corporation is directly affected when it issues new stocks at the primary market. ...... using ERP systems such as SAP/R3, Baan, PeopleSoft or various workflow ...
Information Systems and IT Architectures for Securities Trading Feras T. Dabous and Fethi A. Rabhi School of Information Systems Technology and Management The University of New South Wales Sydney NSW 2052 Australia {f.dabous,f.rabhi}@unsw.edu.au
Summary The chapter gives an introduction to the basic principles of securities trading by presenting some essential concepts such as financial instruments, order types and market structures. It describes the securities trading cycle as consisting of a number of business processes within and across different institutions such as brokerage houses, investment banks, and markets. It classifies the different types of information and transaction systems and their role in the trading cycle. For illustration, the chapter focuses on two commercial systems in terms of their architecture. The chapter also examines the architecture of such systems and surveys the different levels of integration and enabling technologies that are currently being used. We also discuss current trends in utilizing service-oriented computing, business process automation and integration principles in order to achieve the development of stable, flexible and extensible business processes. Keywords securities trading, financial markets, capital market systems, service-oriented computing, business process integration, IT architectures
Introduction Electronic finance (e-finance) is mainly concerned with the automation of traditional activities in the finance domain and their associated processes. In particular, trading has become more and more a global activity because of the recent technological developments that have facilitated the instantaneous exchange of information, securities and funds worldwide. Trading decisions have also become more complex as they involve the cooperation of different participants interacting across wide geographical boundaries and different time zones. Although traditional communication
2
media (such as phone, email and fax) are steadily being replaced by automated systems, there are still many areas that require human intervention. The evolutions in IT middleware and integration technologies (particularly those based on the Internet) are providing endless opportunities for addressing such gaps (Banks 2001). One of the first challenges for software architects and IT professionals is getting sufficient knowledge about existing practices, business processes systems and architectures. This chapter should be primarily regarded as a contribution to the understanding of the domain of capital markets trading by illustrating some of its associated architectures, technologies and systems. The remaining part of this chapter is organized as follows. Section 2 introduces financial markets in general and reviews the concepts of capital markets structures, trading instruments, and trading order types. Section 3 discusses the information flow across the trading cycle in a capital market. It also addresses different types of information systems that are in use across such a cycle. Section 4 provides a closer look to the software architecture of two selected systems. Section 5 discusses an integration model that illustrates the different levels of abstraction currently being used in capital markets information systems. Section 6 concludes this chapter.
Financial markets: Introduction and terminology Introduction to financial markets Financial markets have the role of facilitating the movement of funds among people and organizations. The surplus units (e.g. funds suppliers) are seeking to generate returns (profits) by supplying their funds to the deficit units (e.g. the borrowers). Financial markets have evolved in such a way that enhances efficiency with which trades may take place and maximizes the fairness of trading in order to attract investors. The evolution of financial markets is a continuous process in response to changing needs, new technologies and changes in public policies. Financial markets are constantly introducing new financial instruments with various levels of risk management that equip the market participants with more confidence and risk management to conduct trades. Financial markets are classified into primary and secondary markets. The former is used by deficit units such as businesses, governments and households to raise funds for the expenditure of goods, services and assets.
3
The latter is used by surplus units such as the investors to trade what has been bought in the primary market. Financial markets are also classified into money markets and capital markets (Viney 2000). Money markets bring market participants together to trade market supported instruments offering high levels of liquidity. Australian money market participants include banks, superannuation funds, money market corporations, fund managers, building societies, credit unions, cash management trusts and companies. Money market instruments include exchange settlement account funds, treasury notes, intercompany loans, interbank loans, etc. On the other hand, capital markets have the primary responsibility for the operation of both its primary and secondary capital markets. Companies that need to raise capital by offering new stocks use the primary market. These new stocks can include initial public offerings (IPOs), right issues, placement and company options. Once those newly issued stocks are sold in the primary market, they can continue to be traded through the secondary market, called the stock exchange market. Capital market participants include traders (e.g. investors), intermediaries (e.g. brokers), statutory authorities (e.g. securities commissions) and third parties such as commercial information vendors (e.g. Bloomberg or Reuters). The market structure is determined by the trading rules and trading systems used by that market. These rules and systems determine who can trade, what to trade, when, where and how they can trade. They also determine rules governing market information dissemination to traders and other market participants (Harris 2003). Understanding market regulations, mainly trading rules and systems, is crucial for conducting trades. Normally, trading strategies that are applied to one marketplace are not applicable to most other markets that have different regulations and structure. Such differences are motivated by different understanding and research outcome on how to facilitate fairness and efficiency of a marketplace. A framework for market characterization has been defined in (Aitken et al. 1997). The framework has six crucial elements which are: technology, regulation, information, participants, instruments and trading protocol. It also presents four characteristics to achieve market fairness and efficiency. These characteristics are liquidity, transparency, volatility, and transactions costs. Market organizers have the role of achieving the best combination of the six factors in such a way that satisfies required levels of a fair and efficient market. In the remainder of this section, we illustrate
4
key elements in markets microstructures that are: trading instruments, types of trading orders, and types of markets microstructures.
