WO2008088206A1

WO2008088206A1 - A bond analysis system

Info

Publication number: WO2008088206A1
Application number: PCT/MY2008/000004
Authority: WO
Inventors: Lee Lwei Sia; Aik Beng Neo; Chee Seng Thang; Mustapha Kamal Abdul Wahid
Original assignee: Cimb Group Sdn Bhd
Priority date: 2007-01-18
Filing date: 2008-01-18
Publication date: 2008-07-24
Also published as: MY142560A

Abstract

A bond analysis system, including a selection module for searching trade data for attribute data of a proxy bond, the trade data comprising attribute data of bonds traded on a bond market, the proxy bond being selected when the attribute data corresponds to predetermined filter and searching-parameters for a selected-bond, and an analysis module for applying an attribute analysis process to generate price data for the selected bond by processing- attribute data of the proxy bond.

Description

A BOND ANALYSIS SYSTEM

Field

The present invention relates to a bond analysis system.

Background

Data processing systems have been developed to store and communicate the extensive financial data associated with instruments traded on the world's financial markets. The instruments, such as shares, stocks, securities, etc, have a wide variety of parameters associated with the manner and the markets in which they are traded. The variety of the instruments gives rise to an extensive set of complex parameters associated with the instruments that are communicated by various reporting systems and analysed by parties participating in the markets. Considerable scientific and technological resources are devoted to producing systems for analysing particular instruments, such as shares and stocks.

For debt securities, i.e. bonds, bond markets are supported- by systems for trading bonds and providing access to and communicating parameter data associated with the bonds. For example, the Ringgit Bond Market involvesifre^~trade^~of^aτiumberτ>f-bond instruments, including Malaysian Government Securities (MGS), Malaysian Treasury Bills (MTB),

Government Investment Issues (GII), Bank-Negara- Malaysia Bills (BNB), Private Debt

Securities (PDS), Agency Bonds and a variety of other fixed, floating and steps coupon securities, which for the purposes of this description-will-all be referred to as bonds. A

Bond Information and Dissemination System (BIDS) provides a computerised centralised database for bonds of the market, providing data on the terms of issue, details of trades done and user information on the various bonds issued by both the government and the private sector. An automatic tendering system, FAST, is also able to electronically handle invitations to tender, bid submissions and the entire tendering process for particular bonds.

The bonds are typically traded on a yield basis and have a tenor, normally expressed in yeais, at which the bond matures. The yield is normally specified as a percentage rate of Yield-To-Maturity or Yield-To-Call. Bloomberg L.P. has developed a number of sophisticated reporting and analysis systems for generating indicative yields on the various bonds traded on the Ringgit Bond Market. Examples include the systems for generating the Cl 13 index for MTB, the C128 index for MGS, the C120 index for Cagamas, the C104 index for. Khazanah and the BNM index-for PDS.

In reality, liquidity, the market's preference, investors' knowledge, differentregulatory-and^~ accounting treatments, and other non-quantifiable factors are factors that cause a bond's price to deviate from curves of the generic indicatives yield data provided by the reporting and analysis systems mentioned above.

Furthermore, the technical systems used to perform analysis on attributes of bonds rely on the accuracy of the parameter data provided by the various data sources, in particular the extent to which the data is current. Particular bonds may not have been traded for some time or the parameter data associated with the bonds may be far from current. A significant problem is providing an analysis system that is able to generate useful data for a bond when attribute data in relation to that bond is absent or no longer current

Accordingly, it is desired to address the above or at least provide useful alternative.

Summary

In accordance with the present invention there is provided a bond analysis system, including: a selection module for searching trade data for attribute data of a proxy bond, said trade data comprising attribute data of bonds traded on a bond market, said proxy bond being selected when said attribute data corresponds to predetermined filter and searching parameters for a selected bond; and an analysis module for applying an attribute analysis process to generate price data. for said selected bond by processing attribute data of said proxy bond. Preferably said attribute analysis process is selected from one of a plurality of analysis processes based on the filter and searching parameters applied to select said proxy bond.

Preferably the system includes a data extraction module for accessing said trade data. The trade data may be stored by said system or accessed from a data source system accessible over a communications network.

Preferably the system further includes a test module for analysing the generated price data by using said selection module to select a test proxy bond for the selected bond at a time when the selected bond was traded and using said analysis module to generate test price data by processing attribute data for said test proxy bond, and for comparing said test price data to price data for the trade of said selected bond at said time.

The present invention also provides a bond analysis process, including: a selection module for searching trade data for attribute data of a proxy bond, said trade data comprising attribute data of bonds traded on a bond market, said proxy bond being selected when said attribute data corresponds to predetermined filter and searching parameters for a selected bond; and an analysis module for applying an attribute analysis process to generate price data for said selected bond by processing attribute data of said proxy bond

Description of the Drawings

Preferred embodiments of the present invention are hereinafter described, by way of example only, with reference to the accompanying drawings, wherein: Figure 1 is a block diagram of a preferred embodiment of a bond analysis system connected to a communications network; Figure 2 is a flow diagram of an analysis process performed by the analysis system; Figure 3 is a flow diagram of a selection process performed by the analysis system;

Figure 4 is a flow diagram of a first relative spread analysis process performed by the analysis system;

Figure 5 is an example AAA yield curve used in an illustrative example of the first relative spread analysis process;

Figure 6 is a graph of two risk-free yield curves used in the illustrative example of the-first relative spread analysis process;

Figure 7 is a flow diagram of a second relative spread analysis process performed by the analysis system;

Figure 8 A is a AA yield"xurve^~used in an illustrative example of the second- relative- spread- analysis process; Figure 8B is a graph of two AA yield curves used in the illustrative example of the second relative spread analysis process;

Figure 8C is a graph of two-risk free yield curves used in the illustrative example of the second relative spread analysis process;

Figure 9 is a flow diagram of a third relative spread analysis process performed by the analysis system;

Figure 10 is an example AA yield curve used in an illustrative example of the third relative spread analysis process; and

Figure 11 is a flow diagram of a back testing process performed by the analysis system.

