion. However, the global FX market bounced back quite nicely with a strong volume growth in 2010, reaching US$4.1 trillion (Figure 1). Figure 1: Growth of Global FX Market
Average Daily Trade Volume (In US$ Billions) $5,000 $4,500 $4,000 $3,500 $3,000 $2,500 $2,000 $1,500 $1,000 $500 $1998 2001 2004 2007 2008 2009 2010

Source: BIS, Bank of England Foreign Exchange Joint Standing Committee (JSC),New York Foreign Exchange Committee, Singapore Foreign Exchange Market Committee, Canadian Foreign Exchange Committee, Tokyo Foreign Exchange Joint Standing Committee, and Aite Group

D I VE RS I F I C AT I O N O F M A R K E T PART I C I PA N TS
One of the most significant changes that the FX market is currently undergoing is the substantial increase in trading activities from non-bank financial institutions (i.e., other financial institutions in Figure 2). In 1995, banks accounted for 64% of all trading, but that figure declined to below 40% by 2010. During the same time period, trading from non-bank financial institutions increased to 48% from 20% in 1995, thereby becoming the largest counterparty in the FX market.

2011 QuantHouse. All rights reserved. Reproduction of this report by any means is strictly prohibited.

Automated FX Trading

July 2011

Figure 2: Diversification of Market Participants


Reported FX Market Turnover by Counterparty 70% 60% 50% 40% 30% 20% 10% 0% 1995 Dealers 1998 2001 2004 2007 2010

Other Financial Institutions

Non-Financial Customers

Source: BIS

Following the theme of declining bank transactions, as banks continue to fine-tune their ability to manage their FX risk books in real-time, the client-to-dealer market has increased its overall market share over the last few years at the expense of the interbank market. By the end of 2010, the client-to-dealer market accounted for 61% of overall FX trading while the interbank stood at 39%. In comparison, the interbank market represented close to 60% of the marketplace in 2001. Figure 3: Client-to-Dealer vs. Inter-Dealer
Client-to-Dealer vs. Interbank 100% 90% 80% 70% 60% 50% 40% 30% 20% 10% 0% 1998 2001 2004 2007 2008 2009 2010 64% 59% 53% 43% 41% 40% 39% Client-to-dealer Interbank 36% 41% 47% 57% 59% 60% 61%

Source: BIS

M A R K E T FR AG ME NTAT I O N
2011 QuantHouse. All rights reserved. Reproduction of this report by any means is strictly prohibited.

Automated FX Trading

July 2011

The global FX market remains highly fragmented, represented by the still significant voice market, as well as numerous electronic venues, including single bank, multi-bank, and interbank venues. In recent years, the term FX ECNs has emerged to describe mixture of multi-bank and interbank venues that support ECN-like functionality, including streaming live quotes. Examining the FX ECN market alone, EBS currently holds the top spot with 23% market share followed very closely in second place by Thomson Reuters. While the CME Group is an exchange with its FX futures product compared to the OTC products of the other FX ECNs, it is clear from a trading value perspective that the CME Group has become a major player in the global FX market (Figure 4). Figure 4: Market Fragmentation in FX ECN Market
FX Venues Competitive Landscape (As of January 2011) Hotspot FX 7% Fxall 10% ICAP (EBS) 20%

FX Connect 13%

Thomson Reuters 19% CME Group 17%

Currenex 14%

Source: ECNs, Aite Group

I NC RE AS E D A D OP T I O N I N E LE C T RO N I C T R A D I N G
Unlike other over-the-counter (OTC) markets, driven by early acceptance in the interbank market, electronic trading adoption continues to increase in the global FX market. Given that markets remain fragmented, the need to source multiple liquidity pools simultaneously has only strengthened the overall position of electronic trading. At the end of 2010, electronic trading accounted for 68% of all FX trading. This figure is expected to reach more than 70% by end of 2012 (Figure 5).

