CP we all know of PPM and a lot of the rubbish they produce at times but the released a set of Impact values about 20 years back, these were very successful for me back then as mentioned on another post i wrote my first programs in Qbasic based around these figures, it is possible i still have an excel sheet somewhere with the original figures, i will see what i can find unless someone has the article from back then.

Well put C/P.

I guess this is what people do not understand,
As to why we do not lay horses.

Simply put it makes more money & is,
Percentage wise better for me to back 2-3 horses against,
The Favourite when it is poor statistically than it is to Lay it.

However I never bet in races where a horse is odds-on.

Good point with the I/V'S particularly when you get a positive set,
Many times they goes nowhere a profit by themselves but find a group,
Of them together & you are on your way.

Just another piece in the jigsaw puzzle I guess.

You probably know of the author who lives way down south who has written many books on just this subject with graphs & explanations.

Think I have purchased all but one of them.

Hope some more people can contribute here on this thread,
I look forward to reading other peoples thoughts on how to,
Make something manual for EVAJ to assemble using filters.

Positive Impact Values are a great start "BUT" that involves,
Buying a book(s)" OR DATA BASES" which contains them, that leads to spending,
MONEY which the majority will not do.

Post = 11 below me may come to fruition.

Cheers.
Garyf.

One of our ratings in Axis is based on impact values, of about 6 factors if I remember correctly. They do well. As it was mentioned before, some investment is required to obtain these values in form of data acquisition. Whether you do this by yourself or use some tool to do it. The concept is a great idea. However at the end it is another statistical interpretation of data. Its up to the user to make it work.

I got interrupted, so here my post continues.

Most punters can calculate these values for anything rather simply. You need the data and a tool that will crunch that data.

Next you need to analyse a sample of data to get the over all figures of runners vs winners, or the general strike rate as that is what you are going to divide.
Once you have that then running a test over the same sample, using one form filter at a time will, via the results, give you the figures you need to calculate the impact value of that form/rating element.

Something to watch out for, is when running 2 filters at a time or more, 'to save time'. Because then you are going to get the impact value of that 'pair' of filters.

If your tool has analysis function, where you can breakdown the first test (the all figs) then you can simply obtain the strike rates from the analysis in one go rather having to run individual tests.

Another thing to be aware of, the initial ALL sample. In the sample of AWT in CP post, your ALL sample needs to be based on a filter that isolates only AWT tracks. So you need to be careful of what your starting point is and to make sure you are not using a wrong one.

Another good example would be to calculate the IV of jockeys, based on all jockeys performance % vs individual jockey performance %. Here you need to have that ALL sample and then using analysis of individual jockeys strike rate you can simply create these values in a spreadsheet if you can export.

No great mystery, and you don't even need a book! Where it does become interesting is when you start thinking about your IV. Running just blindly ALL as described before will produce IV that everyone else will produce. And that may be good enough. It is the ACADEMIC sample. You know math performed by people or programers that don't punt! However there are further ways that you can drill down to obtain more truer IV ratings, by removing the chaff or if you like horses that otherwise cloud the 'issue'.

A couple of gems of wisdom here from CP. And to add another, I think from Bill Benter (possibly the most successful gambler in history), "Find the horses that CAN'T win first!"....also known as Laying opportunities, for the backers in denial! So if you think in reverse, you might just do better at backing?

The second gem of CP's is about finding characteristics pertaining to a certain group, be they likely winners or losers. It's where I was coming from in an earlier post. Once you collect your data, you need to filter it to find those groups. Here's an example below where I query my database with 1,1,1,1 (as in ranks of a value/rating). When displayed on a scatter plot, the winners are shown in the left hand table and the losers with a price >=$5 Back in the right table. This group haven't won a race yet (duration of data collected). What does that tell you"
If a backer....don't waste your money on selections rated 1,1,1,1 >=$5
If a layer....look to lay the above.

None of it's rocket science and anyone with a bit of excel knowledge, heaps of patience and prepared to put in the time, can create opportunities for themselves.

One point to note is that the normal way to get impact values is to group different factors together as "all in". Take distance for instance, 1200M at one race track is vastly different to 1200M at another. The position of the first turn or the length of straight makes a different style of race.
Similarly a 7 horse race is vastly different to a 15 horse race.
So you must group like with like which may not give enough races to make a true sample.
If you put 15 pots of paint together, you will get absolute rubbish. However if you mix them by base colour in three groups, you may end up with three reasonable colours.

What Wally is saying is good advice, of course. However, I tend to place no limits on any race other than no hurdles, steeples or untested 2yos. That's it, come what may after that. You need to IMO, for AUS races through the week. There might only be 3 meetings on, by the time you eliminate soft tracks, synthetic tracks, maidens etc you might only end up with 3 or 4 qualifying races.

The data I grab, is one size fits all, pretty well, and the analysis is on that basis, so it should all average out in the end. The more filters you apply, the smaller the sample, the less reliable the analysis. Apart from the race types described earlier, I just use Price, Volume, and two free Ratings. Most, if not 90% of the variables, are taken care of in the consensus of those four factors.

So keep it simple to understand, work out and your punting life will be a lot easier. I'm sure the fastidious types will disagree, but each to their own.

Got my vote


What's I been toying around with is the old manual "horse racing programs", in DOS, yeah OK, some of us got a bit naive at the time, but hey, some of these programs have merit, it requires form none the less, and calculates a rating.
The recommended; "the top numbered rating is a Place bet, but in time you will identify the winners", they said so.
I combined all these program's so called "ratings", for the MC race when Saintly won, and got the Trifecta and some other good wins.
But it took time and quality family time. Automation just crept in back then.
My PC was a 486 Win 3.1, ( I thought I had "arrived" )
I have no idea about cracking into these programs to find the "magic key to Aladdin's cave formula" they used, but I tried. Not much help from Google either, or I don't understand what to look for.

Anyway,
It requires the Sportsman, a good set of reading glasses and good typing skills.
NOT that I am going to take it on as a new career path as in what is old is new again. What I can do is show here what these programs required for the input data fields, which is purely form, and why they used this info, which is much mentioned here.
A members only would suit this idea though.

Michal could you point me to some software that can do that kind of analysis?

Quote:
Originally Posted by Michal
Member Restricted forum section would be I think a good idea.

Got my vote as well.
Will try to get more involved in the forum discussions from here on in, as I have learnt a multitude of things from this forum and its many selfless contributors, so to give a little back isn't too much to ask I suppose. (Not that I am saying that I am some sort of genius by any stretch of the imagination).
Anyhow, keep up the good work fellas.
Cheers,
Jose.