Capital markets trading instruments Capital markets have evolved by introducing a number of trading instruments that diversify investment opportunities and attractiveness. These instruments can accommodate different trading styles for investors. They range from trading physical stocks (i.e. shares) to contracts on the willingness to trade stocks on a future date at a price determined today. Equity market The equity market, commonly known as sharemarket, plays the role of both primary and secondary capital markets. It is the place where financial assets are bought and sold. These assets can be the new stocks issued at the primary market or existing stocks that are currently traded at the secondary market. Secondary market transactions do not directly affect the cash flow of the corporation whose shares are being traded. The cash flow of that corporation is directly affected when it issues new stocks at the primary market. However; it has already been suggested that the existence of a well-developed secondary market is of a great significance to a corporation that may be seeking to raise capital in the primary market (Viney 2000). The existence of an active, liquid, well-organized market in existing shares adds to the attractiveness of acquiring new shares once advertised in the market. Fig. 1 (based on (Viney 2000)) summarizes the flow of capital and ownership rights of stocks in secondary capital markets in Australia while showing the participants role in that process.
5
Fig. 1. Roles of secondary capital market participants and the flow of funds in Australia
Derivatives market The derivatives market provides instruments that focus on risk management (interest rate, foreign exchange and price risks). These instruments such as options and futures have their value derived from the underlying security in the equity market or the value of a specified index for the index based derivatives. The unit of derivative trading is a contract. Once the contract is written, it can be traded at the derivatives market at a price that depends on the underlying stock price movement or the underlying index value. Each contract is a variant of an obligation or no obligation to trade (sell/buy) the underlying stocks at a predetermined price before a predetermined date or at predetermined date at some cost or no cost. Derivatives investors should be obligated to pay or receive margins during the life of the contract on timely bases. Organized derivatives can also be based on an equity market index (index derivatives) that gives the investor exposure to a certain market index. The advantage here is the exposure to a broad range of securities comprising an index rather than being limited to one particular company. Index derivative are normally cash settled.
Types of orders There are a number of standardized orders that can be used by investors to inform their brokers about the order specification and/or by traders or brokers to inform the market about the intension to sell (ask order) or buy
6
(bid order). Some markets may not support certain types of orders. In this case, brokers may offer services to their clients. Brokers may support a more complicated order specification service. In both cases, the broker service would imply the execution of the order as soon as all its requirements and conditions set by the trader are satisfied. Harris (Harris 2003) has presented a number of common orders’ types some of which are: Market order is an instruction to trade at the current available best price. This type of order is traded immediately but the trading price is not fixed before trading. Limit order is an instruction to trade at a specified price or better. The price of bid limit order is normally equal or less than the current best bid price, while the price of ask limit order is normally equal or greater that the current best ask price. This order is only trader when the market price moves towards it. Any order can carry additional information such as validity, expiration, etc. Some of the validity and expiry instructions are: open-order, dayorder, good-this-week, good-this-month, good-until, immediate-orcancel, fill-or-kill, good-after, market-on-open, market-on-close. Quantity instructions may also be included such as all-or-nothing or minimum-or-none orders. Other instructions may also be attached such as hidden-volume (undisclosed) order and substitution orders.
Classifications of marketplaces There exist two main types of trading sessions: continuous and call. Continuous session allows brokers and traders to place orders at any time during the trading session and the exchange executes trades as soon as a bid/ask match occurs. On the contrary, in call session, brokers and traders can also place orders at any time during the session until the market is called. A classification of market microstructures (Harris 2003) shows the following types of markets: Quote-driven dealer markets: in this market type, the dealers participate in each trade by quoting the bid and ask prices once requested by brokers or traders. Traders request quote and negotiate with a number of the security's dealers and normally select the dealer with the best quote that suites their order. If the submitted order is not matched immediately, then it is processed according to the type of that order and its routing instructions, for instance, it may be cancelled or awaiting as an outstanding order. Matching outstanding orders can occur through a
7
negotiation process in which the dealer moves his offered price to match that outstanding orders and/or the trader updates his order price to match the dealer price. NASDAQ, London stock exchange (LSE) and Reuters200 foreign exchange trading system are examples in which quote-driven market is the dominant structure organized by a dealer association and an exchange. Order driven (auction) market: include oral auction (open outcry), single-priced auction, continuous rule-based two-sided auction and crossing networks. In this type of markets, trader orders for any particular security are centralized in a single orderbook where buyers are seeking the lowest price and sellers are seeking the highest price. An order-driven market uses order precedence rules to match buyer to sellers and trade pricing rules to price the resulting trades. The primary order precedence rule at all markets is the order price where the higher bids and the lower asks have higher priorities for matching. Other precedence rules vary for different markets and used when a number of orders have the same price. These rules include placement time where the earlier submitted orders have a higher priority, order size where smaller/larger order volume has higher priority, public order precedence where public traders have higher priority than floor traders and display precedence where disclosed orders have higher priority than undisclosed orders. Continuous auction markets, which maintain a real-time outstanding orderbook for all securities, are used in most stock exchanges such as ASX continuous trading session. Brokered market: in such markets, brokers take the role of the exchange order matching under certain circumstances which is normally called crossing. Brokers’ role is to search for a counterparty trader(s) for a trading request. This is normally occurs for very large trading orders which is called block orders. In dealer market, the dealer sometimes refuses to conduct risky trades due to expected abnormal price movement or in illiquid securities. In auction markets, block orders could have an impact on the market price especially if that security is illiquid that may cause the market price to move away from the quoted price before matching that block order. The ASX allows brokers under some conditions to arrange trades for block orders through in-market and off-market crossing trading rules. Hybrid markets: combine characteristics of the previous three market types may be presented. Normally, the market structure features of one market type are the dominant while some other features from other markets types are embedded. The NYSE, which is classified as an order-driven market, requires the specialist dealers to offer liquidity if
8
no one else does that. The NASDAQ, which is classified as a quote driven dealer market, requires its dealers to display or execute the limit order.
