computer rating system

I have developed a computer rating system based on a load of historical winning statistics. I enter each horses form details etc. The system then rates each horse as to it's chances of winning.

The selection results of this work over a couple of years has been interesting.

Top rater wins 1 in 4 races and pays average $3.10 so loses money.
second rater wins 1 in 6 races and pays average $6.50 so slowly makes money.

Now for the real eye opener I spotted when reveiwing past results for trends etc.

Bottom raters #3 & #4 producer some big longshot winners. Since begining August those 2 are 75 units up. Past 2 years returned 152% on turnover.

The only explaination is those horses are underbet according to their chances of winning. Long may it last.

Another topic, with many computer based systems in use which now show loses, could it be that the odds available are much lower now than before for favourites? If their strike rate is the same now as before,then that may be the reason for losing money. Of course they could of been marketed with bogus return figures as well.

Horse cents

[PunterPete]

I have slowly been developing a horse racing database and analysis system for Sydney Metropolian Races only over the past 3 years.

I use previous times and last 600 times graphed for each horse to pick winners. One thing I have found is that the pace of the race is everything, and working out the pace of an upcoming race is sometimes difficult. The system works best for 1400-2000m races.

My engineering and software background has helped me make profiles for each Sydney track for horses stepping distance or changing to other tracks.

I thought it would be simple when I started but I have spent countless hours trying to perfect something that will never be perfect.

I have recently included a bet recorder in my program which I use to record every bet I have with the race data so I can find why I'm failing. I am only punt small amounts (under $20 or less bets) and my biggest return was a $78,000 superfecta of which I receive 1.11%. In my 3 years of analysis I have only found 3 horses I was sure would win and they all did with ease.

My program includes audio of each race, stewards reports, and every bit of horse and race data available. I have day time factors, stepping gradients for each track, Location factors, Weight factors, Jockey factors.

Can anyone give a hint at what I could be doing wrong.

[PunterPete]

Is there anyway I can submit a picture (jpg etc) of a sample analysis page of my program to this forum for comment?

[matt]

Pete, there is no room on this site for people to upload images. You can link to an image as long as it is on a web page somewhere. If you don't have your own web site there are plenty of free ones available (e.g yahoo or homestead).
Let's say the url of yorr site is 'yoursite.com' and the image you want to link to is 'yourimage.jpg'.
Then to put a link to your image add the following code to your post:
< IMG SRC="http://yoursite.com/yourimage.jpg" >

Cheers,

matt.

<font size=-1>[ This Message was edited by: matt on 2001-09-27 11:29 ]</font>