how accurate is the back-test from the complex filter?

[Wuuu4D]

Hi, something is horribly wrong with the backt-test from complex filter.
I have a lottery, 5 numbers out of a 49. I have attached a file containing all previous winning number.

I also have a complex filter (https://www.mediafire.com/?0ml5f1c1ben55nn), when I back-test this complex filter the number 16 23 25 26 42 passes the filter.

26.02.2015 16 23 25 26 42 <= this is the last winning number
so far, so good
I delete this number from the winning numbers, and save the winning numbers.

I insert the number 16 23 25 26 42 into a package, I run the complex filter on the package, and to my surprise the number that passed in the back-test didn't pass now.

The Complex Filter looks something like this
Complex Filter (Min to pass 1, Accept)
> Filter Group (Min to pass1, Reject)
>Filter Group (Min to pass2, Accept)

What's going on? I've been doing this to alot of complex filters(guiding myslef using back-test), does that mean that all my previous complex filters are inaccurate?
I'm a upset because I put alot of hours into building many complex filters.

https://www.mediafire.com/?0ml5f1c1ben55nn <== link to the complex filter

[andthen]

hi

there shouldn't be a mistake in the back test facility. Without knowing the lottery you are playing it is not possible to check your filter and work without setting up a special lottery to test it.

If I come across apparent discrepancies in back testing or complex filters an easy way to test the filter is to insert one ticket only...the winning one, ensuring that you have hidden the last draw and press preview to see which filter is rejecting your ticket.

Hope this helps, regards, AH

[Wuuu4D]

@andthen I tried that here is what it looks like


As you can see the ticket is rejected in 3nd group, the ticket is accepted in the 2nd group, shoudn't the 1st group accept the ticket, with the settings min 1 must pass? The ticket passed in the back-test when I back-tested 1st group

Regards, WU

[Jerzy]

Wuuu4D,

Your complex filter contains two Filter Groups and both of them fail. So the Filter Group, or the complex filter as the whole, fails as well.

There is no third filter group in this complex filter.

[Wuuu4D]

When I say the 1st group I'm reffering to the complex filter containint the two groups.
If they both fail why doesn't the ticket get rejected when I back-test the complex filter?
This is what it confuses me, it should also fail when runing back-test on the complex filter

[andthen]

So you need to drill down into the filter group and find which filter is rejecting your ticket..the image you have posted doesn't help.

[stan]

I went through the code and found the problem - the back test executor does not honor 'matching tickets accept/reject' applied to filter groups (individual filters evaluate fine). I'll upload a fix later on today.

[Wuuu4D]

Thank you stan, I have updated EL and now back-test seems to work fine

[Wuuu4D]

I have found another bug in the backtest
I have a single number 2 7 19 28 43 in package




As you can see the ticket is rejected when previewd, but the ticket passes when back-testing

[stan]

I have found another bug in the backtest
I have a single number 2 7 19 28 43 in package




As you can see the ticket is rejected when previewd, but the ticket passes when back-testing

You'd have to hide all draws starting from 20.11.1994 when running the preview to get the same result as in back-test.

[Wuuu4D]

@Stan you are right, I I apologize I tought the filter isn't influenced by past draws since no matter what past draws I selected the Accept/Reject count remained the same

Thank you for your support