Information systems for securities trading This section describes the information flow across the securities trading process. It also illustrates the different categories of information systems that can be utilized across such a process.
Securities trading information flow Market information includes real-time trading data and events such as orders, trades, quotes, indexes and announcements. It also includes historic information such as past trading data and companies’ periodic reports. Additional statistical data and analytics may also be available; however, participants such as brokers can generate their own analytics based on raw historic market data. As mentioned earlier, market information dissemination is very crucial to the fairness and efficiency of a capital market and hence contributes to its attractiveness to traders and encourages companies to be listed on the market. Fig. 2 illustrates the overall information flow across the trading processes for an order-driven market structure which is derived from the way the Australian Stock Exchange (ASX) (www.asx.com.au) operates. It comes in the shape of a continuous cycle in which we identify five phases: pre-trade analytics, trading, post-trade analytics, settlement and registry. In the following is a description of each trading phase.
9
Pre-trade Analytics
Buyer trader
Seller trader
Corporate
Exchange Trading
Secondary Market
Sell orders
Broker-B
Buy orders
Post-trade Analytics
Primary Market
Broker-A New Issues (underwriter holding)
Legend: Historic data flow
Corporate registry
MICROSOFT CORPORATION
Trades info flow Settlement / registry info
Settlement
Registry
Orders info flow
MICROSOFT CORPORATION
$
$
Buyer bank acount
Broker-A
Broker-B
Seller bank account
Fig. 2. Flow of trading information within the trading cycle
Pre-trade analytics phase The trading cycle starts with a pre-trading analytics phase that is normally conducted by the brokerage firms. In this phase, analysts use real-time information (trades, orders, volume, etc.) and correlate it with company announcements and historic information (including periodic company reports) to predict price movement with a degree of certainty. Consequently, the purpose of this phase is to support buy or sell decisions in order to gain capital profit or prevent losing capital respectively. Most brokers provide the outcome of their analysis as services to their clients (investors) and extra fees normally apply. Investors can seek historic information, delayed real-time market information and announcements from the capital market web site or from third party such as commercial information vendors e.g. Bloomberg (Bloomberg.com) and Reuters (www.reuters.com). These services are offered free or for some fee that depends on the level of detail and real-time accuracy.
10
Trading phase The trading phase basically implements a market structure as discussed in the previous section. Buyers and sellers both signal their intentions in the form of orders (different types of orders were also discussed earlier) and different markets have different methods for determining trades depending on the order type, the market structure and the degree of the trade process automation. At the heart of the trading system is the trading engine (TE). Post-trade analytics phase A capital market attracts investors by maximizing the fairness and efficiency of trading. Post-trade analytics focuses of detecting illegal trading behavior based on the available real-time and historic market information. Post-trade analysis can also use this information for other purposes such as analyzing past and current market behavior and trading patterns and predicting future behavior, etc. A capital market surveillance department is an example of a place for conducting post-trade analytics. The major responsibility of a surveillance department is to detect illegal trading incidents and to generate real-time alerts for such incidents. A surveillance department must have real-time access to every piece of market information across all stages in the trading process such as order and trading data, settlement and registry databases, possibly brokerage firms’ databases and investors’ holdings and personal data. Due to information privacy measures that are enforced, surveillance departments may not be able to access some of this information. Instead, they are partially coping by implementing intelligent and smart techniques to detect illegal trading behavior. Because of heuristic nature of these techniques, surveillance is in constant need for improvements. Settlement and registry phases Settlement and registry is a transactional-based process for transferring securities from seller’s profile to buyer’s profile and transferring money from buyer’s account to seller’s account. The transfer of both securities and money is simultaneous and irrevocable once the settlement process is started. The registry phase completes the settlement process by modifying the ownership of shares in the company's registry from the seller to buyer. It is possible that a number of brokers are involved in one single trade, this happens when, for example, in equity market one large bid order is matched with a number of smaller ask orders. The settlement process is complicated since it involves transferring money and ownership of securities between the different parties involved.
