Chris Tham & Tim Posney
Originally published in: SunTech Journal Volume 3 No. 1 (Winter 1990) pp. 46-52
The treasuries of large financial institutions, such as banks, need to constantly control the risks associated with trading. Effective risk management depends upon accurate and timely information from all dealing desks.
There are various methods of minimising, or hedging, financial risks, depending on the type and nature of the asset or commodity being traded. This article discusses the issues and methods of managing risks associated with trading 'options', and provides an overview of the family of Optech Option Trading Systems. Optech systems are distributed systems running on Sun workstations, providing trading facilities with risk analysis and immunisation of option portfolios.
An option is a contract that gives the buyer the right, but not the obligation, to buy or sell a fixed amount of an asset at an agreed price (commonly called the 'exercise' or 'strike' price) for a fixed period of time. This is in contrast with other financial contracts, where the buyer is obligated to meet the terms of the contract. A call option is an option to buy and a put option is an option to sell. An asset is anything that is commonly traded, such as currency, fixed interest securities, company stocks, physical commodities or even other options. The option writer, or seller, charges a small amount, called the option premium, for the option.
The buyer may abandon the option, i.e., do nothing, losing only the premium. Alternatively, the buyer may exercise the option and enter into an agreement to buy or sell the asset at the agreed price.
An option can be purchased as a form of insurance policy, protecting the purchaser from adverse price movements. See Figure 1 for an example.
By buying an option, the maximum loss that the buyer can suffer from an adverse movement is the option premium. However, the purchaser can benefit from advantageous movements by abandoning the option.
Option purchases are also useful for contingent cash flow situations. In the example given in Figure 1, the exporter could buy the option before the sale of goods is finalised and still be protected from adverse price movements.
An option purchase is attractive to buyers with poor credit ratings, since they do not require prior credit approval for the underlying transaction, merely the ability to pay the premium.
Finally, options are attractive as speculative instruments when the price of an asset is expected to move in a specific direction. They have limited risk and are not generally subject to margin calls.
The option writer gains, at most, the option premium from a sale but risks unlimited loss, being exposed to the full extent of price movements in the underlying instruments. Large scale option writing by traders in bank treasuries will create a position with unacceptable risk unless methods are used to contain or minimise the risk due to the exposure. This is done by 'hedging' the portfolio through trades which are performed solely to minimise risks.
Ad-hoc or manual hedging methods typically involve the use of spreads, i.e., the simultaneous purchase and sale of similar options; or straddles, which involve the purchase of equal numbers of both call and put options of the same type. Spreads seek to use option purchase to reduce the risk of option sale, and vice versa. Straddles allow the option writer to be protected from market movement in either direction. However, the effective use of ad-hoc strategies such as these requires careful judgement on the part of the hedge trader, which can only be obtained from experience. The use of spreads and straddles will also result in a reduction in profit from the option sale.
Modern computerised hedging techniques involve the quantitative evaluation of risks, based on mathematical models of option pricing, together with continuous dynamic trading based upon monitoring of positions and a means of mathematically determining hedging trades.
Hedging relies on developing accurate mathematical models for the fair price of option premiums and hedging instruments. A hedging instrument is a financial contract used to hedge the sale of an option, which can be as simple as buying or selling a portion of the underlying asset. The fair price of a financial instrument is the expected value of the instrument if current market conditions prevail.
If we make certain assumptions about distribution of asset price movements, the fair price of an option premium is determined by five main factors:
These are the only quantifiable influences on the premium. The fair price is a function of these factors. In practice, non quantifiable factors such as market expectations will also affect the premium.
The writer is exposed to the risk of changing market conditions, that is, to the risk of any of the above factors changing from the time the option is sold to the time it is exercised.
From the pricing model, the partial derivative of the fair price with respect to any one of these factors can be derived giving a measure of the risk due to a change in that factor.
Risk immunisation consists of constructing a series of hedging trades such that the sum of the risk measures for the hedging trades are equal and opposite to the risk measures of the option. If the option and the set of hedging trades are traded as a group, the resultant portfolio will not be affected by market movement. We have immunised the portfolio against market movement, i.e., the portfolio value does not change appreciably for small movements in market values.
Risk immunisation only occurs for limited periods of time. Large or sudden market movements will cause the portfolio to lose immunisation and must be ‘rebalanced’ by constructing new hedging trades.
Risk management also entails performing sensitivity analyses of how a portfolio will respond to changes in market conditions and simulations of predicted or historical market movement provide an indication of the effectiveness of rebalancing.
Optech is currently in the process of developing options trading systems with advanced risk analysis and control facilities. The primary objective is to design a risk management system that is directly linked to trading floor activities. To achieve this goal, the following properties have been identified requirements in the trading or risk management system:
In order to fulfil the above requirements, we have decided upon a portable fault-tolerant distributed approach. The systems are written entirely in the C and C++ languages, with portability as a design goal. The software currently runs on Sun workstations under SunOS 4.0.3 but is designed to work in a heterogeneous NFS (Network File System) environment. Distributed processing is achieved by spreading activity amongst machines in the network. Replication of critical processes across the network provides fault-tolerant operation.
All database transactions go through a machine running the Sybase SQL database server. Normal file access is provided by networked file servers.
Two kinds of user interfaces are supported. Front-office trading and risk analysis is performed on bitmapped workstations using the X11 Window System. Back office trade modification, confirmation, reporting and maintenance is accessed from a screen oriented interface from inexpensive character terminals connected to a networked minicomputer or fileserver.
