Remove Doubles and triples for Pick 3

You ask, we answer...if we know the answer!
Post Reply
draughtsman
Site Admin
Posts: 112
Joined: Thu Jan 17, 2008 4:22 am
Location: Melbourne, Australia

Remove Doubles and triples for Pick 3

Post by draughtsman » Sun Jan 27, 2008 6:24 am

TRANSFERED FROM OLD FORUM

This following describes how to remove doubles and triples tickets from a Pick 3 ticket listing.

We'll use the rejection filter "Mass ->Dif. Digits" for this.

Remove triples
A triple ticket is the one which has all numbers the same e.g. 333, 777 are triples.
The above rejection filter counts how many different digits occur in a ticket. Obviously the above triples have only 1 digit (3 and 7 respectively). So, to remove all triples, we have to create the above filter and set Value=1 to "No" (the other values should be set to "Yes"). Create this filter in a session and apply Stage 3 calculations using this session. Your triples have been removed!

Remove doubles
A double ticket is the one which has 2 numbers the same e.g. 353, 448, 799 are doubles. We'll do the same as triples; the only difference is we'll set Value=2 to "No" and all other values should be set to "Yes".

Remove doubles & triples
Same as above; Set Value=1 & Value=2 to "No" and the other values to "Yes".

Hi there at Lotto Architect,

I am exploring rejection filters and the report process where you can see the hit rates etc plus the success rate. I have opened up my reports to show 100 predictions and I am getting numbers like 100% 99% 98% 97% etc and this is where the prediction results appear to mostly fall, but occasionally I see 95%. this of course must mean 5% inaccurate. Can you advise what sort of tolerance one might accept here. Is for example 99% OK and if that is the case then where should one not accept the performance of a filter ie 94%?



many thanks

relowe

Filtering with rejection filters seems to be the most inaccurate procedure of all rejection systems included. The addition of reports aims to get a clear view on this matter. The most draws tested against an algorithm, the better view you have about the accuracy output of the algorithm. Rejection rates of 95%-100% are quite acceptable. But you may get 100% if you test only 5 last draws and you may get 100% over the last 100 tested draws. Intuition says that the 100 tested draws provide a more steady accuracy.
What you should look for is high accuracy with high rejection rates. This will allow better filtering with fewer filters and it is a very important factor for correct rejection.

As a quick start guide, you can generate all possible filters in all categories with the "out of bounds" algorithm. This will ensure rejection of tickets that have never experienced in the whole history draws (use by 1 draw filter for this) and therefore unlikely to come up at the next draw. You may wish to use this "out of bounds" approach using other draw filters: e.g. "last 500 draws" so you still maintain a large statistical data and result in greater rejection rates.
Your next step is to determine filters that have a high accuracy and high rejection rates at the same time. I cannot offer advice of what filters these might be but you can use the report to determine that. Also note that you can try several draw filters as I have experienced good results with only "last 50 draws" or "last 100 draws" draw filter. This offers the desired properties we want most of the time.

Hope this make rejection filters usage more clear.

Post Reply

Who is online

Users browsing this forum: No registered users and 11 guests