11
Categories of capital markets’ information systems We now take a close look at the different types of computer-based systems that participate in the securities trading process. These systems include those that facilitate information dissemination to market participants and contribute to market’s transparency. Other systems facilitate order routing to and from the exchange and across the trading cycle. Remarkably, each phase in the trading cycle is supported by one or more computer information system. In the following classification (based on (Harris 2003)), one system may perform tasks related to more than one category. Information collection and distribution systems These systems have replaced the clerks (market reporters) who used to observe and record all trading activities. Information collection includes reordering all market activities such as orders, cancellations, amendments, trades, quotes and announcements. Most exchanges provide real-time market feeds that can be accessed by outside systems. The distribution of market information is directly related to market’s transparency. Distribution can occur in real-time or delayed basis governed by market regulations and the type of requester. Most markets provide delayed quotes and companies historic information as a free service for public. Real-time services are only available registered members such as brokers and data vendors. Data vendors, such as Bloomberg and Reuters, often support query and broadcast services of its customers. Query services provide information on demand while broadcasting provides continuous streams of information as it is received from the participant exchange. Broadcast services include trade reports (ticker tape) and/or quotes (quotation feeds). Broadcasting delayed quotes service in normally available free on the watch board of the market while query services of delayed quotes are also available for free on the Web sites of markets or data vendors. Trading engines The trading engine (TE) is at the heart of an automated trading system. The main function of the TE is to match bid/ask orders using order precedence set by the market regulation. In an order-driven market such as the Australian Stock eXchange (ASX), the TE implements a number of bid/ask order-matching algorithms that depend on the security concerned, the current trading phase (opening, continuous, closing etc.) and the type of order. Once a security's best bid price matches its best ask, an
12
irreversible trade is executed. In-house developed trading engines are often the choice by many large markets such as SYCOM in Sydney Futures Exchange (www.sfe.com.au), SuperDot in New-York Stock Exchange (www.nyse.com) and SuperMontage in NASDAQ (www.nasdaq.com). Other markets choose a third-party trading engine. For example, SWX Swiss and Shanghai Stock Exchange use OMX’s X-STREAM trading engine (www.omxgroup.com) which will be described in more detail later in this chapter. Another example is OMX’s CLICK trading system that is utilized in a number of equity options markets such as the International Securities Exchange and the ASX. Order routing systems Order routing systems transmit orders between the appropriate market participants and systems. Traders use them to send orders to the broker or dealer. Brokers use them to transmit orders to a dealer or an exchange and to report back to the trader about the status of the order. Dealers and exchanges use them to report trades to the brokers and/or traders and to the registry and settlement systems. In ASX, order routing systems are utilized by the SEATS message based Open Interface (OI). Order routing systems of Australian brokers such as IRESS IOS (www.iress.com.au) are conformant with the SEATS IO. American ITS (Intermarket Trading Systems) facilitate trading across different equity markets in US. Some Electronic Communication Networks in the US are participants to order routing systems. Order presentation systems Order presentation systems present information about orders and quotes to the market participants (e.g. brokers). In order driven markets, this information is presented on traders screen connected to the screen-based trading system. Brokerage firms support their own presentation/visualization systems that receive feeds from ORSs. For example, IRESS system has its own facility for order presentation to the brokers whereas SMARTS (to be described later) has two subsystems called SPREAD and REPLY which support orders visualization facility to market researches and surveillance personnel. Dealers markets often use messaging systems that allow participant to communication via messages.
13
Settlement and registry systems Settlement and registry systems facilitate the registration of the trade volume to the buyer and transferring the trade value to the seller. We often distinguish between settlement and registry systems. Settlement is about the transfer of money in return for title and the legal title is often (not always) then held in what is known as a depository. The settlement and depository is often done in the same organization which are in general called “CSD” – central securities depository. The registry system is really a duplication of this information but is a specialized function because its role is to facilitate the communication between a company and its shareholders (such as voting, annual meetings, distributing annual reports and dividends etc) and is becoming an increasingly important role given that issues of corporate governance are becoming more prominent. In the ASX, CHESS is the official settlement system and depository that acts upon receiving trades from SEATS with the cooperation of the trade participating brokers. CHESS is capable of communicating with issuer sponsored registry if one of the trade counterparties is not CHESS sponsored. Surveillance systems Market surveillance is facilitated by computer systems that try to detect illegal market behaviors that violate the market regulations. Surveillance systems should have access to every piece of real-time or historic market information. For example, the Financial Services Authority (FSA) (www.fsa.gov.uk) in UK is tendering for a supplier to upgrade its SABRE surveillance system. SMARTS (www.smarts-systems.com) is a universal surveillance system that is used in more than 16 countries.
Case studies This section presents two selected systems that participate in the securities trading cycle, namely the SMARTS surveillance system and the XSTREAM trading engine.
