GB2403031A

GB2403031A - Method and apparatus for generating a broadcast graphic from multiple data sets

Info

Publication number: GB2403031A
Application number: GB0314012A
Authority: GB
Inventors: Kevan Moretti; Nick Bryan
Original assignee: Iknowledge Ltd
Current assignee: Iknowledge Ltd
Priority date: 2003-06-17
Filing date: 2003-06-17
Publication date: 2004-12-22
Also published as: GB0314012D0

Abstract

A method for generating a broadcast signal includes accessing multiple data sets, where each data set corresponds to a respective entity and comprises a sequence of data values. An index value is generated from each data set. The index value is calculated on a statistical basis from the data values in the data set. The index values are normalised so as to ensure comparability between index values from different data sets. A graphic is created representative of a group of entities. The graphic includes an identifier and a corresponding index value for each entity in the group, with the identifier and the index value for any given entity being displayed in association with one another. The graphic can now be incorporated into a broadcast signal. In one embodiment, the entities are horses, and the index values and the graphic are used to summarise the form histories of the different horses in a race.

Description

2403031. . ..

METHOD AND APPARATUS FOR GENERATING A BROADCAST GRAPHIC

FROM MULTIPLE DATA SETS

s Field of the Invention

The present invention relates to analysing and providing graphical summaries of data sequences, especially for use in a broadcast environment.

0 Background of the Invention

Many fields of human endeavour involve making some prediction about the future. Examples of such activities include investigations into possible climatic change, studies of disease epidemiology, stock market investment policies, capacity planning for web servers, just in time compilers for computer software, risk and reliability management of complex systems, etc. In many situations, these predictions are derived on a statistical basis from historical data using time series analysis techniques such as autoregression.

The betting industry naturally has a particular interest in making accurate predictions of the future. Commercially the most important segment of this industry relates to sports betting, especially horse racing. One attraction of such betting is that there is a certain element of skill. Thus knowledgeable betters should win money over the long-term at the expense of novice betters (with a certain amount of course being skimmed off by the bookmakers). In contrast, in lotteries and casino games based purely on chance, all punters are equal, and must lose over the long-term to the house.

A typical betting activity is for punters to study the form of each horse in a race (i.e. the placing of the horse in its most recent races). Other variables, such as the length of the race in question, the draw of the horses, the quality of the going, etc. are also considered. The relevant information about the race and previous form is available to punters from newspapers and other information services. This À .e.. be: À : .. a. À. .: À À À * information is then studied to try to predict the likely outcome of a race, thereby allowing bets to be placed accordingly.

Statistical sporting data are also of widespread interest for their own sake, s irrespective of the betting industry. For example, in tennis the performance of a player may be reflected in a variety of figures, such as percentage of aces and double faults, rate of unforced errors, maximum service speed, etc. Such data are useful to support and inform sports commentary and discussion on television and other media.

0 The past few years have seen a rapid increase in the number of ways in which sports events may be viewed, for example via conventional network television, via satellite television, via digital channels, via the Internet, and so on. Since sporting rights can be expensive, broadcasters have sought to maximise the return on a given event by providing increased analysis and discussion before and after the event. This in turn has led to a demand for facilities to support such programming, including graphics to present various statistics pertinent to the event in question. In addition, competition between different channels ensures that broadcasters are keen to provide viewers with the most informative and stimulating sports coverage.

Summary of the Invention

Accordingly, one embodiment of the invention provides a method of generating a broadcast signal. The method involves accessing multiple data sets, typically from a local or remote database, or potentially from a live data feed. Each data set comprises a sequence of data values and corresponds to a respective entity.

Note that different data sets may be retrieved from different sources, for example some data sets may be loaded from a CD-ROM, while other data sets may be downloaded over the Internet.

The method further includes generating from each data set an index value for the data set. The index value is calculated on a statistical basis from the data values in the data set. The index values are normalised so as to ensure comparability between index values from different data sets (which may contain different numbers of data values). This normalization may be performed as a separate operation on the initially À À ce. .. ec: Be:: cee ce as e. I: À: À À generated index values, may be integrated into the overall calculation of the index values.

The method further includes forming a graphic for a group of entities. The s graphic incorporates an identifier and a corresponding index value for each entity in the group. The identifier and the index value for an entity in the group are displayed in association with one another in the graphic. This is typically achieved by positioning the identifier next to the index value representation, or possibly these may be superimposed one upon another. The graphic is now incorporated into a broadcast 0 signal for transmission to television viewers via cable, satellite, conventional broadcast, and so on.

