Filters – A Path to ‘The Edge’
 

Hi All,

I have been thinking of filters and when The Schmile raised the issue, I thought it’s a perfect time to have a look. But it is a question I have been wrestling with for years now. Are filters a good or bad thing? Are these little tweaks a way to sharpen our selection strategy, driving us towards that punting “Edge” we all so desperately search for, or are they in fact a way that punters can cushion the psychological blow that we are destined to suffer when the hours and hours of work we put in to system construction, application and refinement still, eventually, push us over the “Edge” sending another system in the abyss. I have read on many different threads here and the debate rages on and probably will for eternity.

To put this into perspective, when I started looking at systems I ended up purchasing a commercial tool to help test systems and find future selections. I was like a kid at Christmas, as I spent hour after hour, day after day, pumping new filters in. And I still can see the smile on my face after the system test was run and the magic profit was achieved. I mean, a system test of 2 years which showed a profit………. It will surely continue. I can also remember the shock on my face when this profitable system was tossed in the bin 6 months later when it had shown results nothing like those of the test.

Back-fitting - I was under the impression that was part of a bra fitting that the missus has (and uses to justify spending 100’s of dollars on the stuff – give me my $2 Target specials)

Chi squared – I thought that was the 2 4 Tuesday special down at the local Thai restaurant

When I look back at some of these system filters, I shudder now. Things like:-

Race Distance – 1249 – 1651
Win % - 23%-46%
Place % - 51%-99%
Victorian Provincial and Country meetings on Tuesday, Thursday, Sunday

And worst of all, I was guilty of purchasing some of these systems, thinking that these people smarter than me, must be on the right track. I wish I could sit this bright eyed punting newbie down, put my arm around his shoulder and then slap him as hard as I can in the back of the head while screaming at him, if the system was making money, why would the bloke be selling it and not living on an Island somewhere.

I am not anti-filter. Far from it. But I am very interested to see what others think.

I will look at filters for this system, justify why I think we should use it or leave it and the stand back as the idea is attacked and defended by all in sundry. Friendships will be tested, stretched, destroy and rekindled. And when the dust settles, hopefully we have a nicely refined system that keeps ticking along.

Cheers

Brett

Conundrum:

A logical assumption that defies resolution, a difficult and intricate problem!

RP

Backfitting a system with ANY database is a total waste of time as the conditions being tested can Never replicate the millions of little things that occurred when the past race was run.
The vendors of these products will defend their businesses to the hilt I expect but LOGICALLY examine what I have just written.

My advice.Dont waste your money.

As you,ve found past results never live up to future predictions.

Cheers
darky

Backfitting i will say properly wont work at all , but sorting things like >10% win and 30% place and 2 wins or more, to try and narrow done the horses and the fields to find the better horses and races may be of help the bookies are so good at there job there cant be that many areas of profit left. It may be best used to find the areas that are really bad % loss wise over the long term and dont go there to start with and even that would be hard just ask the layers. It needs to have reason for me so i look for the better horses in the better races metro and try to find the horse or horses that are in or seem to be in with a real chance - first 4 in my ratings while i may win outside there it doesnt make that much differnce to my betting as it doesnt rely on the longshot.

For The Testing Process
 

I will include the selections I had before I went ‘live’. I am aware that there are only 85 selections and as such, the sample size is very (too) small.

Race Filters 1 – Exclude Maidens
 

Thinking –

As part of this system is based on ratings, and the thinking behind ratings is that the more data you can gather, the more accurate the ratings will be, then maiden ratings will be less accurate and produce results that are poorer

Results –

Non-Maiden
Selections – 62
Wins – 23
Win % - 37.1%
Outlay – $62.00
Return – $75.80
POT - 22.3%

Maiden
Selections – 23
Wins – 9
Win % - 39.1%
Outlay – $23.00
Return – $31.00
POT - 34.8%

Justification –
As the ratings are free from the internet, I am unaware what data is used to come to the ratings. This may also explain why some of the other ratings performed so poorly. As the results show, there is very little difference between the 2 sets of data and as such, I would not use this filter in this particular case.

On the subject of ratings i.e.Price

Where do the on course bookmakers get their ratings from? Is it a closed door service? They frame their market based on information supplied, from where?

RP

Horse Filter 1 – Starting Price
 

Thinking –

The top 4 in the market historically produce a lot of winners.

Results –

SP Under $3
Selections – 36
Wins – 20
Win % - 55.6%
Outlay – $36.00
Return – $46.20
POT - 28.3%

SP $3-$4.9
Selections – 33
Wins – 10
Win % - 30.3%
Outlay – $33.00
Return – $48.30
POT - 46.4%

SP $5 and Over
Selections – 16
Wins – 2
Win % - 12.5%
Outlay – $16.00
Return – $12.30
POT - -23.1%

Justification –

I will admit this is one filter I do already have an opinion on and use. I tend to operate using either first or 2nd favourites. I very rarely bet on a horse with and SP over $5. This is for 2 reasons. First of all, my personality and punting mind set struggles with long losing streaks. My nerves and patience struggle to deal with it. I have always taken the view that the horses I back under $3 are for my psyche, and the horses over $3 are for my bank. In saying that, I know a lot of people who regularly back winners at any old odds and have been doing it for a long while. I'm also aware of the fact that 2 or 3 winners at $5+ would realign their overall results and 16 selections are not very many. When using this system, I do use the exclude SP $5 filter.

The two most common would be above tab no 7 or tab no 5 and above and placed last start also broken into 1,2 or 1,2 3 combos or 1,2,3 ,4 last start

or beaten lengths last start 2 l under 1600 and 3l a 1600 and above ???

To let the tab no, sort class, bit rough but seems to work and try to see if the horse is in form rough again but not with out merit.

A nother couple are just the place strike rate if you are in the shorter end try 45% or greater and number of starts for the horse or all horses in the race as in must all horses must have had 3 starts min so less chance of a suprises.

Another lot is to break them up into ages 2 , 3 ,and 4,5,6 and maybe 7 year olds, and older into another lot cut of should be over 10 years ???