Odds Wizard vs horse racing

[Alex]

Many UK users ask horse racing to be included in OW. Here is the common answer that become available after certain efforts in math analysis.

Horse racing is a kind of competition that appears to be extremely hard to predict because of the following reasons.

Let's say that you have measured individual performance of the participated horses in terms of 1) math expectation of the running time (mu), and 2) its math variance (sigma^2). That data is a mandatory statistical basis of all further computations.

Let's also assume that running times obey Gauss distribution.

Now the formal solution can be expressed in terms of N-fold semi-infinite integrals of N-dimensional Gauss distribution, where N is the number of horses.

I tried to compute those integrals using the program Mathematica (version 3, 5, etc. doesn't matter). 3-fold integrals took 3 minutes to be computed in 2 GHz system. 4-fold - over 30 minutes with the minimum accuracy. Yet the number of integrals to be computed is equal to N! (factorial). 9! = 362880 9-fold integrals for 9 horses will be computed all the time of the Universe being.

My conclusion is that existing math formalism does not allow to compute desired probabilities/odds in a reasonable time in a horse racing.

To substantiate the answer, I attach the formulae for just 3 horses, where p(ABC) means that A was 1st, B was 2nd, and C was 3rd. Naturally there are 3! = 6 combinations and respective integrals:



Anyway, the work is still continued, and there is a hope to find out some workaround for the above integrals.

[danniego]

To minimize computation time to reach the solution(s), in due time:(?)
1: Cloud computing
2: Parallel environment
3: Using the GPU to offload
4: Group of users who run an client-host routine to distribute the computation amongst the group of members and send data back to main.
5. Asking NASA to share cpu time on their main frames

[Alex]

Wow, so many forgotten instruments
Just turn on intuition, and there is nothing necessary at all!

[push.button.bang]

hi alex

any developments on the horse racing?

as a keen horse race gambler i know this tool would be a great asset to us all.

cheers
PBB

[Alex]

Yes, we do have certain advances. But we'll still need the same input: statistics of runs for each horse.

[TheBluesBrother]

I cannot see anybody coming up with a method to price up horse racing because of the variables involved.
I have been trying to come up with a method for the past 30Years+.
At this moment in time I have been scraping the sportinglife.com website to remove 16 years of data i.e. winning times etc to produce accurate standard times for each distance and each class.
I hate copying and pasting so I have a webscrape program I have setup downloading the information into a SQL database, which takes 30 minutes per 1000 pages scraped.
I am not running the program too fast as I do not want to mess up their bandwidth.
In a matter of weeks I should have a complete database of results covering some 150,000+ pages and 16 years of results (1994-2010).
As far as I am aware there is nothing like this available on the net.

[Alex]

Thanks. Just browsed sportinglife, and found the following horse details:

A lot of info there but still missing running times which are essential for evaluating the odds.
Is there a place on that website where running times can be found?

[TheBluesBrother]

Hi Alex, if you click on the date it will give you the full results list for that race, which includes the winning time, but this is only for the winner.

These are the pages that I am web scraping, removing 9 pieces of information, like course, distance, age, race title, going etc.

I am working from an excel database of URL's to scrape due to the makeup of the end of the URL i.e. ,00.html.
If the URL ended in 'http://www.mysite.com?page=' I could have used a SQL statement like

Code: Select all

SELECT 'http://www.mysite.com?page=' + _COUNT from _COUNTER, _MYTABLE WHERE _COUNT <= _TOTAL_PAGES

to count the pages.

example:

Code: Select all

http://horses.sportinglife.com/Arch_Race/0,12658,129514,00.html
http://horses.sportinglife.com/Arch_Race/0,12658,129515,00.html

it is just as easy to make up multiple URL's in Excel.