Your almost their barny the same problem i had years ago your having now.
Trying to get a set of mechanical rules to make your selections is almost impossible based on the ever changing things that happen in racing.
I collected 10-15 years of sportmans truth formguides herald sun liftouts etc.
Then attempted to induce a selection criteria based on any number of variables days form patterns barriers p/money p/post order etc etc.
These would work for a while then eventually lose so i would go back in time apply new filters to weed out the losers thus creating a new system.
Would then start again with obviously a new system same result lose again
After several years i gave up and joined several rating services while i won for a while eventually the one's i were in lost so back to keeping the day job.
i then joined punters choice who analysed most rating and mechanical data
i joined the most profitable ones unfortunately most only bet saturday's also purchased every profitable mechanical system.
These would grind out a few % profit then eventually die off years later.
Attending the races here in melbourne religously every saturday for years i met a couple of professional punters one particular saturday i summoned up the courage to ask how they made a living out of the punt.
Both said they purchase the ratings from successability or superform eagleform think they published a guide back then called the wizard
then made there own adjustments wait for it based on mechanical rules.
The trick then is to get a set of ratings (1st) zipform r+s( don scott) or the (neurals )these will way up each horses chances against each other using whatever criteria they use you can then apply the statistical side of it to improve the s/rate.
AT least this way you will have the selection criteria being assessed exactly the same way for each race.
Failing that the only other way is to purchase a data base for several $1.000 dollars and go back in time using ratings and mechanical rules.
Unfortunately this is impossible using r+s as their system only keeps the data for a couple of days and is then deleted due to the vast amount of information that is presented on the site.
cheers
garyf.

A qoute from the late don scott.
Show me the person who can find the class horse in the race.
That person will win as much money from racing as they want.
garyf

hi barny
Once you have your set of rules in place you will then have to sit tight, recording all the bets using the exact same rules.
After a several hundred or 1.000 bets if possible you can then go back in time to make the necessary improvements.
Any alterations will then become a new system thus the testing and recording starts all over again.
The things we punters do to find the elusive pot of gold so to speak.
cheers
garyf

Interesting comments guys.

I also believe the key to succes is to find a set of ratings and apply mechanical filters or rules based on a studied knowledge of the ratings and where their edge is.

Get to know your ratings,thats your best investment.

Dont ratings lead to the top four ? ..... then your doing what everyone else is doing. And everyone else (98%) lose ?

What's a poor old bloke missing here please ?

Not always Barny but the correct way of using ratings is the asessed prices and only betting those that are better than the asessed price BUT putting the money on at the asessed price.
i.e asessed $4 (3/1) price available $6 (5/1).
In this scenario you would bet $33 (the asessed price)
Many proffessionals will only bet if they can get double the asessed price.

Cheers
darky

hi barny.
far from it all ratings no matter where they are derived from regularly deliver prices all across the board i actually use the gtx ratings from t.r.b.
This is a pay for based neurals rating for a free version go to r+s website.
the only set of ratings i use on the free version is this.
(CP=5)(CF=5)(T/A=5)
I use this set on sydney country races plus canberra(ACT)
I use the top=3 only.
I bet on non-age related races eg no mares f+m 2yo 3yo cge races etc.
Of those races left i eliminate races where all horses haven't had 2 starts>
So today at p/macquarie the bettable races were races(1-4-5-6-8)
Race 1=1-7-4.
Race 4=6-5-2
Race 5=9-5-3
Race 6=14-6-1.
Race 8=8-5-1
If you check the prices of those selections you will see some short some are long etc obviously the larger bets go on the t.r.b ratings but i still invest on the free neurals as the selections there are profitable and different.
I usually don't post my selection criteria on the post's occasionally i will mainly i like posting filters to improve them more my go.
this is just an example of the broad range of prices for this set.
Tomorrow will post a filter i haven't seen yet posted here but have only been aboard a short while that will seem ridiculous in fact i bet a mate of mine $100.00 it wouldn't improve my best ratings system and lost badly.
cheers
garyf.

gary,

I think many professional punters go through the process you did but they became professional punters because of the journey. In the years it took you to move from mechanical patterns, to ratings, to your final selection method you would also have increased your skills in knowing what to look at and what is rubbish.

Experience is the key and dedication.

Correct on everything wes including losing years in that learning process. Which was the most horrible experience i ever encountered including then going back to work again to build up a bank after a losing year on the punt
and rebuilding a selection process i thought previously was a winning one.
That's why i can relate to barny in his attempts and frustrations as with many others to build a successful set of selections.
Always remember money lost nothings lost.
Confidence lost everythings lost.
cheers
garyf.

Can I just add my two bob's worth to the backfitting debate?
I came up with a longshot system that actually works (so far anyway) and is in profit to the tune of $199.1 for the calendar year after 1135 x $1 bets. i.e. POT approx 17.5%. I do have further data going back a few weeks of last year; not to hand but very similar S/R & POT.
On examination of my records today I noticed that 89 bets on system selections having their first start where they were the only first starter in the race produced a return of $120, whereas 207 bets on horses having their first start where there were more that one such horse in a race produced just a single win ($27.7).
I will not be betting betting these horses in future. Will that be backfitting or is it just sensible tweaking in the light of empirical evidence?
Regards
Ron