Description of Preferred Embodiments

A bond analysis system 100, as shown in Figure 1, includes a data extraction module 110, a proxy bond selection module 112 and an analysis module 114. The analysis module 114 uses a number of relative spread analysis (RSA) process components 120, 122 and 124. The system 100 also includes a back testing module 116 for adjusting time parameters utilised by the selection module 112 and the analysis module 114 in order to generate data for use in producing metric data to monitor the performance of the analysis system 100.

The data extraction module 110 extracts trade data for the analysis system 100 and stores the data in structured databases 130 and 131 of the system 100. Database 130 includes historical trade data and attribute data associated with bonds and includes, for each bond, data representing identification of the bond, the date and yield for the last trade, the tenor, valuation terms, rating, callable features and coupon types. The database 131 includes historical indicative yield curve data extracted from Bloomberg or any other data provider as follows: (1) daily Malaysian T-BiIl curve data (may include 3 tenors, e.g 3m, 6m and lyr);

(2) daily MGS yield curve data (may include 12 tenors, e.g. 3m, 6m, lyr, 2yr, 3yr, 4yr, 5yr, 7yr, 8yr, 9yr, lOyr and 15yr);

(3) daily CMCTCagamas yield curve data (may include 10 tenors, e.g. 3m,6m,lyr, 2yr, 3yr, 4yr, 5yr, 7yr, 8yr and 9yr); (4) daily Khazanah yield curve data (may include 8 tenors, e.g. 3m,6m,lyr, 2yr, 3yr,

4yr, 5yr and 7yr) and (5) weekly BNM Indicative yield curve data for different ratings after interpolation and extrapolation (may include 5 tenors, e.g. 3yr, 5yr, 7yr, lOyr and 15yr for AAA,

AAl, AA2, AA3, Al₅ A2, A3, BBBl, BBB2, BBB3, BBl₅ BB2, BB3, Bl₅ B2, B3 and C bonds, and 4 tenors, e.g. 3yr, 5yr, 7yr and lOyr for Old Cagamas and New

Cagamas bonds).

The extraction module 110 communicates with a number of data source systems over a communications network 140 to obtain attribute data for bonds traded on one or more bond markets. The data source systems include BIDS₅ systems maintained by Bloomberg L.P. and other information service providers.

The data extraction module 110 extracts trade data on a daily basis from the data source systems.

The bond analysis system 100 performs an analysis process 200, as shown in Figure 2, which commences at step 202, and involves generation of a user interface that allows a particular proxy bond to be selected for analysis. The bond is selected from securities traded on the market for which data is extracted by the extraction module 110. Once a bond is selected, data identifying the bond is provided to the analysis module 114 which then seeks to access attribute data for the bond from the databases 130 and 131 (step 204). Current attribute data for the bond- may also be extracted by the data extraction module 110.

The analysis module 114 determines (step 206) whether the bond has been traded within a predetermined time, based on a time parameter which may represent one day. If the bond has been traded on the valuation day, then the analysis module 114 generates price data (208) representing the yield for the last trade. If the bond has not been traded within a predetermined period of time determined by the time parameter, then the selection module 112 is invoked to perform a proxy bond search process (210). The proxy bond- selection module 112, as described below, uses a set of filtering rules defined by filter parameters to search the trade data to locate a bond having attribute data that may be used as a proxy bond to generate price data for the selected bond. The price data represents a useful and accurate yield for the selected bond, notwithstanding the bond may be inactive or illiquid, and not traded for some time.

The analysis module 114 performs a first RSA process (step 214) if the selection module 112 has located a proxy bond having attribute data that meets or corresponds to a first set of filter parameters (212). If the first set of filter parameters are not met, then a determination is made as to whether a proxy bond has been selected having attribute data that meets a second set of filter parameters (216). If so, the analysis module 114 applies a second RSA process (218) to generate price data. If a proxy bond cannot be selected using the selection module 112, the analysis module 114 applies a third RSA process (step 220) to generate price data for the selected bond. The RSA processes 214, 218 and 220 are performed respectively by the RSA components 120, 122 and 124, and operate on the attribute data of a proxy bond or the selected bond. The attribute data includes yield data for the last trade, tenor spread data, proxy curve movement data, residual risk adjustment data and indicative yield data, as discussed below.

The price data generated is communicated by the analysis system 100 (224). This may involve transmission over the communications network 140 for use by other analysis systems or a trade system to execute a trade for the selected bond based on parameters associated with the price data generated. The price data can also be used to generate a yield displayed as a percentage for the user on the user interface.