The Optech family of option systems are designed to be distributed across a network, hence each system does not consist of one monolithic executable but a group of related applications that work closely together in an integrated environment. Each application is designed to work by itself or in conjunction with other applications in the system, no matter where they are physically located on the network. Interoperability between systems is also designed to be as transparent as possible.
Inter-application communication is performed by the TCP/IP socket mechanism via the X Windows system for point to point communication, and through the Sybase SQL DataServer or remote procedure calls for broadcast communication.
The following applications have been developed for the Optech system:
This is the biggest and most complex application in the system, and runs on a graphics workstation utilizing the X Window interface.
It provides a trader with a window allowing on line trading to be performed. A trader can perform operations on the PTS such as entering option details and accessing live market data and using it to price options.
The PTS allows a trader to view the risk measures associated with a portfolio, and the effect a new trade will have on the risk measures. Hedging trades can also be suggested by the system for new option trades.
The PTS is built using advanced screen design tools, and is customised to the kind of options being traded. Figure 2 gives a relationship diagram showing how trading activity is centred around the PTS.
Figure 3 gives examples of trading screens currently under development.
The GLM is a daemon that continuously revalues and recalculates the risk measures of all portfolios in the Portfolio Database in response to changing market data. The GLM is designed as a transparently distributed application. Multiple instances of the GLM can be started across machines in the network and behave as one application. The GLM is also fault-tolerant as different instances can be started and stopped at any time.
This allows a user to determine exactly how the value of a particular portfolio and associated risk measures will change should the market move in a particular way. It allows the user to enter sensitivity analysis parameters and then view the results as a graph, on screen, or as a report on the printer. The user can alter up to two market parameters independently viewing the effects on portfolio value and risk measures.
This is an application that analyses the effects of rebalancing on portfolio value and risk measures when random or historical market movements are applied. Simulation results are effective for analysing the effectiveness of different hedging techniques. The user interface allows simulation results to be printed as a report or displayed on screen as a graph.
The Clipboard is designed for passing data between parts of the trading system as well as for storing data for later retrieval. Different types of data can be passed not only from one instance of an application to another on one machine, but also across machines of different architectures within the network. The clipboard window shows current clipboard entries and allows users to extract or post entries.
This application retrieves relevant market/economic data from information providers such as Reuters or MicroGnosis. It interfaces to a digital information channel, such as a Reuters RDCDF unit, and automatically updates the market information database. Page translation algorithms convert text screens into relevant data according to predefined formats. Screen pages are archived for future analysis and simulation purposes. Sophisticated error detection and recovery from hardware failure is provided.
The Report Daemon is in charge of generating and printing scheduled reports on an automatic or semi-automatic basis. It is an application that allows reports to be scheduled and printed at certain times of the day or at certain dates.
The Back Office application provides a character terminal oriented interface to Back Office staff allowing day to day system operations such as trade editing and confirmations, database maintenance, accounting and management reports to be generated. It is designed to be run on a minicomputer with terminals attached to it.
All Optech applications are written using the various modules in the Optech system. Each module is designed to handle one aspect of the system, and the source code for every module is encapsulated in an object code library that Optech applications link with.
This module incorporates the complex mathematical models that define a financial instrument, allowing it to be priced and its risk measures computed.
Provides an application programming interface to the SQL DataServer. Applications can access and modify information in the database without resorting to SQL. An audit log of transactions is also generated.
Besides being a date function library, this library allows queries into a database containing information about trading days, holidays and instrument maturity.
This library allows access to and addition of historical market data from a database which is automatically updated by the COM application.
Allows access to and modification of market values stored in the database. Market values may also be overridden manually by privileged users.
This module is in charge of validating user access and allowing the creation and deletion users. Access rights are organised in a hierarchical fashion allowing rights to be granted and revoked on a group basis. Rights can also be transferred on an individual basis.
Applications interface to the clipboard via routines in the clipboard library, which allows an application to post things into and retrieve things from the clipboard.
This library provides report generation tools for applications. Reports are generated as NROFF/TROFF input files, allowing flexibility of report output generation on a wide variety of printers.
Graphical User Interfaces and networked workstation technology is gaining rapid acceptance within the finance industry as a tool for management of trading floor activities. The volume and rate of trading and the complexity of risk management demands fast response to market changes. In a rapidly expanding market, a distributed approach provides the extensibility that is required. Although designing a distributed systems imposes substantial challenges in software engineering, it also offers many benefits including true concurrency and fault-tolerance. The resultant implementation can be made efficient despite the networking overhead through careful division of labour throughout the network.
The Optech FX (Foreign Exchange) options trading system is presently undergoing alpha testing and the development of an interest rate system is already underway. The system currently consists of over 250,000 lines of source code and takes up over 5 Megabytes of storage.
Stripped executables range from a few hundred kilobytes to over one megabyte in size depending on the complexity of the application. A port of the system to the IBM RT workstation under the AIX operating system is nearing completion and a port to Digital's DECstations running VMS will commence soon.
A system as large or as complex as a trading system cannot be developed by one or two persons and space precludes us from listing all the members of the development team. We would like to mention Andrew Carroll, Garry De Jager and Michelle Rembel for their invaluable contributions in the project. We would also like to thank John Cornwall, Ralph McKay, Robert Lee-Johnson and Jenny Spence for their help in preparing this article.