One application of the above method is where the sequence of data values represents sports results, and the entity represents an entrant in a sports competition.

In one particular embodiment, the entities represent horses, the data sets represent form histories for the horses (i.e. the placing of a horse in its most recent outings), and the identifiers represent the names of the horses. The same technology may also be used for other sports, such as greyhound racing, motor racing, and athletics, as well as team sports such as football. In such cases, the index value can be considered as an indicator of the current form of the team or competitor (as appropriate).

A particular advantage of the index values is that they provide a succinct representation of current form. This can be easily represented as a graphic, such as a bar chart, which provides the most intuitive and user-friendly form of presentation for television. This tends to increase viewer enjoyment of the programming, and may also increase viewer participation in related activities, such as betting (which might become available on-screen through interactive digital television).

A further advantage of the index values is that they are directly comparable with one another. In other words, the index values can be easily ordered with respect to one another, thereby leading to a quick and direct assessment of the relevant current strengths of the different entities (horses or whatever). Note that in contrast, it is difficult to order the original data sets per se with respect to one another, since they are multi-dimensional.

À À ee. ee. ce: .e: ae e. .: À À * The index values described herein have been found (in practice) to be a reasonably reliable indicator of form. Thus if index values are generated for entrants in a competitive event (e.g. horses in a race), then the index values have been found to provide relatively good predictions of the outcome of the event. In other words, a correlation is observed between the (pre)calculated index values and the subsequent result of events.

In a preferred embodiment, the method further includes pre-processing the 0 data prior to generating the index values. This is in order to produce homogenised data sets. For example, extreme data values may be reduced or amplitude limited, in order to avoid having undue weight on the index value. Note that this pre-processing may be omitted, depending on the quality and nature of the initial data sets.

In one preferred embodiment, calculating the index values on a statistical basis involves the use of a placement weighting function and a historical weighting parameter. The placement weighting function determines the relative weighting to be accorded to different data values, and the historical weighting parameter determines the relative weighting to be accorded to different positions in the data sequence.

Typically, the data values in the data sequence are arranged in chronological order, and the historical weighting function gives more weight to recent data values than to older data values. In one particular embodiment, the historical weighting function decreases in linear proportion to position in the data sequence, but any other suitable form of appropriate decay function could be used (e.g. exponential, reciprocal, step, etc.).

In one embodiment, the method further comprises transforming the index values to accommodate a different set of circumstances. For example, where the data values represent form histories in racing, there may be additional factors relevant to so the outcome of a particular race, such as the draw position of a horse, the distance of a race, and so on. The index values determined from previous results may then be adjusted or supplemented in order to allow for these additional factors.

À À A. e ee: À À . . . . a À À À À . À À In addition (or instead of) being incorporated into a television broadcast, the index values and associated graphics can be supplied over any other appropriate network, such as a mobile phone network and/or the Internet. Note that the detailed form and provision of the index values may be specific to the particular network in s question. For example, mobile phones have a relatively limited graphical capacity and small screen size. In contrast, most Internet clients (i.e. desktop computers) have relatively powerful facilities, and sophisticated graphical support. Thus graphics that look good on one network may need to be modified in order to look good on another network.

A further embodiment of the invention provides a method for summarising the current form of entrants into a competitive event. The method comprises accessing multiple data sets, where each data set corresponds to a respective entrant and comprises a sequence of data values. Each data value is representative of past IS performance of the corresponding entrant in competitive events. The method further comprises generating from each data set an index value for the data set. The index value is calculated on a statistical basis from the data values in the data set using a chronologically-dependent weighting. The competitive events typically comprise sporting activities, such as horse races (in which case the entrants comprise horses).

Another embodiment of the invention provides apparatus for generating a broadcast signal. The apparatus includes a processing unit for generating multiple index values for respective data sets. Each data set corresponds to a respective entity and comprises a sequence of data values. The index values are generated on a 2s statistical basis from the data values in the respective data set. The index values are normalised so as to ensure comparability between index values from different data sets. The apparatus further includes a graphics engine operable to form a graphic for a group of entities. The graphic includes for each entity in the group an identifier and a corresponding index value, which are displayed in association with one another.

The apparatus further includes a broadcast transmission system, which is operable to incorporate the graphic into a broadcast signal.

