PDA

View Full Version : Horse Racing Systems Through Racecard Analysis



mathare
1st February 2005, 13:13
Members will have seen that I have recently published some pretty hardcore research on a horse racing system devised by BigC in the members section of the forum. As Plater commented in that thread I have an analytical mind.

I have been working on a number of football systems also, and some of you have seen the raw data I have been working from and anyone following my Method Madness bets has seen the results.

Football is nice and easy to work with. There are only three runners in each race (home win, draw, away win) and margins of victory are recorded as a matter of course and readily available. A quick check in the papers and you will see the likes of Team A 1 - 0 Team B. There is also data on shots, corners, possession etc available online so football systems are easy to generate as they allow for simple data manipulation. I can use fixture lists, historical results and league tables to perform all sorts of analysis.

So, with that lot in mind I started thinking about horse racing systems. Can anything similar be done for the sport of kings?

Suppose for a minute I have each of the day's racecards in simple to use Excel format (my format of choice for numerous reasons). What can I do with it? What trends can I look at? What analysis can I perform?

My lack of ability to tip horses is getting to me and it is something I want to address but I don't have the time, inclination or ability really to read through formlines, comparing horses recent runs across different grades and on different grounds. I want to write a spreadsheet or computer program that can do some of that for me. I have a few questions I'd like some opinions on:

1) Does the racecard in it's raw form (say the ones on Sporting Life rather than the more verbose ones on RP) contain enough data to gain any sort of insight into how a race ought to turn out?
2) If not, what else would I need to know?
3) Am I wasting my time going down this road? RSB and RaceXpert have all the stuff I am after and more. Should I bide my time until I have reached the point in my gambling career where I feel I can afford to spend cash on the software?

All thoughts welcome...

mathare
1st February 2005, 13:18
Now I have lain down the problem let me chip in with some thoughts of my own:

Football fixtures and results are in the same format (almost). Fixtures are written Team A v Team B and results as Team A 1-0 Team B. Close enough to make analysis dead easy. Horse racing fixtures (racecards) are in a very different format to the results normally so analysis is harder.

I think I may be wasting my time :(

I don't have a clear view on what is possible so I don't know what I am trying to achieve by thinking of going down the Excel analysis route. I would need to build up databases of results with dozens of variables for each race and runner. This has all been done before by RSB and RaceXpert and so on. It's not worth re-inventing the wheel is it?

Workshy
1st February 2005, 13:22
I'd be happy to throw up ideas aswell as have a go myself Mat. Its something that has always crossed my mind but I'm put off by the time need. Now if Excel were to cut corners for me :D

I'd definately look at topspeed/RPR (last 6 & course/distance/going).

http://www.racingpost.co.uk/horses/rpr_ratings.sd?race_id=367977

mathare
1st February 2005, 13:36
Ideas is what I am after and that's a good start Worky :)

Obviously speed ratings matter more in sprints than distance races so we'd need to set some rules on how much weighting to give the speed figures related to distance etc. I like the thinking though.

I'm beginning to Wonder if Excel is best though. Is this more a database thing?

Theoretically though, if I can get the data into Excel or a database you can have hundreds of shortcuts to systems in place. It could quickly tell you which horses meet certain criteria etc.

eastyorkshireracing
1st February 2005, 13:45
Hi mathare,

To work the systems you will need to use excel, but to compile histories etc, you will need to use Access (or some other DB software that works with excel). Pref work the two together.....

Good Luck!

Gordon

mathare
1st February 2005, 13:45
RaceXpert is about £75 plus £6.50 a week for data updates if you order 40+ weeks. I make that over £400 for the first year (£350 if the ten weeks of updates come after the initial purchase). I am a long way of spending that each year for the purposes of gambling so I am looking to develop something on my own for now. It may not be as good but it will be cheap :) Plus it will give me something to play with (oo-er!) and hone my coding skills.

I won't be starting anything for a couple of months I reckon but all ideas gratefully received. Contributors will of course get the benefits of anything I turn out

plater
1st February 2005, 14:12
Mat, I had a similar idea years ago when I used to burn the midnight oil with Qbasic :yikes: , now with spreadsheets and databases easier to use I cannot see why you could not create something that would be useful.

In a lot of papers and online you can find ratings for all the races, these I suspect are generated by a program of one type or another, but when I was playing around I found some papers throw more weight at speed figures etc, where others tended to have a more form reading type view, going / course etc, wish I could remember who done what but it was years ago. :D, could do no worse than anaylise those already generated by these sources.