The proxy bond selection module 112 performs a proxy bond selection process 300, as shown in Figure 3, to determine a proxy bond for the selected bond. The selected bond can be referred to as a Mark-To-Market (MTM) bond. Attribute data on all potential proxy bonds is obtained using the data extraction module 110 in step 302, so only potential bonds belonging to the same category are selected at step 304. Step 304 differentiates Coupon- bonds^~fromrZero-coupon bonds and short-dated bonds from long-dated bonds. A cut-off level is used to determine whether a bond is short-term or long-term and that may be set at two years. If no potential proxy bonds belonging to this category, then^" no proxy bond is selected (step 334). Similarly, if there are no potential proxy bonds with the same issuer as, or classified as homogeneous bond to, the mark-to-market (MTM) bond (hereafter both refer to as "same issuer"), as tested at step 306, then no proxy bond is selected (step 334). If the pool of potential proxy bonds includes at least one bond belonging to the same category and issued by the same issuer as the MTM bond, then a proxy bond is selected for either the first RSA process (214) or the second RSA process (218). If there are more than one potential proxy bonds identified at step 306, the selection process 300 searches for the proxy bond with a tenor as close as possible to the MTM bond, and a not-to-distant trade date.

In steps 310 to 320, the selection process 300 steps through anumber of selection criteria searching forihe most relevant proxy bond At step 310 it is determined whether a proxy bond was traded more recently than a selected recent cut-off duration, the recent cut-off duration being set by a first historical time filter parameter (shown as five days in Figure 3). If the criteria of step 310 are not met, step 312 searches for a proxy bond with a tenor similar to that of the MTM bond, where similar corresponds to a tenor time filter parameter, in this case plus or minus seven years from the tenor of the MTM bond. The other criterion in step 312 is that the proxy bond be traded within the time frame set by the first historical time filter parameter (e.g. five days in Figure 3). If an appropriate proxy bond is not found in steps 310 and 312, the selection process 300 continues with two further similar^" steps 314 and 316, where that particular bond (314) or a same bond with similar tenor (316) is searched that has been traded in a time more recent than a second time historical time filter parameter (twenty days in Figure 3). If an appropriate proxy bond has still not been found to fit the selection criteria, steps 318 and 320 from- a similar selection to steps 310 and 312, except extending to bonds traded more recently than a third historical time filter parameter (sixty days in Figure 3). If any of the searching steps 310 to 320 are successful in identifying one or more bonds, these bonds are used as proxy bonds for determining the estimated yield of the MTM bond. At step 322, if there is more than one potential proxy bond matching the selection criteria, one-of these-is-seleeted-withrthe nearest matching tenor to the MTM bond, and the proxy bond attribute data is transferred to the first KSA process (214) at step 324. The first RSA process generates price data representing an MTM yield estimated for the MTM bond.

If the selection process steps 310 to 320 fail to select one or more proxy bonds with appropriate attributes, then steps 326 and 328 sequentially search for that particular bond (326) or same bond with similar tenor (328) to the MTM bond that have been traded more recently than a fourth historical time filter parameter (120 days in Figure 3). One or more bonds match the selection criteria in step 326 or 328, the proxy bond with the tenor closest to that of the MTM bond is selected at step 330, and the attribute data of the selected proxy bond is transferred to a second RSA process for generating price data representing an MTM yield (at step 332).

If no proxy bond is selected by the selection steps 310 to 320 or 326 and 328, it is because there are no bonds with sufficiently similar tenors, or no proxy bonds traded more recently than the fourth historical time filter parameter or no potential proxy bonds traded by the same issuer, and price data representing the estimated MTM yield of the MTM bond is generated using the third RSA process (step 336).

The historical time filter parameters, the criterion for similarity between tenors and the maturity tenor cut off level in bond selection process 300 are selectable filter parameters set to optimise the bond selection process 300, and thus the accuracy of the generated MTM yield price data. In an alternative configuration of the proxy bond selection process 300, the search steps 318 and 320 result in the second RSA process 218, i.e. that steps 318 and 320 lead to steps 330 to 332. For example, the second RSA process 218 can start earlier at 60 or 20 day segments, and the third RSA process 220 can start at 120 or 60 days segments. Configuring the process 300 to result in the selection of the third RSA process 220 or the second RSA process 218 when alternative search criteria are met may be selected by a user.

The first RSA process 214, shown Figure 4, begins by receiving attribute data on a selected proxy bond that has been chosen to represent similarities to the MTM bond. The first RSA process 214 retrieves the attribute data, including the last done yield of the proxy bond stored in database 130 (step 402, which is the known yield of the proxy bond at the time when it was last traded). At step 404, the first RSA process 214 accesses the downloaded indicative yield curve data that are extracted by module 110 and stored in database 131. At step 404, an appropriate indicative yield curve will be chosen to represent the term structure of the MTM bond and proxy bond. It also extracts the risk free yield curve at two points in time: firstly representing the risk free yield curve at the valuation date, and secondly representing the yield curve current at the time when the proxy bond was last traded (i.e. the second downloaded yield curve corresponds to the yield in step 402). Based on the information analyzed at step 402 and 404, the process 214 generates a raw tenor spread representing the percentage difference in the yields of bonds traded with different tenors as determined from the most recent yield curve data (step 406). If the proxy has the same tenor as the MTM bond, the raw tenor spread will be 0%. The raw tenor spread from step 406 is adjusted at step 408 to account for "pull-to-par effects", At step 408, the raw tenor spread is refined after taking into account the change in tenor of the proxy bonds over time. For example, the price of a 3 -year bond traded 6 months ago represents a yield for 3.5-year bond, not a 3-year bond. The difference of the price of a 3-year bond and 3.5- year bond exhibits a "pull-to-par effect" as a result of the fact that the price for the bonds will eventually converge to par (100.00) at maturity. After generating the tenor spread, process 214 generates a parameter indicative of the movement of the risk-free yield curve (for example a MGS curve) between the time and that the proxy bond was. last traded and the risk-free yield curve at the valuation date; the movement of the risk-free yield curve over the observed period is expressed as a percentage (step 410). Finally, the first RSA process 214 generates estimated attribute data for the MTM bond (specifically the current yield of the MTM bond) from: the last done yield of the proxy; the adjusted tenor spread; and the proxy curve movement, which is represented by risk-free yield curve movement (at step 412). The estimated MTM yield price data is then generated by summing the data representing the las1rdOne^~yield^~of- proxy ^'bond, the tenor spread and the-proxy-curve- movement. The first RSA process 214 assumes the proxy bond is traded recently enough to keep the credit-spread constant over the observed period. Thus, risk-free yield curve movement between the valuation date and last traded date is used to explain the interest rate movement in the market.