2011 QuantHouse. All rights reserved. Reproduction of this report by any means is strictly prohibited.

Automated FX Trading

July 2011

Figure 5: Projected Electronic Trading in FX


Projected Electronic Trading in FX 80% 70% 60% 50% 40% 30% 20% 10% 0%

Source: Aite Group

I NC RE AS E D A D OP T I O N I N AUTOM AT E D T R A D I NG
Automated trading in FX is poised to grow quite rapidly over the next few years, as the firstgeneration automated trading firms are joined by an influx of next-generation equity and futures automated trading firms looking to capture uncorrelated alpha in FX. In addition, new automated trading firms have emerged in recent months, formed by FX quants and traders who have left large banks looking to capture new opportunities on the other side of the market. At the end of 2010, automated trading accounted for 29% of overall trade volume. This figure is expected to hit more than 40% by the end of 2012 (Figure 6).

2011 QuantHouse. All rights reserved. Reproduction of this report by any means is strictly prohibited.

Automated FX Trading

July 2011

Figure 6: Projected Adoption of Automated Trading in FX


Estimated Automated Trading in FX 45% 40% 35% 30% 25% 20% 15% 10% 5% 0% 2003 2004 2005 2006 2007 2008 2009 2010 e2011 e2012

Source: Aite Group

I NC RE AS E D I N T E R N A L I Z AT I O N BY S I N G L E BA NK P LAT FOR M S
Another key trend over the last few years has been the increasing effectiveness of large FX banks in managing their risk books when trading against customers. By utilizing sophisticated pricing engine and real-time internalization capabilities, large FX banks have become quite adept at showing varying prices to different types of customer segments as well as efficient at deciding when to internalize vs. utilize traditional interbank markets to lay off their risk. In a way, the aftermath of the credit crisis of 2008 has only reinforced banks need to internalize, particularly as regulators and politicians continue to emphasize banks need to lower their overall risk profile. Consequently, the need to better segment customer flow has been on top of banks overall client-facing trading strategy so that they can optimize their balance sheets and better manage their profit and loss (P&L).

2011 QuantHouse. All rights reserved. Reproduction of this report by any means is strictly prohibited.

10

Automated FX Trading

July 2011

NEW OPPORTUNITIES AND CHALLENGES


Against these diverse market realities, plenty of opportunities currently exist for those firms looking to capture alpha in the global FX market. At the same time, however, obvious challenges exist, driven by the OTC nature of the FX market.

O P P ORT U N I T I E S I N P R I C E AG G R EG AT I O N
From a technology perspective, clear opportunities currently exist in the FX market in terms of providing effective price aggregation services to overcome the challenges that stems from market fragmentation and lack of industry-wide consolidated tape. Price aggregation is a difficult proposition in the FX market due to the differences that exist across multiple venues and how they operate. While the larger FX players have developed price aggregation capabilities internally, recent efforts made by third-party vendors hold hope for those smaller players looking for a level playing field. From strategy development to actual execution, getting the price aggregation piece right is the first important steps towards developing a winning automated trading operations.

C H AL LE N G E S I N M A N AG I N G M U LT IP L E T R A D I N G VE N U E S
Not so long ago, the traditional FX venues, such as EBS and Reuters dominated the market, making it less cumbersome for traders to figure out where the market is headed. However, the recent trend of growing market presence of newer ECNs as well as single bank platforms, coinciding with declining market share of EBS and Reuters have made it essential that firms attempt to source liquidity from various venues. A few of the challenges arising from multiple trading venues include the following: Different locations and built-in distance latency: FX execution venues are located in many different regional and financial centers, resulting in inherent distance latency for most firms. Different frequency of price distribution: Each venue has its own timeframes in terms of price distribution. In cases of single bank platforms, they will be distributing prices to many different FX ECNs, in addition to their own in different latency. Depth of book: Wide differences exist in terms how much of the order book is actually represented in each venue. Tick size and order types: FX venues also support different tick sizes and order types, adding another layer complexity in terms of firms trying to figure out the best location for execution at any given time. Constant upgrades: Venues are constantly going through enhancements of new functionality, order types, reduction in latency and more. These changes could have an impact on effectiveness of strategies so firms need to stay on top of these changes and tweak strategies as needed. 11