SMARTS SMARTS (www.smarts-systems.com) is a securities markets surveillance and information system that provides easy-to-use alert management,
14
reporting and research capabilities for exchanges. The system allows accurate alert algorithms to be created. SMARTS is a software tool which originates from the research labs at Sydney University and was developed under the direction of Prof. Mike Aitken and two PhD students Thomas Jones and Tim Cooper. After the involvement of Computershare Ltd in 1996, the tool changed its focus from a research tool into a surveillance and compliance product. SMARTS is now used in many stock exchanges including Jakarta and Hong Kong, SWX Swiss Exchanges, ASX, Tokyo Commodities Exchange, SE of Thailand, Singapore, Moscow. The SMARTS architecture comprises the following important elements: 1. A number of application components that perform surveillance and research functionalities on real-time and historic financial data. 2. A data model called MPL which stores all the market real-time and historic information in an optimized manner 3. A number of software modules (called engines) for the creation and manipulation of MPL data. The most commonly used are the tracking and alerting engines. 4. The Alice language that is utilized to specify the rules and conditions that can cause the issuing of alerts. Application components SMARTS comprises a number of application components (or modules) which carry out predefined and autonomous tasks on behalf of users. There are different types of components. The surveillance-oriented components are: ACCESS (benchmark visualization tool depicting the normal behavior of trading metrics representing liquidity, volatility, transparency and transactions costs), REPLAY (replays historical order and trade data), SPREAD (an interactive graph of trading behavior on price and time axes), ALDIT (Alice language editor), and ALMAS (Alice Management System), COLLUDE (depicts trading activity between brokers). Besides real-time detection of illegal behaviors in capital markets (e.g. insider trading and market manipulation), SMARTS can also be utilized as a platform for capital market analysis and research purposes. The analysis/research oriented components are: EVENTS (Market research), FACTS (Market statistics report), INFORM (Information Editor for news such as company announcements), PORTFOLIO (visual representations of trading metrics), and REPORT (lists of transactions, such as trades and orders, pertaining to a single or multiple security, house, trader and/or client).
15
SMARTS MPL data model The MPL data model manages all information relevant to SMARTS application components. As Fig. 3 shows, there are two types of data: transient data which is held in the main memory and persistent data which is held on disk. The MPL data model can manage and process real-time transactions as they arrive from the trading engine. It also provides the required infrastructure for monitoring multiple markets (either different trading engines or different types of markets such as options and futures). The MPL data model manages the following market relevant data and concepts: trades, orders, quotes, historic indexes, interest rates, dividends and stock splits, static data such as companies names, shares-on-issue, index weighting, special fields for options, futures, bills and bonds, broker house or trader cash position, inventories of holdings, and news announcements. MPL data model Transient data (main memory)
Persistent data (disk storage)
Shared memory RAM
Track data (FAV files)
Full orderbook
AM
Temporary memory
Name
Daytot
LAN MPL manipulation components
Application components
Tracking Engine
Alice Program Interpreter
Alerting Engine
External systems Trading engine feed
Third part software
Fig. 3. SMARTS architecture
One feature of MPL is that it relies on new data (i.e. market transactions) being added to track data (FAV format) in append mode only. The “never delete anything” feature is needed for storing a full history of events so that, at any time, a surveillance department can regenerate the exact conditions as they occurred in the past for research and investigation purposes. The concepts of “append-mode only” and “never delete any data” allow creating redundant data in different formats and indexing for performance needs such as AM data (indexed by security and contains information such as best bid/ask, trade info, security status control info and
16
liquidity info for each trading day) and Daytot data (daily summaries indexed by date including open price, close price, max price, min price, average price, and turnover). MPL comprises a set of APIs that can be invoked from SMARTS application components, C programs or through the Alice language commands. At the lowest level, MPL data is stored using binary proprietary formats that provide efficient access and storage conditions. MPL manipulation engines The alerting engine is the most significant part of any surveillance system. This engine implements the specialized alerting language Alice. Once the alerting engine is started (by the “alertserver” program), it interprets all text alert rules from the companion Alice code into a convenient data structure that will direct the alerting engine processing for that trading day. Normally, the alerting engine examines those alerts rules that are activated by the type of the FAV message that has been processed or on a timely basis. The Alice language SMARTS uses the in-house built programming language Alice for coding and creating alerting rules based on historical or real-time trading data. It also used for generating reports based on historical or real-time trading data. Alice programs make it clearly possible to separate SMARTS functionalities from the alerting rules. Alerting rules (i.e. Alice program) can be modified at any time to incorporate new or modified rules without the need to recompile SMARTS. The alerting engine interprets the contents of the respective Alice program to build the memory structure that represents those rules. Alert engine then uses this structure to control its behavior on when (following which events) and what rules to be check. Event can be the type of trading event (i.e. FAV message type) that is currently being processed. It also can be on timely basis (e.g. every 5 minutes). A converter program needs to be written for every trading engine that SMARTS connects to. This conversion program generates FAV formatted messages out of the trading engine output (trades, orders, control info, announcements, etc.). Information from more than one trading board can be combined in a single FAV. Indeed, output of the trading engine can form more than one feed to SMARTS where each feed data is collected in a separate FAV file. Normally, order/trade/control data compose one feed,
17
announcements (news) compose a second feed and index data composes a third feed.
X-STREAM X-Stream, previously called ASTS (Automated Securities Trading System), is a system built and commercialized by Computershare (and recently purchased by OMX of Sweden). It is primarily designed as a trading engine and market management system but it is also able to keep track of users’ accounts, forward market information, and provide a personal instant messaging service to users on the system. X-Stream can be customized for different markets and system configurations and is one of the most widely used in the world. Orders and Instruments There are a large number of different types of instruments that are tradable with X-Stream. X-Stream was developed to work seamlessly with securities, options, futures, bonds, swaps, commodities, foreign exchange etc. As such there must be a defined standard for how X-Stream treats these different types. With these different types of instruments comes the ability to trade in different ways. An order is an “Instruction to a broker/dealer to buy, sell, deliver, or receive securities or commodities that commits the issuer of the “order” to the terms specified1. The X-Stream Architecture The X-Stream software architecture is broken up into several components. Fig. 4 depicts the organization of these components, described next.