Another embodiment of the invention provides apparatus for summarising the current form of entrants into a competitive event. The apparatus includes multiple s À À ece.. c: e: ce À. i.: À ce data sets, where each data set corresponds to a respective entrant. Each data set comprises a sequence of data values. Each data value is representative of past performance of an entrant in previous competitive events. The apparatus further includes a processor operable to generate from each data set an index value for the s data set. The index value is calculated on a statistical basis from the data values in the data set using a chronologically-dependent weighting.

Another embodiment of the invention provides a computer program product comprising program instructions on a medium. The program instructions are to executable by a machine for causing the machine to access multiple data sets, each data set corresponding to a respective entity and comprising a sequence of data values, and to generate from each data set an index value for the data set. The index values are calculated on a statistical basis from the data values in the corresponding data sets, and are normalised so as to ensure comparability between index values from different data sets. The program instructions further cause the machine to form a graphic for a group of entities. The graphic includes for each entity an identifier and a corresponding index value, which are displayed in association with one another. The graphic may be incorporated into a broadcast signal, or utilised in any other appropriate manner.

Another embodiment of the invention provides a computer program product for summarising the current form of entrants into a competitive event. The computer program product comprises program instructions on a medium. The program instructions are executable by a machine, causing the machine to access multiple data sets. Each data set corresponds to a respective entrant and comprises a sequence of data values. Each data value is representative of past performance of the corresponding entrant in competitive events. The instructions further cause the machine to generate from each data set an index value for the data set. The index value is calculated on a statistical basis from the data values in the data set using a chronologically-dependent weighting.

It will be appreciated that such systems and computer program products can generally utilise the same particular features as described above in relation to the method embodiments. It will also be appreciated that the program instructions are À À act.. Be: : e: À À generally loaded from a memory into a processor for execution. Some or all of these software components may be pre-installed onto one or more hard disk units, or loaded off some portable storage medium, such as a magnetic tape, CD-ROM, DVD, etc. Alternatively, some or all of the software components may be downloaded via a transmission medium over a network. Note that software obtained from a transmission or portable storage medium may be saved onto a hard disk for subsequent (re)use by the system, or may be loaded directly for execution into system memory.

0 Brief Description of the Drawings

Vanous embodiments of the invention will now be described in detail by way of example only with reference to the following drawings in which like reference numerals pertain to like elements and in which: Figure 1 is a flowchart illustrating processing in accordance with one embodiment of the present invention; Figure 2 is a schematic diagram illustrating an example of a broadcast graphic that may be generated in accordance with one embodiment of the present invention; and Figure 3 is a network diagram illustrating the system architecture in accordance with one embodiment of the invention.

Detailed Description

Figure I is a flowchart illustrating processing in accordance with one embodiment of the invention. The method starts with receiving data (step 110) in the form of a time series of previous values. More particularly, the received data comprise multiple time series, which are then individually processed as described below. Note that the data may be received dynamically, such as from an on-line feed, or the data may be retrieved from a static source, such as a database. In addition, the data may be received from a variety of different sources, and then combined as appropriate.

Each received time series is pre-processed (step 120) on receipt in order to form a homogenised data sequence. This typically involves removing nulls from the Àe;:: Àe e'.; À.e:.e Àe data, performing amplitude limitation, etc. It will be appreciated that such pre- processing may be omitted, depending upon the quality and consistency of the incoming data set.

A filter is now applied to the homogenised data sequence (step 130), which can be regarded as calculating a form of weighted moving average. In particular, the filter is arranged to give priority to more recent values in the time sequence of data compared to older values. The output of the filter is now normalised (step 140) to give a single index value. Such normalization overcomes variations in the length of 0 the time sequence for one data set compared to another, and so provides a consistent and comparable set of index values. Note that the filter itself could be arranged if desired to produce a normalised output directly (rather than having this as a separate processing step).

The index values calculated by the normalization process typically pertain to a default set of circumstances. In some cases however, it may be desired to adjust or modify an index value so as to be more appropriate to the particular set of circumstances in question (step 150). This can be regarded as a form of bias adjustment or compensation, and results in an index value having improved accuracy.

One way of implementing this is to aggregate the index values calculated at step 140 with data representative of these additional circumstances.

The set of (modified) index values generated from the input data sequences is now passed to a graphics engine (step 160). This engine may be part of the same system used for the data processing of step 110 through to 1 50. Alternatively, the graphics engine may be part of a separate system, which receives the calculated index values over some appropriate data feed or connection.