In an example of the first RSA process 214, the selected MTM bond is a 3 -year AAA rated bond, and the selected proxy bond, 'AAA05', has an AAA rating and a 5-year tenor and was traded 20 days ago with a yield of 5.5%. Attribute data (traded information) for this proxy bond is downloaded at step 402. AAA yield curve data is downloaded at step 404 representing the yield curve from last Friday (i.e. the most recent yield curve, published by BNM on weekly basis) and 2 risk free curves at 2 points in time: one risk-free curve as at valuation date, and another risk-free curve as at 20 days ago. The raw tenor spread is generated (step 406) as the difference in yields between 3-year tenor and 5-year tenor bonds derived from the AAA yield curve term structure. The curve shown in Figure 5 is an example of the AAA yield curve term structure, being the most recent yield curve corresponding to a bond with the rating (AAA) of the proxy bond, AAA05. The AAA proxy yield curve shows a yield of 6.0% for a 5-year bond and a yield of 5.0% for a 3-year bond. As a 3-year bond has a lower yield than a 5-year bond according to the AAA yield curve in step 406, the tenor spread adjustment is a negative number and is equal to 5.0% minus 6.0% or -1.00%. The raw tenor spread is adjusted for "20 days" pull-to-par effects at step 408 which cause no shift in this example. To account for interest rate movement between the last traded date and valuation date, the first RSA process 214 assumes the credit spread remains constant for 20 days. Therefore, interest rate movement is deteraiined by measuring risk-free curves 20 days ago and the risk curve of the valuation date; the risk-free yield curve is published by the information provider on daily basis, hi the example of Figure 6, the risk-free yield curve at the valuation date is higher than the yield curve of 20 days ago. For 5-year bonds, the yield has increased by 0.5% from 4.3% to 4.8%. Finally at step 412, the estimated yield price data of the MTM bond is generated to represent a yield of 5.5% adjusted down 1.0% for the shorter tenor of the MTM bond, and adjusted up by 0.5% due to movement of the proxy curve in the last 20 days. The MTM'yield is therefore, 5.50% - 1.0% +0.5%-= 5:50%.

The second RSA process 218, shown in Figure 7, ^"begins by receiving attribute data on a selected proxy bond that has been chosen to represent similarities to the MTM bond. The RSA process retrieves the attribute data, including the last done yield of the proxy bond (step 402). At step 604, the second RSA process 218 downloads the proxy yield curve data for the two points in time, the last traded date and most recent yield curve date, as it is published by BNM once a week. In step 606, the second RSA process 218 downloads data for two risk-free yield curves, using the data extraction module 110, which indicate the most recent movement of yield curves in the market since last Friday until Today. The risk-free curve data is published by information service provider on daily basis. The raw tenor spread is generated (step 406) and adjusted (step 408) to account for pull-to-par effects of the proxy and MTM bonds. After generating the tenor spread, process 218 generates proxy curve movement data indicative of the movement of the yield curve between the time that the proxy bond was last traded and time of the most recent yield curve (step 610). At step 614, the yield estimate is updated to the date of valuation by generating residual risk adjustment data, representing a percentage, indicative of the movement of the risk-free yield curve between the date of the most recent yield curve (from step 604) and the date of the most recent risk-free yield curve, which is likely to be today. Finally at step 616 the second RSA process 218 generates the estimate yield price data of the MTM bond by summing the data representing the last done proxy yield, the tenor spread, the proxy curve movement and the residual risk adjustment.

In an example of the second RSA process 218, the selected MTM bond is a 7-year AA 000004