1
Computershare Technology Services PTY LTD. Programmer’s References for ASTS Client SDK. Version 1.6. 2002
18
Relational Database Server
Trading Engine
Relational Database Server
Trading Engine
Log Files
Log Files
LAN
Governer Trader Workplace
Gateway
Client Developer Software
Fig. 4. X-Stream Architecture
Trading Engine The Trading Engine is the main transaction component and its main function is to take orders and process the order book. It also manages all the user accounts on the system. On start-up, the Trading Engine requests all of the market and system data from the relational database server. It then stores this data in log files updated in real-time. If the Trading Engine ever crashes, then it can be restarted from data in these log files. Relational Database Server All of the parameters in X-Stream (such as the number and type of users, which securities to trade, security boards, currencies and etcetera) are held within the relational database server. The flexibility of the system is provided through the database server because all of the system parameters can be defined easily within the relational database. Backup Trading Engine The Backup Trading Engine processes the same orders as the Trading engine. In the event of a failure of the Trading Engine, the Backup Trading Engine can be used as the primary trading engine transparently. The
19
Backup Trading Engine starts up after the main Trading Engine, so it simply downloads the market information from the Trading Engine’s log files. In all other respects, the Backup Trading Engine is the same as the main Trading Engine. Governor The Governor handles the fault management of the whole system. After an initialization protocol with the Governor, the two trading engines stay synchronized. Upon failure of either service there will be a breakdown in communication and both services will attempt to contact the Governor, and the one that is successful becomes the only trading engine. Gateway As there could possibly be many client machines connected with the Trading Engine, the Gateway component is used to load balance the different transactions and queries and to distribute all the market data to the clients. The Gateway processes all the same transactions as the Trading Engine so that it maintains a complete picture of the market. Trader Workplace The Trader Workplace is the user end interface developed by Computershare to the X-Stream software. It has different functions depending on which user is logged into the terminal. Its main priorities are to display market information, to submit orders, allow communication between brokers, and to submit market sensitive information and to control and operate a market. Client Developed Software X-Stream also comes with a complete software development kit, called Comet SDK, for interfacing with all aspects of the system. The SDK was developed as a complete way to interface to the system. Any client that wishes to develop or modify their own Trading Engine’s client can easily interface through Comet. The main usage of the SDK is to interface to the Gateway as a client application, however, because of its extensiveness; anyone can connect to any part of the system. It allows enough flexibility for developing a different Gateway or Governor. Software components communicate with each other using the AMP Messaging Protocol, which is X-Stream’s main messaging protocol for handling all transaction and
20
query information and is formally described in the message definition language ASN.1 (ISO/IEC 8824-1:2002).
IT integration architectures and technologies We now turn our attention to the middleware technologies that are used to facilitate the integration between the different systems described earlier. As these technologies intervene at different levels of abstraction, we first define a layered conceptual model that will be used to classify the different technologies involved. Services Trading
Trading data
Broker
Surveillance
Settlement
...
Content (data) level: FIX, FIXML, XML, FIN, FAV, etc.
System X
System Y
Business Process level: Placing a large order, Determining a trade signal, etc.
Transport level Network protocols, Remote Procedure Call, APIs, etc.
Fig. 5. Conceptual integration model for middleware technologies
Fig. 5 shows three possible levels of integration between two different systems X and Y. In the upper level, the business process level is the highest level in the hierarchy. It offers business services in a way that “hides” the underlying data formats or middleware platforms. It represents the appropriate level for defining integrated services that correspond to particular business models. The content (data) level defines the structure and format of messages between the software components. The transport level defines the interfaces of communication between the various software components. We now present a brief review of existing technologies classified in relation to this model.
21
Transport level For security reasons, most financial systems traditionally interact with each other using private or dedicated networks. An example is the SWIFT Transport Network (STN) based on the X.25 protocol. As communication software is non-standard, this is a very expensive solution. Other than for historical reasons, there is less and less interest in integration at this level because of the low level nature of the corresponding programming interfaces. Even private networks are now moving towards IP-based solutions (Intranets and Extranets). For example, SWIFT has developed a new generation of interactive services (called SWIFTNet) using an IPbased network called the Secure IP Network (SIPN). Besides loosely coupled communication by means of messages and file transfers, systems can also interact with each other in a tightly coupled way using dedicated Application Programming Interfaces (APIs) accessible through the network protocols. A number of possibilities exist for such interactions such as the use of remote procedure calls and remote objects.