The graphics engine then generates a graphical representation of the index values. For example, a bar chart may be used, where each time sequence is depicted by a bar whose length is dependent upon the index value calculated from that time sequence. This graphical representation is then incorporated into a television broadcast transmission (170). Note that this television broadcast may be made via terrestrial network, cable, satellite etc., and may be in the form of a conventional ce- .. Be: c. I: I: a.:' À À analog broadcast, an interactive digital television service, etc. It is also possible to use the generated graphics with a wide range of other media and networks, including the Internet, wireless and other private data networks, private broadcast networks, handheld computing devices, cellular phones and similar mobile communication devices, touch-screen devices, betting terminals and kiosks, and so on.

Although the operations shown in Figure 1 have potential applicability to a wide range of situations, it has been found that they are particularly advantageous in the realm of horse and greyhound racing. Thus in the case of horse-racing, the data 0 received at step 1 10 comprise a form history for each horse in a race. Such a form history is available from various information services, and in one implementation might be provided in the following format for two horses (named here as SOLENT and POMPEY): SOLENT: 60142-2 POMPEY: 5/232 In these strings, a number in the range 1-6 indicates a placement corresponding to that number, in other words a 1 indicates a first place finish, a 2 indicates a second place finish, and so on. A 0 (zero) indicates that the horse finished further down the field than sixth (or failed to finish at all). In addition, various symbols are used to indicate other events of interest in the data sequence (apart from places) . For example, a "-" denotes a break in racing for the horse, while a "/" denotes a change in racing (e.g. from jumps to flat). The results in the form history are presented in chronological order, with the early results at the beginning (i.e. to the left) of the string, and the recent results at the end (i.e. to the right) of the string. Thus in the above example, both SOLENT and POMPEY finished second in their last outing.

Once the above data sequences have been received, they are pre-processed at step 120. In one embodiment, this pre-processing involves removal of all non- placement indicators, such as the "-" and the "[', and replacement of the value 0 (zero) by a value of 7 (bearing in mind that a value of 0 in the form sequence indicates that the horse finished no higher in the field than seventh). Consequently, the pre r; a, ., processed (i.e. homogenized) data sequence for SOLENT now comprises 671422, and the pre-processed data sequence for POMPEY now comprises 5232.

The homogenised data sequence is now ready for filtering to calculate an index value. The intent of the index value is to provide a single number that is representative of the form history of the horse.

If we denote the (homogenized) form history of the horse as a time or data sequence: P/, P2, P3, À PN, where the suffix value increases with age, i.e. P/ represents lo the most recent placement of the horse, end pN the earliest, then we can specify the filtering procedure as:

N

I'= wi.f ( pi) (Eq. l) i=l where I' represents a preliminary index value, and: (a) Spit represents a placement weighting function to be calculated from the placements, in other words it determines the relative importance to be given to a first place compared to a second place, a third place, and so on. In one current implementation,f(p)=pi, in other words, the rank position is used directly.

(b) wi represents a historical weighting parameter for the time sequence, in other words it determines the relative importance of older values compared to more recent values. The historical weighting parameter is given by: wi=2.(T+l-i) for i=1 to T wi=0 for i>T Here T represents the maximum number of runs in the form history to be considered in deriving the index, and is set in one embodiment to T=6. In other words, only the six most recent placements from the form history are taken into consideration when deriving the form index.

Applying these formulae to the examples given above produces the following preliminary index values: 3 0 SOLENT= 122 =(2x6)+(4x7)+(6x 1)+(8x4)+( 1 0x2)+( 1 2x2) POMPEY= 100 =(6x5)+(8x2)+(1 0x3)+(12x2) at. ;: .. tIe; s. a. :. l

Here the first number in each pairing represents the placement weighting function, and the second number represents the historical weighting parameter.

Of course, it will be appreciated that other embodiments may use different formulae for the placement weighting function and/or for determining the historical weighting parameter. Various possibilities include multivariate statistical analysis techniques, such as partial least squares, multiple linear regression, reduced rank regression and the like, as well as using a neural network. These multivariate techniques are known (of themselves) to those skilled in the art, and so will not be 0 described in detail herein. In addition, other embodiments may use a somewhat different approach to that of Equation 1 to perform the filtering for calculating an index value. I In the preliminary index values derived from Equation 1, the better the performance of a horse, the lower the preliminary index value will be. Since people generally associate higher values with better performance, the preliminary index values are now inverted in order to produce a more intuitive outcome. This is achieved by subtracting a preliminary index value from the maximum possible (i.e. worst) preliminary index value that a horse could in theory have achieved. In the present embodiment, this corresponds to assuming that the horse finished below sixth; place in all its races.