- 12 -

rated bond, and the selected proxy bond, 'ABC09¹, has an AA rating and a 9-year tenor and was traded four months ago with a yield of 7.30%. Attribute data (traded information) for this proxy bond is downloaded at step 402. Two Proxy Yield curves (AA Yield Curve in this example) are downloaded-at- step 604-represeήting the -yield curve from last Friday (the most recent yield curve, published by BNM on weekly basis), and the yield curve from four months ago (the-yield- curve from-the- last proxy date). The raw tenor spread is generated (step 406) as the difference in yields between the 7-year and 9-year tenor based on the AA ρroxy-yield-curve._Ihe_curve shown in Figure 8a is an example yield curve, being the most recent yield curve corresponding to a bond with the rating (AA) of the proxy bond. The-yield curve shows a yield-of 7.9%-for_a 9-year bond, corresponding to the proxy, and a yield of 7.6% for a 7-year bond corresponding to the MTM bond. As a 7- year bond has a lower yield than a 9-year bond according to the proxy curve in step 406, the tenor spread adjustment is a negative number and is equal to 7.9% minus 7.6% or - 0.3%. The raw tenor spread is adjusted for pull-to-par effects at step 408, which causes no shift in this example. A measurement of proxy curve movement is determined from the AA curve of four months ago and the AA curve of last Friday as shown in Figure 8b. The proxy curve movement is expressed as a percentage indicating the vertical movement of the curve for bonds with a tenor equal to the tenor of the MTM bond. In the example of Figure 8b, the most recent yield curve is higher than the yield curve of four months ago, and for 7-year bonds, the yield has increased by 0.2%. At step 614, the yield estimate is further refined to cover the interest movement between last Friday and the valuation date by generating residual risk adjustment data. The residual risk adjustment data represents a percentage, indicative of the movement of the risk-free yield curve between Friday (the date of the most recent yield curve from step 604) and the valuation date risk-free yield curve, which is likely to be today, as shown in Figure 8C. In this case the yield for risk- free bonds has risen from 3.8% to 4% since last Friday; the yield is considered for bonds with a tenor equal to the tenor of the MTM bond, i.e. 7 years. The residual risk adjustment in this case is +0.20%. Finally at step 616, the estimated yield price data of the MTM bond is generated to represent a yield of 7.3% adjusted down 0.3% for the shorter tenor of the MTM bond, and adjusted up by 0.2% due to movement of the proxy curve in the last four months until last Friday, and further adjusted up by 0.2% due to the movement of risk-free curve since last Friday until the valuation date. The MTM yield is therefore, 7.3% - 0.3% +0.2% + 0.2% = 7.4%.

The third RSA process 220, as shown in Figure 9, commences by accessing the most recent yield curve data (step 402) and the risk-free yield curve data (step 606). The yield of the MTM bond in the third RSA 220 is determined from the indicative yield of a bond of the MTM tenor (step 806) and a residual risk adjustment to take into account any possible-recent- movement- of the yield curve since the most recent yield curve -was published (step 614). The MTM yield price data is then generated at step 810 by summing the indicatrve"yieldrandiiϊeτesidualτisk adjustment. An example- of the third- RSA process illustrates estimation of MTM yield for an MTM bond with a tenor of 7 years and an AA rating. The process estimates the yield of a bond with the tenor of the MTM bond directly from the most recent AA yield curve (step 806). As shown in Figure 10, the AA yield curve from last Friday shows a yield of 7.6% for a bond with a tenor of seven years. Step 614 gives a residual risk adjustment of 0.2%. Finally, in step 810 the estimated yield of the 7-year MTM bond is determined by summing the yield from the AA curve of last Friday and the residual risk adjustment. The MTM yield is therefore, 7.6% + 0.2% = 7.8%.

The back testing module 116 performs a back testing process 1000 to compare attribute data for traded bonds with the price data generated for those bonds by the analysis module

114 at the time of the trades. This enables assessment of the accuracy of the system 100.

The first step 1002 in the -back testing process 1000, shown in Figure 11, generates a dummy portfolio which includes attribute data from a range of bonds that have been traded once or more over the back testing time period. The data in the dummy portfolio preferably includes data from bonds of different tenor, class and rating. The attribute data of the dummy portfolio is captured using data extraction module 110 over a back testing time period, e.g. the past three months or past 65 days. This attribute data includes all traded market prices for the last 65 working days. The number of observations required for comparison may be set to be 30 pairs or more. Each observation consists of 2 yields for comparison, which is the traded yield and estimated yield for all the bonds in the dummy portfolio over the back testing time period. At step 1006 in the back testing process 1000, the full analysis procedure of the analysis module 114, as shown in Figure 2, is performed on each of the bonds in the dummy portfolio on the days they are traded, but under the assumption that they were not traded on the particular day, i.e. the modelled price data must be extracted from one of the RSA processes 214, 218 or 220 in combination with the selection process 300. Step 1006 generates, for each bond in the dummy portfolio, price data indicative of the yield which may be compared to the actual price data of the bonds that were traded. To quantify the degree of similarity between the modelled prices, the modelled price data of the dummy portfolio and the price-data- corresponding- to- actual- trades, a statistical process is applied in step 1008. This statistical process generates a t- statistic, using the equation below, where xi is the statistical mean of the model price data, and X₂ is the statistical mean of the market (or traded) price data:

Xl -X2 t =

1I₁ n₂

where:

₂ _ Cn₁ -I)S²I +(n₂ -l)S²2 o p — n₁ +n₂ -2

5² ₁ = Variance of the model price data

5²2 = Variance of the market price data ni = Number of observations for model price samples n₂ = Number of observations for market price samples

The degree of freedom is reflected in the denominator, (tiχ + n₂ - 2).

For the t-statistic to be meaningful, the type of bonds chosen in the dummy portfolio are well balanced and the number of observations for comparison in the dummy portfolio are greater than 30 pairs. Once the t-statistic is generated, a decision rule is applied in step 1010: if the t-statistic lies within selected confidence limits, the model is regarded as accurate, and thus the rules and parameters in the analysis module 114 require no further adjustment. The rules are used to define elements and criteria of the processes executed by the analysis module 114. If on the other hand the generated statistic falls without selected confidence limits at step 1010, the process 1000 enters an error handling procedure 1014, where the parameters and rules of the analysis module 114 are adjusted, and the back testing process 1000 is reiterated to confirm that the adjusted analysis module 114 is now operating with more relevant parameters- and-rules. T-he=-baek-testing- process 1000 is performed on a periodic basis, for example on a quarterly basis.