Contents (data) level The content layer provides languages and models to describe and organize information which enables the participating applications to understand the semantics of content and types of business documents. The objective of interactions at this layer is to achieve a seamless integration of data formats, data models, and languages (Medjahed el al. 2003). The two major enabling technologies at this layer are Electronic Data Interchange (EDI) and eXtensible Markup Language (XML)-based frameworks. Electronic Data Interchange (EDI) (www.edi-information.com) is welldefined and well-established in the industry in the sense that it provides both the defined syntax and defined vocabularies for values of the EDI message fields (Bussler 2001). Trading partners exchange business documents via value-added-networks (VAN) using EDI standards. EDI is an established technique for B2B integration that focuses on interactions at transport layer where VANs are used to handle message delivery and routing. EDI standards also provide a single homogeneous solution for the interactions at the content layer where the set of supported document types is limited. In addition, EDI standards, as currently defined, do not support interactions at the BP layer. EDI has been in use long time before the emergence of the Internet. It depends on a moderately sophisticated
22
technology infrastructure based on proprietary and expensive VANs and ad hoc development that needs direct communication between partners. Therefore, it remains unaffordable for the majority of the business community. The emergence of the Internet as a global standard for data exchange has influenced EDI to expand in many directions. The Open Buying on the Internet (OBI) (www.openbuy.org) is an initiative that leverages EDI to define an Internet-based procurement framework thereby utilizing the free use of the Internet instead of the expensive VANs. In parallel with XML developments, a number of international organisations were already working towards the adoption of common standards in the financial area. One of such standards is the Financial Information eXchange (FIX) (www.fixprotocol.org) protocol for communicating securities transactions between two parties. Since FIX does not define how data is transported, an XML version of FIX called FIXML (www.fixml.org) has been defined. FIXML facilitates real-time delivery of messages and it supports multiple currencies and financial instruments types. It also specifies how the trading orders are advertised, acknowledged, placed, rejected and executed. It also can be used to manage settlement instructions between the trades’ parties. Another standard is the Financial Products Markup Language (FpML) (www.fpml.org) supports business information exchange standard for electronic dealing and processing of financial derivatives instruments. It establishes a new protocol for sharing information and dealing in financial derivatives over the Internet. The FpML is currently focuses on partial derivatives instruments such as interest rate swaps and forward rate agreement. Research Information Exchange Markup Language (RIXML) (www.rixml.org) is an initiative for developing an open industrial standard to tag and deliver investment research. It provides a structure for classifying financial research in a way that will enable users to define specific sorting, filtering and personalization criteria across all research publishers. In each case, a working group defines a Data Type Definition (DTD) that defines the XML tags and how they can be legally combined. These DTDs can then be used with XML parsers and other XML tools to rapidly create applications to process and display the stored information in whatever way is required. Other related standards include ISO15022 (www.iso15022.org) which defines a standard for messages in banking, securities and related financial services. This standard is suitable for inclusion within an XML format. Another standard is SWIFT’s messaging service, FIN, which runs on its private network (STN) and enables financial institutions to exchange standardized financial messages worldwide.
23
Business process level The business process layer is concerned with the conversational interactions among the activities that comprise a BP. The objective of interactions at this layer is to allow autonomous and heterogeneous partners to come online, advertise their terms and capabilities, and engage in peer-to-peer interactions with any other partners (Medjahed et al. 2003). Interoperability at this higher level is a challenging issue because it requires the understanding of the semantics of partner BPs. A few approaches have emerged for the purpose of facilitating BP interactions. We focus here on two popular ones, namely Workflows and Web Services. Workflows Current BPs within an organization are integrated and managed either using ERP systems such as SAP/R3, Baan, PeopleSoft or various workflow systems like IBM's MQ Series or integrated manually on demand-basis. There are many Workflow Management System (WFMS) products available each of which proposes a different specification that makes it hard to replace and ineffective when considering the needs of B2B interactions (Medjahed 2003). For this purpose a workflow reference model has been defined (Hollinsworth 1994) which motivates some standardization efforts such as jointFlow, the Simple Workflow Access Protocol (SWAP) (Bolcer and Bolcer 1999) and Wf-XML message set (www.wfmc.org). However these efforts provide little support for interenterprise BPs. The purpose of inter-enterprise workflows is to automate BPs that interconnect and manage communication among disparate systems. An Inter-Enterprise Workflow (IEWf), is a workflow consisting of a set of activities that are implemented by different enterprises, that is, it represents a BP that crosses organizational boundaries. Delivering the next generation of IEWf Systems is having potential attention with the objective to interconnect cross organization BPs while supporting the integration of diverse users, applications, and systems (Yang and Papazoglou 2000). However, no current approach seems to be extensively used in the financial sector. Web Services The Web Services model (Alonso 2004) is emerging steadily as a looselycoupled integration framework for Internet-based e-business applications. Unlike traditional workflows, Web services support inter-enterprise workflows where BPs across enterprises can interact in a loosely coupled
24