2011 QuantHouse. All rights reserved. Reproduction of this report by any means is strictly prohibited.

Automated FX Trading

July 2011

C H A NG I NG R E LAT I O N SH I P B E T W E E N BA N KS A N D AUTOM AT E D T R A D I NG FI R M S
While the banks have gone through a series of consolidations, leading to fewer banks making markets, the FX market continues to evolve with new types of customer segments beyond the traditional corporate and asset manager customer base. Since 2002, actively trading hedge funds and proprietary trading firms have made a huge impact in the FX market, driven by a robust IT infrastructure and development of automated trading strategies. In fact, some would argue that the large FX banks learned a painful lesson between 2002 and 2006, driven by latency arbitrage strategies in the first wave of automated trading firms. As a direct consequence of this experience, the banks initiated massive overhaul of their trading infrastructure, not only focusing on drastically lowering latency levels within the single bank platforms, but also on developing more efficient pricing engine and internalization capabilities to better manage their risk books against different types of customer segments. It was also during this time that the banks decided to kick out those automated trading firms whose relationships they deemed unprofitable. Since 2008, however, banks and certain segments of the automated trading community are attempting to peacefully co-exist. As banks continue to increase their internalization efforts, potential liquidity from automated trading firms has become more attractive. On the other hand, automated trading firms have come to realize that banks have a vital position in the FX market; in order to ensure continued success, co-opetition has become a competitive necessity. One potential change that could change the balance in the market is successful implementations of centralized clearing in the OTC marketplace. If non-bank automated trading firms can become direct clearing members for OTC products, and also illustrate their commitment to taking more risk as a legitimate liquidity provider, banks stranglehold in the FX market could be weakened and hence open up a new phase of intense competition.

D EC LI N I NG SH E LF - L I F E O F T R AD I NG ST R AT EG I E S
While there are long-term quant strategies that have generated great returns, there is a growing group of high-frequency trading shops relying on implementation of short-term alpha discovery strategies. This requires constant monitoring of the strategies, and the ability to isolate ineffective variables or assumptions rapidly so that the strategy can be fine-tuned and launched live into the market again. With a notable increase in the total number of quantitatively-driven funds, the pressure to come up with the next big quant model has grown exponentially. This reality has also fueled the need to streamline the overall workflow process and shorten the duration of the entire investment selection life cycle to compete more effectively. On average, the entire process of idea generation to implementation can take anywhere from 10 to 28 weeks. Given the fact that certain short-term strategies only remain effective for three to four months, rapid construction and implementation of alpha models become that much more urgent. As a result, the search for a single unifying platform that can create a more centralized and streamlined process and functionality appears inevitable.
2011 QuantHouse. All rights reserved. Reproduction of this report by any means is strictly prohibited.

12

Automated FX Trading

July 2011

And as the shelf life of alpha strategies continues to decrease driven by competition and changing market conditions, there is growing recognition that the links between alpha discovery and execution cannot be ignored.

2011 QuantHouse. All rights reserved. Reproduction of this report by any means is strictly prohibited.

13

Automated FX Trading

July 2011

KEY INGREDIENTS IN AUTOMATED TRADING


U ND E RSTA N D I NG T H E ST R AT EGY D E V E LO P M E N T WO R K F LOW
There is no industry standard when it comes to strategy analysis workflow. Depending on the firm and also depending very much on whether or not they have long-term or short-term trading philosophies specific steps within the workflow and their duration will vary. Figure 7: Alpha Generation Workflow