The bond analysis system 100 may be implemented using a number of different technologies. In one example, the components 110, 112, 114, 116, 120, 122 and 124 may include computer program instruction code written in a language such as Visual Basic, Java, C++, etc., and either interpreted or compiled to run on an operating system, such as Windows, UNEK or Linux, running on a standard personal computer or server, such as that produced by IBM, Lenovo Corporation, Dell, HP, etc. The databases 130 and 131 maybe implemented using a spreadsheet such as Microsoft Excel, etc or a database application, such as Microsoft Access, Microsoft SQL, Oracle SQL, MySQL, etc. The executable code and the database are stored on computer readable storage memory of the computer. Alternatively, the components 110 to 124 may include dedicated hardware circuits, such as ASICs or FPGAs, that perform at least part of the processes performed by the components 110 to 124. The system 100 can be executed independently as a stand-alone application (i.e. installed and executed at the same location) or via a communication network 140 (i.e. installed and executed at separate locations). The communications network 140 may include a LAN, WAN, the Internet, and other public and private communications networks.

Many modifications will be apparent to those skilled in the art without departing from the scope of the present invention as herein before described with reference to the accompanying drawings.

Claims

1. A bond analysis system, including: a selection module for searching trade data for attribute data of a proxy bond, said trade data comprising attribute data of bonds traded on a bond market, said proxy bond being selected when said attribute data corresponds to predetermined, filter and searching parameters for a selected bond; and an-analysis-module"for-appl3dn^an^~attribute analysis process to-generate-price^data- for said selected bond by processing attribute data of said proxy bond.

2. A bond analysis system as claimed in claim 1, wherein said attribute analysis process is selected from one of a plurality of analysis processes based on the filter and searching parameters applied to select said proxy bond.

3. A bond analysis system as claimed in claim 2, wherein said attribute data represents a category of the traded bonds, and the filter and searching parameters include category data representative of a category of the selected bond.

4. A bond analysis system as claimed in claim 3, wherein said category data differentiates coupon-bonds from zero-coupon bonds.

5. A bond analysis system as claimed in claims 3 or 4, wherein the category data differentiates short-dated bonds from long-dated bonds.

6. A bond analysis system as claimed in any one of claims 3 to 5, wherein said proxy bond has a category the same as the category of the selected bond.

7. A bond analysis system as claimed in any one of claims 2 to 6, wherein said attribute data represents an issuer of the traded bonds, and the filter and searching parameters include issuer data representative of an issuer of the selected bond.

8. A bond analysis system as claimed in claim 7, wherein said proxy bond has an issuer the same as said selected bond.

9. A bond analysis system as claimed m clamr2,^~6 or 8, wherein when- said'proxybond is not selected, said one of the analysis processes includes the steps of: generating risk-free curve movement^" datairom- a- risk-free-yield -curve on" the last traded day of the selected bond and a risk-free yield curve on a most recent day, using the terrarof- the=selected-bond; and generating the price data for the selected bond from price data of the selected bond on a last traded day of the selected bond,

10. A bond analysis system as claimed in claim 9, wherein the price data of the selected bond is generated by summing: percentage yield data of the selected bond on a last traded day of the selected bond; and percentage change data indicative of the risk-free curve movement data.

11. A bond analysis system as claimed in any one of claims 2 to 10, wherein said attribute data represents a last trade date of the traded bonds-, and the filter and searching parameters include date range data representative of a selected range of last trade dates.

12. A bond analysis system as claimed in claim 11, wherein said attribute data represents a tenor of the traded bonds, and the filter and searching parameters include tenor data representative of a tenor of the selected bond.

13. A bond analysis system as claimed in claim 11, wherein the selected range of last trade dates corresponds to the last five days.

14. A bond analysis system as claimed in claim 13, wherein the tenor of the proxy bond is the same as the tenor of the selected bond.

15. A bond analysis system as claimed in claim 13, wherein the tenor of the proxy bond is within seven years of the tenor of the selected bond.

16. A bond-analysis system- as claimedin-either one of claims 14 or 15, wherein said one of the analysis processes is a first analysis process.

17. A bond analysis system as claimed hi claim 12, wherein the selected range of last trade dates corresponds-to the last-twenty-days.

18. A bond analysis'system- as claimedάn claim 17, wherehrthe^~tenor of the proxy bond is the same as the tenor of the selected bond.

19. A bond analysis system as claimed in claim 16, wherein the tenor of the proxy bond is within seven years of the tenor of the selected bond.

20. A bond analysis system as claimed in claim 18 or 19, wherein said one of the analysis processes is a first analysis process.

21. A bond analysis system as claimed in claim 12, wherein the selected range of last trade dates corresponds to the last sixty days.

22. A bond analysis system as claimed- in claim 21, wherein the tenor of the proxy bond is the same as the tenor of the selected bond.

23. A bond analysis system as claimed in claim 21, wherein the tenor of the proxy bond is within seven years of the tenor of the selected bond.

24. A bond analysis system as claimed in claim 22 or 23, wherein said one of the analysis processes is a first analysis process.