For the filtering function of Equation 1, together with the placement weighting function and the historical weighting parameter defined above, the maximum possible 2s preliminary index value for a sequence of N data values is given by: 7.k(2T+l-k) (Eq. 2) where k=min(N, T). (This formula assumes that 7 is the maximum placement in the homogenized data sequence, but can be adjusted accordingly if a different maximum is used instead).

For the horse SOLENT, N=T=k=6, implying that the inverted index value is I given by: 294-122=172. For the horse POMPEY, N=4, T=6, k=4, implying that the inverted index value is given by: 252-100=152.

'Àe.. ee: I À .. À : .. .e À .: Àc À Note that these two (inverted) index values are still not directly comparable, since they have been calculated over different lengths of form history - i.e. from time sequences containing different numbers of data values. Accordingly, the inverted index values are now normalised to a percentage value. This can be done by s multiplying the inverted index value by 100, and dividing by the maximum possible preliminary index value as calculated above (see Eq. 2). The result is a final index value that lies on a scale of 0-100. (As previously mentioned, this normalization procedure can be integrated if so desired into the filter calculations).

0 Applying the above normalization process to the two example horses gives the following results: SOLENT= 58% (=172. 100/294) POMPEY= 60% (=152.100/252) (Note that in both cases, the index percentage value has been rounded down to the nearest whole number). These two index values therefore imply that POMPEY has slightly sponger current form than SOLENT, but there is relatively little to choose between them.

Once the index values for form history for a set of horses in a race have been determined (as above), they can be utilised in any appropriate manner. In one particular implementation, they are fed to a graphics engine which generates charts and other graphics for use in broadcast television. The output from such an engine is shown schematically in Figure 2, which lists by name the horses in a particular race.

Associated with each horse is a bar whose length corresponds to the index value derived from the form history of the horse.

In the particular race shown in Figure 2 there are five runners, and their index values are as follows: Bassett 54, Pompey 62, Shirley 45, Solent 60, and Totton 72.

Note that this graphic may be superimposed over live background pictures of the horses at the racetrack or any other suitable images. Furthermore, the graphic will typically use colour and other design features to improve clarity and visual appeal.

For example, the bars may start off blue on the left, and become increasingly red to the right to indicate increasing levels of success.

:.. ee..: ee. ces; À :..

The generation of an index value such as shown in Figure 2 has two main benefits compared to working with full form histories. Firstly, the index value represents a highly compressed indicator of form compared to the full form history, and secondly the index values are directly comparable with one another. This is s especially significant in a medium such as television, where clarity and visual impact are extremely important. Thus the graphic of Figure 2 provides a compact and succinct visual summary of form history that can be readily accommodated within a standard television screen.

lo In contrast, it is difficult to represent conventional form histories as anything other than a string of numbers. These can be hard to interpret for those not familiar with horseracing, and indeed may hinder newcomers from becoming interested in the sport. In addition, the form strings are not directly amenable to display in graphical form, which is normally the most intuitive mode of presentation for humans and the one having highest impact over a primarily visual medium such as television.

A further problem with form swings per se is a lack of direct intercomparability. Of course, if a horse has won all its previous races then it is natural to expect it to beat a horse that has previously always come third. In practice however, the situation is usually far less clear- cut, and horses under consideration may have had mixed results in previous races. In these circumstances, the provision of a single (one-dimensional) and user-friendly measure of form history provides a readily understood characterization of current form that is particularly attractive for those new to the sport. 2s

It will be appreciated that Figure 2 provides just one example of the potential graphical display of index values for horses, and the skilled person will be aware of many further possibilities for suitable graphics. For example, the index values may be presented numerically in conjunction with (or instead of) the corresponding bar. Alternatively, an animation of a horse might be used, where the speed and

progress of the animated horse are proportional to the index value.

The index values and associated graphics such as shown in Figure 2 can be utilised for a wide variety of purposes, including entertainment, competitions and he eese À e e e ee e e e e e e e e eve wagering. The index values quickly indicate the likely contenders based on current form, how close a contest might be, and so on. Hence the index values provide a sports analytic to inform commentary and fuel debate. The index values and graphics are also a useful addition to the betting industry, since they provide information of interest to both punters and the bookmakers themselves. The former can use the information as input to their betting decisions. The latter may potentially use the information to help setting race odds, one attraction here being the fact that the index values can be generated automatically (race odds have conventionally been set by hand in a rather labour intensive process).

