(continued from part 1) Other uses of systems include the more widely accepted and contested idea that a system can just pinpoint horses without form study.
This is true, however the form study is still involved; in this case it is retrospective. Using a system to isolate selections is step one, the next step is to find out how this performed in the past and eliminate further loosing contenders based on other filters. This retrospective filtering is the form study.
Quote:
|
Originally Posted by Brett V.02
I have tried everything in the last month to continue my system's winning start. But ... (everything) has failed to resurrect the once heady heights of the systems beginning.
So it has made me think (as has Darky's continued challenging of all I have ever believed in when it comes to racing - past results are the promised land).
|
Do not abandon your belief in past performance, I think you will find that Darky is just winding people up. There is absolutely no possible chance of isolating winning contenders in races without reference to past; be that Form, Ratings or just the look of horses or any other way of determining what has a chance. Everything is based on previous experience.
The main point of past experience is; HOW MUCH PAST EXPERIENCE have you experienced?
This question ties in together with the retrospective form study needed for systems. Horse races are not a random occurrence, their outcomes can be predicted, the issue is that there are so many factors that it makes it impossible to account for them all. You can still however create methods that cut through the many factors, isolate a few and more importantly isolate a pattern that can repeat in these elements.
In order to properly isolate a pattern you need YEARS of data, not weeks, or months. Also in order for this to work you need a large sample size over this period. You also need a program, the workload involved in isolating a profitable edge using a newspaper and a pen and paper is enormous and takes hundreds of wasted hours. You are against others that have the tools to cut through the workload in minutes. Programs do however have their drawbacks; this comes in the form of back-fitting. This can get out of hand and result in a system with a 120 selections over 3 years, that's back fitting, is just as useless.
Mathematical variance suggests that unless you have thousands of selections in order to work out the mean/average performance then you cannot truly determine the performance of your method. Only when you have a sample size large enough can you have the confidence to use the method. In the case of a few selections per year it could take centuries to gather the evidence. Mathematical laws cannot be beaten and there is no dispensation for anyone, only way to prosper is to recognise them and work with them. The caveat is that combining 100 systems with 120 selections each does not work either.
Brett you need to answer these questions (for yourself). How much data did I analyse? How many selections did my sample show.
Look at our Free Favourite system method thread, so far we have tested 6 months live; with bets every day, at this stage there is no form curse over the last 7 weeks that we have placed the bets here live. It's not like we didn't have a bad run or 2 but in the end we know what the performance is and in the end the method returns to that performance. This is from a system that was devised in an hour or less whose selections are generated automatically from Axis to show what is possible. The tools, information you use and their application are the difference between success and failure.
Please read the other threads and posts on our Ratings2Win forum here, it is filled with information that will shed light on what you experienced and will give you pointers to improve your punting. We have covered mathematical variance and system development many times.