25. A bond analysis, system as claimed in claim 2, 16, 20 or 24, wherein a first analysis process, of the plurality of analysis processes, includes the steps of: generating raw tenor spread data from most recent yield curve data of the proxy bond, using a tenor of the selected bond and a tenor of the proxy bond; generating-adjustedi:enorspread data from the raw tenor spread by applying a pull- to-par adjustment; generating-risk=free curveτnovement data from risk-free yield curve data on the last traded day of the proxy bond and risk-free yield curve data on a most recent day, using the tenor of the selected bond; and- generating the price data for the selected bond from price data of the proxy bond on a last-traded day of the proxy bond, the adjusted tenor spread data; and the risk-free curve movement data.

26. A bond analysis system as claimed in claim 25, wherein the price data of the selected bond is generated by summing: percentage yield data of the proxy bond on a last traded day of the proxy bond; percentage change data of the adjusted tenor spread data; and percentage change data indicative of the risk-free curve movement data.

27. A bond analysis system as claimed in claim 12, wherein the selected range of last trade dates corresponds to the last one hundred and twenty days.

28. A bond analysis system as claimed in claim 27, wherein the tenor of the proxy bond is the same as the tenor of the selected bond.

29. A bond analysis system as claimed in claim 27, wherein the tenor of the traded bonds is within seven years of the tenor of the selected bond.

30. A bond analysis system as claimed in claim 28 or 29, wherein said one of said analysis processes is a second analysis process.

31. A bond analysis system as claimed in claim 2 or 30, wherein a second analysis process, of the plurality of analysis processes, includes the steps of: generating raw tenor spread data from yield curve data of the proxy bond on a most recent yield curve provision day, using a tenor of the selected bond and a tenor of the proxy bond; generating adjusted tenor spread data from the raw tenor spread by applying a pull- to-par adjustment; generating proxy curve movement data from yield curve data of the proxy bond on a last traded day of the proxy bond and the yield curve data of-the-proxy-bond-on-the-most- recent yield curve provision day, using the tenor of the selected bond; generating risk-free curve movement data from risk-free- yield curve data on the most recent yield curve provision day and risk-free yield curve data on a most recent day, using the tenor of the selected bond; and generating the price data for the selected bond from price data of the proxy bond on a last traded day of the proxy bond, the adjusted tenor spread data, the proxy curve movement data, and the risk-free curve movement data.

32. A bond analysis system as claimed in claim 31 , wherein the price data of the selected bond is generated by summing: percentage yield data of the proxy bond on a last traded day of the proxy bond; percentage change data of the adjusted tenor spread data; percentage change data indicative of the proxy curve movement data; and percentage change data indicative of the risk-free curve movement data.

33. A bond analysis system as claimed in any one of the preceding claims, further including a data extraction module for accessing said trade data.

34. A bond analysis system as claimed in any one of the preceding claims, wherein the trade data is stored by said bond analysis system or accessed from a data source system accessible over a communications network.

35. A bond analysis system as claimed in any one of the preceding claims, further including a back testing module for analysing the generated price data by using said selection module to select a test proxy bond for the selected bond at a time when the selected bond was traded and using said analysis module to generate test price data by processing attribute data for said test proxy bond, and for comparing said test price data to price data for the trade of said selected bond at said time.

36. A bond analysis system as claimed in claim 35, wherein the back testing module compares the test price data to price data-forme-trade-of at Jeast=thirty-selected bonds.

37. A bond analysis -process, including: searching trade data for attribute data of a proxy bond, said trade data comprising attribute data of bonds traded on a bond market, said proxy bond being selected when said attribute data corresponds to predetermined filter and searching parameters for a selected bond; and applying an attribute analysis process to generate price data for said selected bond by processing attribute data of said proxy bond.

38. A bond analysis process as claimed in claim 37, wherein said attribute analysis process is selected from one of a plurality of analysis processes based on the filter and searching parameters applied to select said proxy bond.

39. A bond analysis process as claimed in claim 38, wherein said attribute data represents a category of the traded bonds, and the filter and searching parameters include category data representative of a category of the selected bond.

40. A bond analysis process as claimed in claim 39, wherein said category data differentiates coupon-bonds from zero-coupon bonds.

41. A bond analysis process as claimed in claims 39 or 40, wherein the category data differentiates short-dated bonds from long-dated bonds.

42. A bond analysis process as claimed in any one of claims 39 to 41, wherein said proxy bond has a category the same as the category of the selected bond.

43. A bond analysis process as claimed in any one of claims 38 to 42, wherein said attribute data represents an issuer of the traded bonds, and the filter and searching parameters include issuer data representative of an issuer of the selected bond.

44. A bond analysis. process_as-jdaimedJn_clainL 43, wherein said proxy bond.has_an- issuer the same as said selected bond.

45. A bond analysis process as claimed in claim 38, 42 or 44, wherein when said proxy bond is not selected, said one of the analysis processes includes the steps of: generating risk-free curve movement data from a risk-free yield curve on the last traded day of the selected bond and a risk-free yield curve on a most recent day, using the tenor of the selected bond; and generating the price data for the selected bond from price data of the selected bond on a last traded day of the selected bond, and the risk-free curve movement data.

46. A bond analysis process as claimed in claim 45, wherein the price data of the selected bond is generated by summing: percentage yield data of the selected bond on a last traded day of the selected bond; and percentage change data indicative of the risk-free curve movement data.

47. A bond analysis process as claimed in any one of claims 38 to 46, wherein said attribute data represents a last trade date of the traded bonds, and the filter and searching parameters include date range data representative of a selected range of last trade dates.