Source: Aite Group

A high-level, generally accepted alpha discovery workflow goes as follows: Alpha thesis/philosophy: The initial starting point for most firms is the basic thesis or philosophy of alpha discovery. During this abstract phase, analysts and managers are contemplating basic questions like instrument alternatives, value versus growth, domestic versus international and other items such as decisioning variables. All funds are built around a specific alpha philosophy. Data acquisition and preparation: Perhaps the most time-consuming and tedious part of the entire workflow, in order to accommodate testing and analysis of a given model, appropriate data must be acquired, typically transformed and normalized, and then stored and ready to be mined and manipulated. Depending on the type of firm, types of data would include fundamental and technical data as well as historical and real-time tick data. In addition, firms have used a wide variety of data storage options (including Excel, proprietary programs, relational databases, and high performance databases) depending on the type of data, size of data sets, and acceptable level of latency in testing and analysis. Alpha discovery: Once the data has been loaded properly, quants will mine the data sets to identify patterns, events, and market conditions that may lead to alpha discovery. During this phase, the quants will also either acquire or code various mathematical, statistical, and technical functions and routines to be used to formulate their alpha models. Typically using C, C++, C#, Java or other languages, 14

2011 QuantHouse. All rights reserved. Reproduction of this report by any means is strictly prohibited.

Automated FX Trading

July 2011

quants will create specific parameters for the models. Most quants use popular third-party statistical packages (i.e., MATLAB, S+, and R) during this process; often the same package they have used during their academic training. Back-testing: Using relevant sets of historical data, quants will back-test alpha models through varied market and economic conditions, multiple combinations of parameters, different portfolio attributes and more to validate or debunk the various alpha models. Analysis and optimization: The results from back-testing are analyzed, and new factors are added and/or additional coding takes place to fine-tune and optimize the alpha model. Use of visualization and various portfolio optimization tools would be used during this phase to ensure appropriate risk parameters and diversifications. Simulations: Once the alpha model has been tightened up, intensive simulation takes place using either historical or live data under varying known and hypothetical market conditions. These simulations determine the appropriate levels of risk and investment parameters to ensure effectiveness and profitability of given alpha models. Production and live launch: Once the alpha model has been validated, it will be coded into production, after which integration work takes place to create an automated investment environment. In order to ensure automated trading, the model also needs to communicate with execution management systems (EMS) typically via FIX.

The entire workflow in quantitative modeling is an iterative process, a system of continuous finetuning. The quant manager tests the investment model for validation in light of ever-changing market conditions.

C H A L LE N G E S I N FX FOR AU TO M AT E D T R AD I NG
While there is certainly a growing market demand for low latency trading infrastructure within the FX market, the FX market is relatively slow compared to the U.S. cash equities market (where the accepted level of latency has fallen below hundreds of microseconds), typically operating in the hundreds-of-milliseconds range. Still, continuing efforts to eliminate precious milliseconds off matching engines and trading infrastructure have led to most firms being able to develop internal trading latency levels of single-digit milliseconds. Unfortunately, the globally distributed nature of the FX market with three major FX centers New York, London, and Tokyo has made it tough to alleviate latency caused by physical distance. As a result, depending on the location of the trader and the matching engine, latency levels can vary widely from less than 15 milliseconds (i.e., local transaction) to close to 300 milliseconds (cross-border transactions). In the short-term, basic latency issue is faced by all market participants. As a result, most firms have focused on addressing the three major technology areas to address the latency issue:

2011 QuantHouse. All rights reserved. Reproduction of this report by any means is strictly prohibited.

15

Automated FX Trading

July 2011

Network optimization; Low latency connectivity; and Colocation/proximity solution.

In addition to the obvious latency struggles, traders in the global FX market face more pressing issues, driven largely by the fact that the FX market is still an OTC marketplace. Consequently, firms engaged in automated FX trading must overcome the following barriers to ensure that their trading needs are met: Decentralized, highly fragmented across numerous single bank and ECN execution venues; Unregulated for the most part, with lack of an industry benchmark and best execution obligation; Venue-specific market structure and communication standards; Lack of a comprehensive, consolidated view into the entire market; and Dispersed price discovery process.

