A Brief Explanation of Ratings

All ratings offered within GTX are generally derived from historical form. However, each uses a different methodology based on an individually developed model. Most ratings use an age old method of identifying what is called a Qualifying Run (QR) where the program selects a past run likely to be representative of what could reasonably be expected of that horse today. Each rating process has different criteria for selecting the QR.


GTX Ratings
 


GTX Ratings are generated from a formula and do not rely on any other variables such as bonuses or penalties to be added or subtracted. Many of these variables will already have been used in the formula to create the rating.

The GTX Rating formula was derived using logical rules entered into what is called an ‘Expert System’ and optimised using Artificial Intelligence.

A couple of published definitions of an ‘Expert System’...

  • ‘Expert knowledge is a combination of a theoretical understanding of the problem and a collection of heuristic problem-solving rules that experience has shown to be effective in the domain. Expert systems are constructed by obtaining this knowledge from a human expert and coding it into a form that a computer may apply to similar problems.’ [George F. Luger, Artificial Intelligence: Structures and Strategies for Complex Problem Solving, 4th Edition, 2002. Addison-Wesley. Section 1.2.3 - Expert Systems]
  • ‘An expert system is a computer program that simulates the judgement and behaviour of a human or an organization that has expert knowledge and experience in a particular field. Typically, such a system contains a knowledge base containing accumulated experience and a set of rules for applying the knowledge base to each particular situation that is described to the program. Sophisticated expert systems can be enhanced with additions to the knowledge base or to the set of rules’ [www.whatis.com]

The GTX Rating formula was replaced in July 2005 with a new and enhanced algorithm which now uses some 60+ variables in several parameter sets, for example different weighted values used for each variable for say, 2yo races and hurdle races. Unlike the former GTX Rating formula, the new formula gives a rating to horses having their first career start where information is drawn from the sire’s history, the trainer’s history, the jockey and several other known factors.

Additionally, the GTX rating is dynamically altered when track conditions are downgraded or when jockeys are replaced.

The GTX Market Parameters, located under Setup>Ratings (Extended), include a parameter called 'Market Distribution'. This setting can be altered to suit the user but it is important to note that altering the ‘distribution’ will alter the GTX Net Ratings assigned to each horse and also the difference in Net Ratings between each runner in a race.

Formline Ratings
 

Weight Ratings are sometimes referred to as a ‘Class Ratings’. This method was originally developed by renowned racing author and ratings expert, Rem Plante in the early 1960's and published in his book ‘Australian Horse Racing and Punter's Guide’.

Formline Ratings are expressed in kilograms, generally within the range of 10 - 70kgs. Each kilogram equates to approximately 0.75 lengths, but is variable by distance of race.

Whilst entering race results, expert form handicapper Marcel Plante assesses performances of horses against expectations to determine an accurate Class Value for that race. This is then locked in against each historical run.

The GTX program selects a Formline QR (qualifying run) or what is in effect considered to be the most suitable historical run from the last 16 form runs. The value stored against this run determines the Base Rating. Occasionally the program selects a second QR and makes an adjustment between the two (QRs) within the ‘Other’ adjustment. If default settings are used, the program starts with the base rating and subtracts the weight over the limit, (WOL) then adds a jockey allowance, consistency bonus, subtracts a jockey penalty and either adds or subtracts a value to compensate for the acceleration/deceleration effect (ADE), and a value for ‘Other’, which can be a composite of several variables such as expected improvement and, as indicated above, an adjustment between two QRs


Trakline Ratings
 

These are form-based and time-influenced ratings that are calculated at the post race results stage by an artificial Neural Network and manually adjusted by a form handicapper. Each historical run has a rating assigned to that run. The GTX program selects a QR (qualifying run), or what is in effect considered to be the most suitable historical run from the last 16 form runs. The value stored against this run determines the Base Rating.

Ratings are expressed in ‘points’, generally within the range of 0 - 70 points. Each point equates to 1/10 second, or 0.65 lengths, but is variable by distance of race. The GTX program selects a Trakline QR (qualifying run) from what is considered to be the best of (generally) the three most recent runs. If the horse is resuming from a spell the QR is chosen from a previous first up run. The value stored against this run determines the Base Rating. If default settings are used, GTX starts with the base rating then adds a consistency bonus, subtracts a jockey penalty and either adds or subtracts a value to compensate for the acceleration/deceleration effect (ADE), and a value for ‘Other’, but does not add the weight over limit or the jockey allowance. The consistency bonus is not added for two-year-old races.

Base ratings for first starters are derived using a formula based on jockey penalty and class value.