It will be appreciated that there is a very wide range of potential distribution mechanisms for the index values and associated graphics, depending upon the target audience. They are especially suited to sports programming on television, but can also be used in a wide range of other media and distribution networks, such as newspapers, the Internet, DVDs, computer games, mobile phones, and so on.

Although the index values are especially useful for predicting or speculating upon the outcome of particular events, competitions, races, etc., they can also be used as a form of relative measurement. For example, in certain sports which are largely based on individuals (e.g. golf and tennis), there are also some team competitions (e.g. the Ryder Cup and the Davis Cup for golf and tennis respectively). The index values might then be used as a guide for team selection purposes by comparing the current form of various potential team members.

The index values illustrated in Figure 2 relate to form history only. However, a more accurate forecast for a particular event or situation may be made by taking into consideration specific factors relating to that event or situation (this corresponds to the bias adjustment of step 150 of Figure 1). Thus in the case of horse racing, such factors might include distance, the type of going, left-hand or right-hand course, jockey form, trainer form and so on. In tennis such factors might include surface, maximum duration of matches (3 or 5 set), indoor or outdoor, and so on. These additional factors can be combined with the form history in order to produce a final forecast for the event in question.

:e c.: e.e ce; ee:. :.

A typical mechanism for accommodating these additional factors is to look for a statistical correlation between the various factors in question and results from previous races. This correlation can then be used to perform the relevant adjustments to the forecasts (if no such correlation is found, the factor(s) in question can be s ignored). In one particular embodiment, the final forecast is achieved by adding together values for a variety of factors, such as form history of the horse, recent jockey performance (which could itself be represented as a form history), distance, and so forth. The contributions of the various factors to the final overall forecast may be weighted as appropriate. In this case, the graphic produced may look similar to 0 that shown in Figure 2, but reflect multiple criteria (not just form history).

Figure 3 is a diagram illustrating a system architecture in accordance with one embodiment of the invention. The main processing of form histories to create the index values is performed on computer 303. In one embodiment, this is a personal computer running the Windows operating system, and the index values are derived from the form history using the Excel spreadsheet program (Windows and Excel are trademarks of Microsoft Corporation). Nevertheless, it will be appreciated that other hardware and software platforms could be used instead for performing this processing. For example, computer 303 might be a laptop, a handheld system, a server, a minicomputer, a mainframe, a dedicated microprocessor or an embedded ASIC (application specific integrated circuit). Computer 303 may also represent a distributed system, with different portions of the processing being performed on different units.

2s The spreadsheet for determining the index values may be accessed by computer 303 from a local storage device, such as a hard disk drive, which may in some cases include a removable storage medium, such as a CD- ROM. Alternatively, the spreadsheet may be loaded into the memory of the computer 303 over a network 302, which may comprise a local area network (LAN), the Internet, or any other appropriate form of wired or wireless data link. The data (e.g. form histories) for use in the spreadsheet (if not already incorporated into it) can be obtained from a variety of locations, including database 301 (accessible over network 302), a live results data feed 304, or a removable storage medium such as a floppy disk, CD-ROM or DVD insertable into computer 303.

c. À .. Àe Àe. .e À . . . Once the computer 303 has computed the index values (as described above), these may be passed to broadcaster 310. In some cases the index values may be routed via a third party (not shown in Figure 3) that transforms the numerical index s values into a corresponding graphic, such as illustrated in Figure 2. Alternatively, this transformation may be performed by broadcaster 310 itself, or as part of the processing on computer 303.

The broadcaster then combines the form history graphics with the rest of its 0 horse-racing coverage as appropriate, and transmits the resulting images over one or more networks. For example digital and/or analog television signals may be sent over terrestrial networks via aerial 312, as well as via satellite disk 314, or over a cable network (not shown in Figure 3).

s The index value graphics may be received in a variety of formats and on a range of devices. Thus Figure 3 illustrates a television 326, a handheld device 324 and a cell phone 322 as potential recipients of the index values (note that many 2G and 3G mobile phones now support image reception). It will be appreciated that there may be various modes of network connection to these different devices from different Jo information suppliers. In addition, broadcaster 310 may not necessarily supply all of cell phone 322, handheld system 324 and television 326.

In one embodiment, television 326 may receive public broadcasts, typically for home viewing. Alternatively, television receiver 326 might be connected to a 2s private network and installed in a beuing shop for viewing a dedicated (internal) channel of the relevant beuing agency. Cell phone 322 might be able to access the index values by dialling a special number, and having an image corresponding to Figure 2 downloaded onto the cell phone.