A trader attempting to trade even a very liquid currency pair might need to capture market data from numerous locations scattered around the different time zones to ensure optimal trading execution. Balancing between the inherent distance latency hurdles and decentralized nature of the FX market is not a simple chore. For most of the large automated trading firms, solving the market fragmentation and latency issues has meant mostly internal development with close cooperation between FX ECNs. In recent years, however, a few technology service providers have emerged to provide trading infrastructure to facilitate those firms looking to jumpstart in automated FX trading.

AUTOM AT E D FX T R A D I NG I N F R AST R UC T U R E
Despite these obstacles, firms have been implemented automated FX trading for many years and developing and maintaining a robust automated FX trading infrastructure has been vital to their continuing success. Some of the key components of an automated FX trading infrastructure include the following:

2011 QuantHouse. All rights reserved. Reproduction of this report by any means is strictly prohibited.

16

Automated FX Trading

July 2011

Figure 8: Automated FX Trading Infrastructure :

Source: Aite Group

Access to real-time and historical data: Considered to be the most important piece time data: of the automated trading environment, access to accurate real time and historical real-time data drives the development of strategies as well as live implementation of those strategies to capture alpha. Key features of this would include the following: Ability to capture, store, replay and update live data Support for handling different time slices and types Integration with market leading data vendors and feeds egration

Robust strategy development framework: Once normalized data can be acquired, framework firms can start developing, back testing, and optimizing strategies prior to going into back-testing, live production. In todays automated trading technology market, there are viable trading third-party alpha generation platforms that help quants move from idea generation party and development to production in a seamless fashion. Key features of this would include the following: Support for industry standard development languages standar Seamless workflow from development and testing to optimization and production Support for multiple asset classes and currencies

2011 QuantHouse. All rights reserved. Reproduction of this report by any means is strictly prohibited. . prohibited.

17

Automated FX Trading

July 2011

Full integration with external applications, including leading statistical packages, such as MATLAB Robust back-testing and optimization environment Trading engine: Once the strategy has been thoroughly back-tested and optimized, traders can put it into production and leverage the trading engine to send out and manage orders to various execution venues. Key features of this would include the following: Seamless transition from development to production Ability to start, modify, and stop strategies Position and order management capabilities Integration with third-party trading front ends Low latency connectivity: Firms must have access to low latency connectivity to ensure that opportunities can be identified and acted upon in a timely manner. Key features of this would include the following: Global connectivity to leading execution venues and brokers Access to colocation and/or proximity services Reliability and predictability of connectivity Trade analytics: As executions take place, firms can utilize various trade analytics to measure overall performance of automated trading strategies. Intraday trade analytics can provide real-time feedback on overall execution quality so that the trader can make necessary adjustments to underlying strategy assumptions. Key features of this would include the following: Real-time monitoring of live orders and positions Ability to tweak working strategies on the fly Feeding performance measurement data back to development environment to improve effectiveness of strategies

2011 QuantHouse. All rights reserved. Reproduction of this report by any means is strictly prohibited.

18

Automated FX Trading

July 2011

CASE STUDY
Founded in November 2009, Fisycs Capital is a Paris-based systematic, quantitative hedge fund focusing exclusively on liquid markets. Fisycs Capital is currently authorized and regulated by the AMF (Autorit des Marchs Financiers) in France. Current list of strategies implemented by Fisycs Capital include the following: FX Intraday U.S. Equity Market Neutral Global Macro FX Fundamental

In the process of setting up its investment and trading technology infrastructure, Fisycs Capital examined many different options, both in-house and vendor supplied platforms. At the end of its analysis, Fisycs selected the QuantHouses QuantFACTORY platform for various reasons: Ability to connect strategy to multiple venues and brokers at the same time. Fisycs Capital has a reasonable number of strategies and QuantFACTORY makes message loading quite simple and seamless Ability to customize and the vendors receptiveness to incorporating requested new features into their development plans.

