Oh no, I broke Combined Statistics. Damn!

[pyst7lot]

How is it possible that I back test (on last draw) a few complex filter groups containing some combined stats and pass and yet when I run the complex filter they fail to keep the winning line? One of the filter groups passed the back test and was successful when I ran it. The other two only passed the back test and failed when I ran them,

[pyst7lot]

Maybe it just doesn't work with unsorted draw numbers.

[stan]

You're probably doing something wrong...

[pyst7lot]

You're probably doing something wrong...

I wish that was true because I really want it to work with unsorted draw numbers.

[pyst7lot]

EGG ON MY FACE ...EGG ON MY FACE

I redid my complex filter and it all worked.
Not sure what I did wrong. Only thing I can think of is that I have a problem with copy/paste from the analyser into the complex filter.
I often 3 times out of 4 have to do a double paste because I re-paste the old data on the first paste. I then have to check that I have four valid pastes and I don't double paste one entry and miss out on one of them.
However, I thought I had allowed for that and had checked each paste to make sure it was new stuff I was pasting.

I hope I have it sorted because this worked so well for my other game which I used 30 sets of 4 combined stats and had no errors.

[tdnl46w]

How is it possible that I back test (on last draw) a few complex filter groups containing some combined stats and pass and yet when I run the complex filter they fail to keep the winning line? One of the filter groups passed the back test and was successful when I ran it. The other two only passed the back test and failed when I ran them,

The “Back-test” compares the filter(s) to the winning numbers tickets, and “Run” compares the filter(s) to the package tickets. So if you have filters that use selected latest draws in them, the Back-test and Run might have different ticket results. I’ve been using this program for years and I still sometimes completely miss read results because of this.

[pyst7lot]

Yes, back testing can get a bit tricky depending on what filters you use or how you use them. I mostly just back test the last draw to see which of my filters failed for that specific draw. I'm forever hiding and unhiding the last draw when I'm trying (testing) new things - maybe I forgot to hide or unhide a couple of times. I thought I was being very vigilant about that though.
It was just so odd that back testing showed success but running didn't.

[pyst7lot]

Still problems. Hope the screenshot helps make my point.
The 3rd group is the problem. Passes the back test but fails when I run the Complex filter which includes the Match Number filter with the same numbers which passed in the back test.
If I change the last groups to Min Pass 3 and Max Pass 3 it fails the back test and passes when I run the Com Filter.

The other two groups are not a problem. They pass both the back test and when I run the Com Filter.

I should have made a 2nd screenshot with Min and Max both at 3. It shows Min 3 Max 3 Back test passed 2 of 6.

2025-06-10_215530.png

Damn!
I changed the 1st group to Min Max both 5 and the 2nd group to Min Max both 4. and they both fail when I run the Com Filter.

Am I getting it all wrong!

[tdnl46w]

Still problems. Hope the screenshot helps make my point.
The 3rd group is the problem. Passes the back test but fails when I run the Complex filter which includes the Match Number filter with the same numbers which passed in the back test.
If I change the last groups to Min Pass 3 and Max Pass 3 it fails the back test and passes when I run the Com Filter.

The other two groups are not a problem. They pass both the back test and when I run the Com Filter. ...

I would need to see the entire complex filter and then test it in your lotto but,...a better/faster way to see which filter(s) are causing the problem, is to disable all filters, and then enable them one by one and back-test/run them to see where the issue happens.

[pyst7lot]

I think my using unsorted draw numbers in my winning numbers table is what is causing problems - it's just not meant to be possible.

What I have done is this.
For testing the last draw I hide it

I make the follow Combined Filters.
Eight filters using Postions 1,2 and 3 and Odd/Even (OE) with Low/High (LH)
OE OE OE
OE OE LH
OE LH OE
LH OE OE
LH LH LH
LH LH OE
LH OE LH
OE LH LH

Load the Analyser with Combined Stats
Check each of the above Combined Filters
Copy the entries which have taken longer than 8 draws and paste them into a Complex Group.

Usually at least half will pass in the next draw and pretty rare for all to pass.

If I could make a package with unsorted lines I don't think I would have the same problems - but unsorted lines means an enormous package with a 6/40 game.
I use it with my 4/40 balls in draw order game which I made an unsorted package for and I don't have the same problems with it.