Figure 3 also illustrates the provision of the index values (in numerical and/or graphical form) to server 330. These data and/or images can then be made available by server 330 over the Internet 332, for access by a wide range of client devices (not shown in Figure 3), as known to the person skilled in the art. Note that server 330 may be maintained by broadcaster 310 or by some different organization.

À À . . Be. .:: s-.e Be: À.e À:. À .

It will be appreciated that there are a wide range of uses for the index values and graphics generated as described above, both inside and outside broadcasting. For example, they may be incorporated into CDs, videos, and DVDs, typically as spin offs from the television forecasting, or used in computer games (whether on-line or off-line).

Furthermore, although the above discussion has focussed on the analysis of form histories for horse-racing, it will be appreciated that the approach described lo herein has a broad range of applicability. Thus various programming examples of the potential use of index values and graphics include: (a) summarising and predicting current form in a wide range of sports, including football, tennis, golf, rugby, motor racing, cricket, athletics, etc. (b) determining or estimating popularity of various contenders in "reality" TV shows, is where the time sequences represent previous votes cast for the contenders; (c) summarising and predicting traffic congestion on various routes, based on previous levels of congestion; (d) summarising and predicting meteorological conditions such as rainfall and temperature at various locations.

In conclusion, a variety of particular embodiments have been described in detail herein, but it will be appreciated that this is by way of exemplification only.

The skilled person will be aware of many further potential modifications and adaptations that fall within the scope of the claimed invention and its equivalents.

Claims

e. ee.

e c e e e e e ce À e e e e c e e À e e are e Claims 1. A method of generating a broadcast signal comprising: accessing multiple data sets, each data set corresponding to a respective entity and comprising a sequence of data values; generating from each data set an index value for the data set, wherein said index value is calculated on a statistical basis from the data values in the data set, and wherein said index values are normalised so as to ensure comparability between index lo values from different data sets; forming a graphic for a group of entities, said graphic including for each entity in the group an identifier and a corresponding index value, wherein the identifier and the index value for an entity in the group are displayed in association with one another; and incorporating said graphic into a broadcast signal.
2. The method of claim 1, further comprising pre-processing the data prior to generating the index values in order to produce homogenised data sets.
3. The method of claim 1 or 2, wherein calculating the index values on a statistical basis involves the use of a placement weighting function and a historical weighting parameter, wherein the placement weighting function determines the relative weighting to be accorded to different data values, and the historical weighting parameter determines the relative weighting to be accorded to different positions in a data sequence.
4. The method of claim 3, wherein said data values are arranged in a chronological sequence, and said historical weighting function gives more weight to recent data values than to older data values.
5. The method of claim 4, wherein said historical weighting function decreases linearly with position in the data sequence.

À: Be. Be.. À c: : :. : : : À À . À À .
6. The method of any preceding claim, further comprising transforming the index values to accommodate a different set of circumstances.
7. The method of any preceding claim, wherein said graphic comprises a bar chart.
8. The method of any preceding claim, wherein said sequences of data values represent sports results.

lo
9. The method of claim 8, wherein said entities represent horses, and said data sets represent form histories.
10. The method of any preceding claim, further comprising making the index values available over a mobile phone network.
11. The method of any preceding claim, further comprising making the index values available over the Internet.
12. A method for summarizing current form of entrants into a competitive event comprising: accessing multiple data sets, each data set corresponding to a respective entrant and comprising a sequence of data values, each data value representative of past performance of said entrant in competitive events; and generating from each data set an index value for the data set, wherein said index value is calculated on a statistical basis from the data values in the data set using a chronologically-dependent weighting.
13. The method of claim 12, wherein said index values are normalised so as to ensure comparability between index values from different data sets.
14. The method of claim 12 or 13, further comprising forming a graphic for a group of entrants, said graphic including for each entrant in the competitive event an identifier of the entrant and a corresponding index value, wherein the identifier and the index value for an entrant are displayed in association with one another.