Focusing on the automated FX market, Fisycs Capital believes that the ultimate benefits of leveraging the QuantFACTORY platform are the following: As a multi-strategy hedge fund, very easy to support multi-desk trading activities and also very easy to support risk management and asset allocation off the same platform. Fairly shallow learning curve in grasping the code needed to test and very easy to use. QuantFACTORY is very good at handling data; can take in multiple data sources very quickly. All strategies modules are separated but can communicate with each other. Openness of the platform, enabling Fisycs Capital to link their own execution algos. Helps Fisycs Capital to quickly improve trading ideas, thus shortening the time from research or recalibration to live execution.

2011 QuantHouse. All rights reserved. Reproduction of this report by any means is strictly prohibited.

19

Automated FX Trading

July 2011

CONCLUSION
Despite the influx of automated trading flow, the banks still maintain enough market clout to hold onto their competitive edge. In fact, the FX market remains very much a two-tiered market; single bank platforms tend to interact against customers who can be perceived either as less sophisticated or less sensitive to explicit transaction costs (i.e., willing to take wider spreads to get the trades done). On the other hand, most automated trading is occurring on FX ECNs. While the liquidity ultimately comes from banks, this two-tiered approach has enabled banks to be more efficient in trading against different types of end customers. In this rapidly evolving automated trading marketplace, technology has become a competitive differentiator, enabling firms to quickly develop and launch new strategies to take advantage of changing market conditions. While the initial stage of technology development has been driven by large firms with significant in-house application development capabilities, the availability of cost effective, robust third-party solutions will go a long way in leveling the playfield for the rest of the marketplace moving forward. In recent months, several traders from major banks have left to start their own firms, taking with them not only their quantitative skills, but also their market structure knowledge, which could potentially pave the way for the next phase in automated trading, in which non-bank trading firms play a larger role in liquidity provision. The first round may be over in FX automated trading, but many more remain before the real winners can be determined in this rapidly evolving marketplace.

2011 QuantHouse. All rights reserved. Reproduction of this report by any means is strictly prohibited.

20

Automated FX Trading

July 2011

ABOUT QUANTHOUSE:
QuantHouse is an independent global provider of low-latency trading solutions. These include low latency ultra-low-latency market data technologies, Automated Trading Platform, trading infrastructure latency Platform, and order routing services to help hedge funds, proprietary desks and low-latency low latency-sensitive sell side firms take the lead. With more than 14 international hosting facilities within or near more than 45 exchanges all interconnected by our proprietary fibre optic network, QuantHouse clients interconnected benefit from a leading global trading infrastructure for ultimate results.

Q UA N T FAC TO RY AUTO M AT E D TR A D I N G P L AT FO R M AU TO FOR


http://www.quanthouse.com/?q=node/19 QuantFACTORY, the firms advanced algo-trading development tool, is an Integrated algo Development Environment (IDE) designed to help clients significantly optimize each step of the automated trading development cycle. The frameworks openness and industry standard language enables quant traders, researchers and developers to quickly build and deploy alpha re models, allowing them to focus on business development.

QuantFACTORY is a suite of products, designed to efficiently handle the different phases of the trading alpha generation workflow process, from data acquisition to alpha discovery and from back-testing to production. Its foundation layer provides a powerful Application Programming Interface (API) to build computerized quantitative trading systems. QuantFACTORY handles all asset classes (futures, forex, equities, bonds etc.) and runs multiple models on different timescales.