fashion by exchanging XML document-based messages. The Web Services model idea is to break down web accessible applications into smaller services. These services are accessible through electronic means, namely the Internet. They are self-describing and provide semantically well-defined functionality that allows users to access and perform the offered tasks. Such services can be distributed and deployed over a number of Internet-connected machines. A service provider is able to describe a service, publish the service and allow invocation of the service by parties wishing to do so. A service requester may request a service location through a service broker that also support service search. There are four emerging de facto standards for Web Services. The first one is Simple Object Access Protocol (SOAP) (www.w3.org/SOAP) which is an XMLbased protocol for exchanging document-based messages across the Internet. The second one is Web Services Description Language (WSDL) (www.w3.org/TR/wsdl) which is a general purpose XML-based language for describing the interface, protocol bindings and the deployment details of Web services. The third one is Universal Description, Discovery and Integration (UUDI) (www.uddi.org) which refers to a set of specifications related to efficiently publishing and discovering information about Web services. The last one is Business Process Execution Language for Web Services (BPEL) (www.siebel.com/bpel) that has emerged based on a combined effort in merging the Microsoft XLANG and the IBM WSFL. This Web services model is not matured enough for wide acceptance, yet it tries to leverage the three layer interaction framework. At the transport layer, SOAP is used as a messaging protocol. At the content layer, a WSDL based XML schema is used. At the business process layer, BPEL is yet to evolve to cope with complex inter-enterprise interaction issues. In a previous study (Dabous 2005), we have justified with examples that some of the functionalities of capital market systems can be reused to assist the automation of many of the domain business processes. Therefore, we consider each of such systems as consisting of one or more e-services that can facilitate the enactment of domain business processes. Such business processes are related to trading activities across the trading cycle discussed earlier. This can include but not limited to activities such as placement of orders, online settlement and registry once a trade is matched, real-time surveillance and real-time information dissemination (announcements, quotes, trades, etc.). In (Rabhi and Benatallah 2002), preliminary service oriented architecture for the integration of capital market financial services has been proposed. In this architecture, each software component provides either a basic service or composite (integrated) service. The former is where the service logic is totally provided by that architectural component. The latter
25
is where the architectural component interacts and requires other services (called outsourced services) to fulfill its role. Therefore, a service-based approach separates between business concerns (i.e. the use of the service) and technological considerations (i.e. the computing and networking infrastructure to support a service). Fig. 6 shows the essential basic and integrated services that are associated with securities trading. Very few capital market systems are integrated at this level. Therefore, the B2B integration at this level in capital markets domain is still an active research area. Exchange Service (ES)
Trading Data Service (TDS)
Issuer Data Service (IDS)
Registry Service (RS)
Settlement Service (TS)
Communication / Middleware Infrastructure
Real-Time TDS (RT-TDS)
Surveillance Service (SS)
Broker Service (BS)
Analytics Service (AS)
Fig. 6. High level service oriented architecture for securities trading
Conclusions The main purpose of this chapter is providing software architects and IT professionals with sufficient knowledge about existing practices, business processes, systems and architectures in the area of securities trading. It introduced the concepts of market structures, order types and described the information flow throughout the trading cycle and the different types of information systems that participate in this cycle. The chapter pays some attention to business process integration issues by reviewing relevant middleware technologies according to their level of abstraction. We have seen that most integration currently occurs at the transport and content layer. Heterogeneous technologies in development and middleware platforms together with the absence of dominant standards have complicated the achievement of an interoperable environment that facilitates the integration among different systems. There are many opportunities yet to be exploited for integration at the business process (BP) level. This is essential for two reasons. On one hand, we are
26
witnessing dramatic changes in computing, data communication and middleware technologies. On the other hand, the BPs themselves are changing as a result of regulations, global competition and new financial instruments or market structures. Future developments may witness the maturity of the emergent technologies especially at the BP level which will facilitate the integration of multiple markets worldwide.
Acknowledgements We wish to acknowledge the Capital Markets Cooperative Research Centre (CMCRC) for funding this work. We are also grateful to staff at SMARTS and Computershare for their support especially in providing information about their systems.
References Viney C (2002) Financial Institutions Instruments and Markets. McGrath’s Michael Aitken M, Frino A, Jarnecic E, McCorry M, Segara R, Winn R (1997) The microstructure of Australian stock exchange. Securities Industry Research centre of Asian Pacific (SIRCA). Harris L (2003) Trading and Exchanges: Market Microstructure for Practitioners. Oxford University Press Banks E (2001) e-Finance: The Electronic Revolution, John Wiley & Sons Dabous FT, Pattern-Based Approach for the Architectural Design of e-Business Applications, PhD thesis,The University of New South Wales, Australia Medjahed B, Benatallah B, Bouguettaya A, Ngu AHH, Elmagarmid AK (2003) B2B Interactions: Issues and Enabling Technologies,The VLDB Journal, pp 59-85, Springer-Verlag, Apr Bussler C (2001) B2B Protocol Standards and their Role in Semantic B2B Integration Engines. Bulletin of the IEEE Computer Society Technical Committee on Data Engineering, 24(1), pp 3-11 Hollinsworth D (1994) The workflow reference model, #TC00-1003, Brussels, Belgium Bolcer G, Kaiser G (1999). SWAP: Leveraging the Web to Manage Workflow, IEEE Internet Computing, 3(1), pp 55-88 Yang J, Papazoglou M (2000) Interoperation Support for electronic business, Communications of ACM, 43(6), pp 39-47, Jun, Rabhi FA, Benatallah B (2002) A Service-Based Architecture for Capital Markets Systems, 16(1), pp 15-19 Alonso G, Casati F, Kuno H, Machiraju V (2004). Web Services: Concepts, Architectures and Applications. Springer