48. A bond analysis process as claimed in claim 47, wherein said attribute data represents a tenor of the traded bonds, and the filter and searching parameters include tenor data representative of a tenor of the selected bond.

49. A bond analysis process as claimed in claim 47, wherein the selected range of last trade dates corresponds to the last five days.

50. A bond analysis process as claimed in claim 49, wherein the tenor of the proxy bond is the same as the tenor of the selected bond.

51. A-bond=analysis-process as claimed in claim 49, wherein-the-tenor of-the proxy-bond is within seven years of the tenor of the selected bond.

52. A bond analysis process as claimed in either one of claims 50 or 51, wherein said one of the analysis processes is a first analysis process.

53. A bond analysis process as claimed in claim 48, wherein the selected range of last trade dates corresponds to the last twenty days.

54. A bond analysis process as claimed in claim 53, wherein the tenor of the proxy bond is the same as the tenor of the selected bond.

55. A bond analysis process as claimed in claim 52, wherein the tenor of the proxy bond is within seven years of the tenor of the selected bond.

56. A bond analysis process as claimed in claim 54 or 55, wherein said one of the analysis processes is a first analysis process.

57. A bond analysis process as claimed in claim 48, wherein the selected range of last trade dates corresponds to the last sixty days.

58. A bond analysis process as claimed in claim 57, wherein the tenor of the proxy bond is the same as the tenor of the selected bond.

59. A bond analysis process as claimed in claim 57, wherein the tenor of the proxy bond is within seven years of the tenor of the selected bond.

60. A bond- analysis process- as claimed in claim 58 or 59, wherein said one of the analysis processes is a first analysis process.

61. A bond analysis process as claimed in claim 38, 52, 56 or 60, wherein a first analysis process, of the plurality-of analysis processes,- includes the steps-of: generating raw tenor spread data from most recent yield curve data of the proxy bond, using a- tenor of the selected bond and a tenorof the-proxy bond; generating adjusted tenor spread data from the raw tenor spread by applying a pull- to-par adjustment; generating risk-free curve movement data from risk-free yield curve data on the last traded day of the proxy bond and risk-free yield curve data on a most recent day, using the tenor of the selected bond; and generating the price data for the selected bond from price data of the proxy bond on a last traded day of the proxy bond, the adjusted tenor spread data; and the risk-free curve movement data.

62. A bond analysis process as claimed in claim 61, wherein the price data of the selected bond is generated by summing: percentage yield data of the proxy bond on a last traded day of the proxy bond; percentage change data of the adjusted tenor spread data; and percentage change data indicative of the risk-free curve movement data.

63. A bond analysis process as claimed in claim 48, wherein the selected range of last trade dates corresponds to the last one hundred and twenty days.

64. A bond analysis process as claimed in claim 63, wherein the tenor of the proxy bond is the same as the tenor of the selected bond.

65. A bond analysis process as claimed in claim 63, wherein the tenor of the traded bonds is within seven years of the tenor of the selected bond.

66. A bond- analysis-process as^~ claimed in claim 64 or 65, wherein said one of said analysis processes is a second analysis process.

67. A bond analysis process as claimed in claim 38 or 66, wherein a second analysis process, of the plurality-of-aήalysis processes,, includes the steps of: generating raw tenor spread data from yield curve data of the proxy bond on a most recent-yield curve provision day, using a -tenor of the selected bond and a tenor of the proxy bond; generating adjusted tenor spread data from the raw tenor spread by applying a pull- to-par adjustment; generating proxy curve movement data from yield curve data of the proxy bond on a last traded day of the proxy bond and the yield curve data of the proxy bond on the most recent yield curve provision day, using the tenor of the selected bond; generating risk-free curve movement data from risk-free yield curve data on the most recent yield curve provision day and risk-free yield curve data on a most recent day, using the-tenor of the selected bond; and generating the price data for the selected bond from price data of the proxy bond on a last traded day of the proxy bond, the adjusted tenor spread data, the proxy curve movement data, and the risk-free curve movement data.

68. A bond analysis process as claimed in claim 67, wherein the price data of the selected bond is generated by summing: percentage yield data of the proxy bond on a last traded day of the proxy bond; percentage change data of the adjusted tenor spread data; percentage change data indicative of the proxy curve movement data;, and percentage change data indicative of the risk-free curve movement data.

69. A bond analysis process as claimed in any one of the preceding claims, further inchiding extracting said trade data from at least one of a plurality of data sources.

70. A bond analysis process as claimed in any one of the preceding claims, wherein the trade data is stored locally or accessed over a communications network.

71. A bond analysis process as claimed in any one of the preceding claims, further including back testing to assess the generated price data by using the proxy bond selection steps to select a test proxy bond for the seleeted-bond-at-a-time-when-the selectedbond was" traded and using at least one of said analysis processes to generate test price data by processing attribute data for said test-proxy bond, and compare said testprice data to price data for the trade of said selected bond at said time.

72. A bond analysis process as claimed in claim 71, wherein said back testing compares the test price data to price data for the trade of at least thirty selected bonds.

73. Computer program instructions, stored on storage media, for use in performing a bond analysis process as claimed in any one of claims 37 to 72.

74. A bond analysis system, including: a selection module for selecting an optimal proxy bond for a selected bond by processing attribute data of bonds traded on a bond market, a traded bond being selected as said proxy bond when said attribute data of said traded bond meets predetermined criteria associated with said attributed data of said selected bond; and an analysis module for applying an attribute analysis process to generate price data for said selected bond by processing attribute data of said proxy bond.