2011 QuantHouse. All rights reserved. Reproduction of this report by any means is strictly prohibited. . prohibited.

21

Automated FX Trading

July 2011

DATACENTER SERVER Data acquisition and preparation reparation With its plug-in architecture, QuantFACTORY can communicate with any market data provider or in prov broker, and new plug-ins can easily be written to connect to new providers. ins Clients can import historical data and store real real-time data in the datacentre server to run backback tests. Clients can also store low latency market data, third party vendors data (e.g. Bloomberg, Reuters RMTD and IDC) and be connected to liquidity platforms (e.g. Currenex, Hotspot, EBS or Hotspot Trayport). The data capture configuration is managed through a web interface (instruments and he hrough timescales). QuantFACTORY can handle different timescales, ranging from daily to intraday bars, different tick-data, best bid-offer and order offer order-book updates or custom data. DEVELOPMENT PLATFORM Alpha discovery The QuantFACTORY development platform is a Visual Studio add-in and as such benefits from its powerful IDE. It provides several tools to ease the model development process. This combination allows clients to manage referential and historical market data and to develop and test models. QuantFACTORY then analyzes the performance of clients models and exports them to a live exports execution platform. Develop, record and bask-test your alpha models test Within the same application, a list of instruments (equities, derivatives, bonds, swaps, multilegged instruments, FX, commodities e etc.) and trading rules (mono- or multi-asset) are available. asset) In addition, the Indicator Editor window contains a library with more than 60 listed indicators but clients can also customize and add indicators. The QuantFACTORY development platform is an event-based application, providing convenient and sophisticated ways of writing models in based provid any .Net language within the Visual Studio environment. Using the libraries and third party add-

2011 QuantHouse. All rights reserved. Reproduction of this report by any means is strictly prohibited. . prohibited.

22

Automated FX Trading

July 2011

ons (for MATLAB, R), clients can import mathematical and statistical functions that will be integrated in to the model. Debug and improve models To further enhance models, a debugging mode is included. Alpha models will run at a time step interval to trace internal events, signals and execution flow with high resolution, allowing clients to easily detect any bugs. Simulate model results Continue by choosing the timescales, parameters for different models and the historical data chosen to back-test models. Models will then be executed. The application has a range of different back-testing statistics views: Performance Summary for Curves and Indicators, Portfolio, Bar Chart, Global Trade Statistics and Equity Curve Statistics. Optimize models With an optimization procedure clients can define and test parameter values in order to obtain the best results and determinate the appropriate levels of risk. EXECUTION PLATFORM Produce and launch model To start trading, clients simply load their precompiled strategy component (generated in the development platform) into the execution platform, configure the adapter that will be used to connect the model to the market data provider and broker or EMS. In addition, clients can create their own alert conditions to warn about position items based on P&L value, set-up notification mode (auditory alerts, via email, color codes), and program actions. From there, clients can monitor orders and observe in real-time the full cycle of orders execution, from order sending to position updates in the model.

Q UA N T FE E D U LT R A LOW LAT E NC Y M A R K E T DATA


http://www.quanthouse.com/low-latency-market-data To enable our customers to manage the ever-increasing demand for low latency market data and to meet the changing requirements of todays trading environment with new trading venues, fragmentation of liquidity and rapidly increasing volumes of data, QuantHouse has developed an end-to-end product offering encompassing data capture within the exchange, ultra-fast data normalization and dissemination over QuantHouses proprietary fibre optic network. Since we design, implement and maintain every single element of the market data chain, we can offer our customers an ultra-low-latency solution, accessible through one single connection to our network, using a unique API, making it very easy to access any market using the same application.
2011 QuantHouse. All rights reserved. Reproduction of this report by any means is strictly prohibited.

23

Automated FX Trading

July 2011

Q UA N T LI N K U LT R A LOW L AT E N C Y T R A DI N G I N F R A ST R UC T U R E
http://www.quanthouse.com/QuantLink QuantLINK is QuantHouses trading infrastructure service to help buy side and sell side trading firms improve their infrastructures to keep up with market demand. Whether you use Smart Order Routing or algorithmic trading applications, your overall performance is linked to the quality of your trading infrastructures. QuantLINK combines QuantHouses proprietary fibre optic network interconnecting the heart of the exchanges with proximity hosting at the source.

2011 QuantHouse. All rights reserved. Reproduction of this report by any means is strictly prohibited.

24

S-ar putea să vă placă și