Trakform Ratings
 

These are pure time-based ratings derived by a process of comparison of each horse’s runs against a standard (a par time) for each track and distance-adjusted for track variant. Ratings generally fall within the range of 25 - 70 points. Each point equates with 1/10 second, or 0.65 lengths, (but is variable by distance of race). The GTX program selects a Trakform QR (qualifying run) from what is considered to be the best of (generally) the three most recent runs. If the horse is resuming from a spell the QR is chosen from a previous first up run. The value stored against this run determines the Base Rating. If default settings are used there are no bonuses or penalties.

Base ratings for horses having their first race start are derived using a formula based on jockey penalty and class value whereby the stronger the Class Value (CV) and the better the jockey, the higher the rating.

 

Winform Ratings
 

 

Time Rating calculated using a process and formula developed by Garry Robinson of Winform Publishing. For information on Winform Ratings please contact Garry on 02-4950 1747 during business hours.

Horses having their first starts are unable to be rated by Winform Ratings.


Combined ratings
 

As indicated by the name, Combined Ratings are derived by the combination of the values of any two or more of the above ratings (but not including GTX Ratings) in user-specified percentage weightings. Possibilities range from combining two rating methods in a 50/50 or other percentage, to using four of the currently available ratings in any percentage mix.

Early analysis of multiple combinations revealed a most effective combination to be:

35% Trakline
35% Formline
15% Trakform
15% Winform (an optional subscription rating)

One user reported good figures over several years (excluding jumps races and races with first starters) using the following mix:
60% Formline
40% Trakform

Warning : When creating a Combined Rating do not tick the option ‘Rating is tuneable’ unless you understand the ramifications. This option was added at the request of one user who wanted the combined rating graphs to reflect the combined rating for every historical run and then fine-tune (adjust) the base predictive rating using the past ratings as a guide.

If you choose this option you must fine-tune the base rating for every runner in every race, manually, as each horse is rated with the Formline base rating only. This base rating is the starting point for adjustment. If you are not prepared to do the work of manual tuning, then the Tuneable Combined Rating is not for you.


Q Ratings
 

These ratings are publicly available information published on the Unitab web site. GTX enables you to view the ratings, and test them through the System Developer, not only as a rating value which generally ranges from about 70 points to 100 points, but also as a probability and a price.

In its native form, this rating does not include any bonus or penalty adjustments. However, the GTX user is free to experiment with this rating by adding the bonuses and penalties within GTX, such as Jockey Penalty, ADE, Consistency, and Barrier Penalty. It is therefore possible for the user to create a new rating that is not used by any other user. There is nothing to stop you from assigning a value to the ADE, although it is not really ‘ADE’ for ratings without QR's.


Consensus Ratings
 


Originally started as a Ranking, Consensus Ratings are now ratings derived by assigning values from -10 to +10 to a number of ranking-type variables according to the user's perceived (or tested) value of relative importance. The Consensus Ranking is now a full Rating derived solely from the values assigned to each ranking, and as such is expressed as a probability and a price.

The user may combine all or any of 20 rankings including the TAB price rankings and derive a net number for each horse which is then treated by GTX in exactly the same way as any other net rating, after the program references the Rating Parameters for the user's instructions. Traffic Rank can not be included, as the Consensus Rating can itself be a component of the Traffic Rating.


Traffic Lights Ratings
 

Originally conceived solely as a visual indication of whether or not a runner should be backed, Traffic Lights is now a powerful user-definable rating which is arguably the single most powerful feature in GTX.

Traffic Lights setup enables the user to create a personalised rating based on that user's personal process of thinking, and it has the added feature of each specific parameter being able to be configured by the user to indicate the relative importance of that parameter. Those users who have the System Developer have the added power of being able to test the independent value of almost all Traffic Light parameters, create the Traffic Light Rating with the chosen parameters and then test the rating over historical data.

With experience and the benefit of information derived from the use of the System Developer this tool will prove to be invaluable for considering a likely outcome. Additionally, when viewing a race in your favoured rating order the user can gain a different perspective by viewing the Traffic Lights which will graphically demonstrate the degree to which a horse passes your rules

The Traffic Lights have a considerable advantage over a system, in that a system rejects a horse unless all rules are satisfied, whereas the Traffic Lights calculate the horse which best meets the user’s criteria, giving a quick visual indication as well as a detailed numeric value if required.

To create the rating, the user may configure any of the 235 available parameters, including the TAB price rankings. A net value for each horse is derived, which is treated by GTX in exactly the same way as any other net rating, in that a probability and price are calculated.


Text Size : A | A | A | A

Photo Finish © TRB.COM.AU Pty Ltd 2003