Computer Systems

[Artemis]

I think the main reason computerised (and manual) systems based on ratings do not work so well is because a lot of people, including layers and odds compilers, are using them. This causes the selections to be overbet and condemns their users to have to suffer poor value. Most of the providers of  form-based ratings tend to pinpoint the same horses as being the best- in or well- handicapped. Also, the additional factors such as C, C/D, D, trainer%, jockey%, wins/runs%, etc. are also heavily taken into account in most systems, again causing some horses to be overbet.

There is no easy way, at least none that occurs to me, of getting round this problem. I think you need to get your own ‘handle’ on these factors and perhaps try different approaches. Also, I feel that the market also needs to be factored in – I understand that professional syndicates who bet on pari-mutuel systems always include the market (i.e. the collective opinion of all backers) in their computer models.  

[LookingForAWinner]

Some interesting points there Artemis.

Perhaps with the right historical information you can use reverse logic.  For example, say that wins/runs% is a strong factor in Novice Chase races, but bears very little correlation in Handicap Hurdles, then you can either:<br>a) Vary the influence of a factor under certain conditions<br>b) Avoid betting where your factors perform worst<br>c) Make a bet based on a more unusal factor that exerts a stronger influence.

In the past I have included betting forecasts in my systems, but I didn’t find this factor influential.  Where it does play a part is if there is insufficient information about a horse, but the forecast reflects other factors such as pedigree or trainer/jockey whispers.  These all play there part, but it depends how much value you place on them.  Nowadays, I use the forecast to highlight unknowns and then make a subjective decision.  This does spoil the hard fact route, and it is also difficult to record for future analysis.  If you were to use market values closer to the race start, then desk jockey’s like myself who only have a lunchbreak browse will miss out.

<br>Hope this adds more thoughts into the mix.

Looking For A Winner.

[JAR]

Hi<br>That’s why I like new Boxform you can think of any factor you like and give it a numerical value depending on how strong an influence you think it may have on the outcome of a race.<br>For instance you could rate a newspaper experts opinion according to his choice of words and factor this into calculations or you could rate a horse on a scale of 1 to 10 whether or not in your own opinion the market supports a horse or not.<br>If you felt pretty horses win races you could mark a horse on how pretty it is .  Any factor could be used and at the end of the day you have a record of points given per horse to see if the factor chosen works or not.<br>It is up to you whether you choose factors everyone else uses and so find little or no value – or are able to discover unique factors and find some value.

I believe that today virtually no horse is giving away more than 5 to 10% value to the punter and this is totally eaten up by the bookmakers margins overall, so in fact no one can win in the long term.   Most people who think they are winning are only winning short term and will lose in the long run.  A gloomy view I know but many are happy to bet just to break even and it is these people that keep bookies in business.

Kind regards

[Artemis]

Jar,

Nick Mordin, writing in the Weekender this week has some interesting observations about coumputerised systems. Briefly, he states that many professionals (presumably including bookmakers) are using them and that they tend to come up with similar results. He goes on to say that this inevitably leads to more efficient markets and less value for people using such systems.

[LookingForAWinner]

It appears that successful punters need to find an angle that other people haven’t cottoned on to yet.  Perhaps we are not just analysing the horses, but other punter strategies in order to find a niche.

Looking For A Winner.

[JAR]

Hi Artemis<br>What Nick says is true.  However the same could be said of using the Racing Post .  Many people use this paper including professionals.  They tend to come up with the same results and this affects value.<br>The best way to use computer programs is either write your own and try and design it so it picks horses others do not.<br>Otherwise find a computer program such as Boxform that lets you use a mixture of inputs of your own choice and try and use inputs others are not using.<br>Using this approach though requires a period of paper testing to ensure the inputs you choose actually produce profitable predictions. In most cases finding an angle others do not use will result in a lower strike rate but hopefully more profit.  The bugbear may be the longer losing runs.<br>Kind regards

[LookingForAWinner]

John,

I have never used Boxform.  Is there any way to load historical data to help assess new systems?

Looking For A Winner.

[JAR]

Hi<br>Not with Boxform at present.  The data must be hand input similar to many earlier computer programs.

It simply allows you to choose an influencing factor on a race.  Say we think a horse has more chance in a particular because it is a beaten favourite Boxform would allow us to award 0, 1 or more points for this to the eventual calculated total.  It also allows you to use from 1 to 5 handicap ratings in its calculations.<br>It has to be said it is a fairly simple system , although I like it because it allows you to assign different value to some of the factors that affect a race.<br>Hope that helps.

Kind regards

[JAR]

Hi <br>I’ve just noticed that the computer program " Compunter" does not seem to be advertised anymore and the web sites associated with this product cannot be found.

I don’t know if  the author of the program has either retired or passed on.

Seems a shame that the program is no longer available as it was a great little program to use with a Racing Post racecard.  I liked the MSDOS version because it used large print which is easy on the eyes for a TOG like me.<br>

(Edited by JAR at 11:44 am on April 27, 2004)

[LookingForAWinner]

JAR,

What is a TOG?

[JAR]

Hi

Terry’s  Old  Gentlemen  

re Terry Wogan’s Morning Radio Show.:biggrin:

[Artemis]

<br>Do I come here often?

[JAR]

I wonder if there is a bookmaker in heaven . :)

[LookingForAWinner]

There certainly will be someone asking us to settle our account!

[boylee]

Hi, when construting a system instead of form figs would it be better to use lenths beaten? I use a little system of compairing the class, value & race class of last 2 runs against todays and taking into acount the distance behind the winner and differance in weight.

                        Barry

sorry about the spelling & english

[LookingForAWinner]

Barry,

I don’t see why not.  Many software systems use speed ratings rather than form figures so there are a few options to tackle.

I suppose it is better to look at attributes that the general racing public ignore so that you can gain an advantage.

How is your system performing?

Looking For A Winner

[boylee]

Hi LFAW, I only use it a little but the system is sound it depends on how you rate the classes. My health stops me getting too involved, it is part of a method I used to use.

             Barry

[Author]

Posts