À À À À À he e À À

C À À

À C
15. The method of claim 14, further comprising incorporating said graphic into a broadcast signal.
16. The method of claim 14, further comprising making the graphic available for download onto a mobile phone.
17. The method of claim 14, further comprising making the graphic available for download over the Internet.
18. The method of any of claims 12 to 17, further comprising using the generated index values for determining odds with respect to the entrants winning the competitive event.
19. The method of any of claims 12 to 18, wherein generating the index values on a statistical basis involves the use of a placement weighting function and a historical weighting parameter, wherein the placement weighting function determines the relative weighting to be accorded to different data values, and the historical weighting parameter determines the relative weighting to be accorded to different chronological positions in a data sequence.
20. The method of any of claims 12 to 19, wherein said competitive events comprise horse races and said entrants comprise horses.
21. Apparatus for generating a broadcast signal comprising: a processing unit for generating from multiple data sets an index value for each respective data set, wherein each data set corresponds to a respective entity and comprises a sequence of data values, wherein said index value is generated on a statistical basis from the data values in the data set, and wherein said index values are normalised so as to ensure comparability between index values from different data sets; a graphics engine operable to form a graphic for a group of entities, said graphic including for each entity in the group an identifier and a corresponding index at. À c. cec. À. .e À À value, wherein the identifier and the index value for an entity in the group are displayed in association with one another; and a broadcast transmission system operable to incorporate the graphic into a broadcast signal.
22. The apparatus of claim 21, wherein the processing unit incorporates a pre processor for homogenizing the data sets prior to generating the index values.
23. The apparatus of claim 21 or 22, wherein generating the index values on a lo statistical basis involves the use of a placement weighting function and a historical weighting parameter, wherein the placement weighting function determines the relative weighting to be accorded to different data values, and the historical weighting parameter determines the relative weighting to be accorded to different positions in a data sequence.

IS
24. The apparatus of claim 23, wherein said data values are arranged in a chronological sequence, and said historical weighting function gives more weight to recent data values than to older data values.
25. The apparatus of claim 24, wherein said historical weighting function decreases linearly with position in the data sequence.
26. The apparatus of any of claims 21 to 25, wherein said graphic comprises a bar chart. 2s
27. The apparatus of any of claims 21 to 26, wherein said sequences of data values represent sports results.
28. The apparatus of claim 27, wherein said entities represent horses, and said data sets represent form histories.
29. Apparatus for summarising current form of entrants into a competitive event comprising: À À À À :. a:: À. as:: Àe.:.

multiple data sets, each data set corresponding to a respective entrant and comprising a sequence of data values, each data value representative of past performance of said entrant in competitive events; and a processor operable to generate from each data set an index value for the data set, wherein said index value is calculated on a statistical basis from the data values in the data set using a chronologically-dependent weighting.
30. The apparatus of claim 29, wherein said index values are normalised so as to ensure comparability between index values from different data sets.
31. The apparatus of claim 29 or 30, further comprising a graphics engine operable to form a graphic for a group of entrants, said graphic including for each entrant in the competitive event an identifier of the entrant and a corresponding index value, wherein the identifier and the index value for an entrant are displayed in association with one another.
32. The apparatus of claim 31, wherein said graphic is incorporated into a broadcast signal.
33. The apparatus of claim 31, wherein said graphic is made available for download onto a mobile phone.
34. The apparatus of claim 31, wherein said graphic is made available for download over the Internet.
35. The apparatus of any of claims 29 to 34, wherein said competitive events comprise horse races and said entrants comprise horses.
36. A computer program for implementing the method of any of claims 1 to 20.
37. A computer program product comprising program instructions on a medium and executable by a machine for causing the machine to: access multiple data sets, each data set corresponding to a respective entity and comprising a sequence of data values; c.e;: e... À:e À À . generate from each data set an index value for the data set, wherein said index values are calculated on a statistical basis from the data values in the data set, and wherein said index values are normalised so as to ensure comparability between index values from different data sets; form a graphic for a group of entities, said graphic including for each entity in the group an identifier and a corresponding index value, wherein the identifier and the index value for an entity in the group are displayed in association with one another; and incorporate said graphic into a broadcast signal.
3 8. A computer program product for summarising current form of entrants into a competitive event, said computer program product comprising program instructions on a medium and executable by a machine for causing the machine to: access multiple data sets, each data set corresponding to a respective entrant and comprising a sequence of data values, each data value representative of past performance of said entrant in competitive events; and generate from each data set an index value for the data set, wherein said index value is calculated on a statistical basis from the data values in the data set using a chronologically-dependent weighting.
39. A method for generating a broadcast graphic from multiple data sets substantially as described herein with reference to the accompanying drawings.
40. A computer program for generating a broadcast graphic from multiple data sets substantially as described herein with reference to the accompanying drawings.
41. Apparatus for generating a broadcast graphic from multiple data sets substantially as described herein with reference to the accompanying drawings.