As to what you req is another matter, whatever you do, you will never see the horses in the paddock, how they ran last time (race watchers get it wrong as well), how they are at home, plus many many more of the things that come into why they run well or badly.

Football, now there's another thing, anyone ever looked at the weather aspect in football matches where it's raining / windy etc, to see if there is a trend or two to consider High / Low scores ?? :wiggle:

Workshy
1st February 2005, 14:16
I really dont know how much you can get a spreadsheet (etc) to do but other angles are stats. If it can extract data regards trainer/jockey SR/LSP in race types/courses/distances/month etc.

mathare
1st February 2005, 14:22
Cheers guys.

The more I think about this the more I am looking towards a database solution. It's all very well being able to get qualifiers from today's racecards but we need historical data too. I need a database containing dates, each data has a number of meetings, each meeting has a number of races, each races has a number of horses, each horse has all it's attributes and a position and SP for that race. Using RP's data archive I should be able to build something up. It will take time and a lot of playing with VBA or another suitable language to get it right.

I know I can only put so much into this and it will never see how the horse looks on the day etc but it should help us find the sorts of trends Workshy is on about.

This is a long-term project though guys - don't expect anything other than hundreds of questions over the next few months

MarcusMel
1st February 2005, 14:57
I was thinking that we might try and collaberate Mather.

mathare
1st February 2005, 15:07
I was thinking that we might try and collaberate Mather.I was sort of hoping you might suggest that MarkisMol :D

If you wish to discuss vie e-mail because it is easier or whatever you have access to all my secrets as you are a mod

I guess a good place to start would be compare what you already have in place with what we want the system to be able to do. We need to look at what functionality we ultimately want then see what data that requires and check the design handles it. Then start adding new functions in bit by bit

MarcusMel
1st February 2005, 18:27
email sent - use the address for msn chat

Onlyforfun
1st February 2005, 18:40
I personally would look for something that RSB / raceExpert can't do.

What about create a list of key horses and search for good runs against these, supported by other decent form. Say any that has finished withing 10L of Marcel running in grade C or below for example.

MarcusMel
1st February 2005, 18:45
Key horses might be ones run in Large Prize money races.

mathare
1st February 2005, 18:46
Where the criteria for selection of a key horse and a good run are fixed in the program or entered by the user? Is "within 10l of Marcel in Grade C or below" a fixed query or do we have within xL of Horse A in Grade ? or above/below where the distance, horse and grade are all variables the user can set

It's a good idea and depending on what MM has in place already may be possible. I need to look at where we are starting from and chat to MM about it a lot.

mathare
1st February 2005, 18:47
I personally would look for something that RSB / raceExpert can't do.To me this is a key point, and we're dependent on users of RSB/RX to tell us what they want from their software that it can't currently do

MarcusMel
1st February 2005, 18:52
What currently interests me is creating a number for each horse that represents its relationship to the other horses in the race and assigning it(the number) a probablity of win. Not sure how to make that number yet but start simple and build in functionality as I go along.

MarcusMel
1st February 2005, 20:22
While the programs generated by RSB(have demo) and Racexpert(not used by me-yet!) are good there is nothing like getting your own idea and seeing where it goes. Yet frustratingly the software available does not always do what you expect. That is where you then save all the results of your system and start analysis outside of the program. For this reason and to take the tedius work summing up numbers, fiddling about with weight numbers etc I set about writting my own software. It still needs a lot of tweeks and as some point a complete rewrite. I am at the stage where I can just about start to do things with it.

Onlyforfun
4th February 2005, 11:53
What about...


...rating each race in terms of its quality, then selecting the runner that meet that level or better.

For example, take the total number of wins/places at grade F, E, D, C separately to decide the class level required to win the race. Then do the same for each runner and compare.

So if the "quality required" comes out at C and runner 1 is max quality D and runner 2 is max quality C, then runner 2 is the selection....

...or something. It's VDW, with a twist!

MarcusMel
4th February 2005, 12:40
Having been done once by missreading the class of horse because I did not take note of the prizemoney offered in past races I think class and prize money are intermingled.

Onlyforfun
4th February 2005, 12:58
Could rate the quality of the race based on criteria of prize money, track etc. and the future success of runners in it.

MarcusMel
4th February 2005, 13:03
Actually classing the effect of courses would be interesting. I hear the comments from ex-trainer Mick saying how tough and easy different courses are so it would be interesting to see if one could find a significant difference between them.