4.1 BETA (since build 432)

[stan]

i uploaded beta version 4.1 build 432 with the following updates:

- bug fixes:
* previous builds didn't handle zeros in bonus numbers while calculating winning numbers statistics
* unsuable compound filters
* sorting of odd/even and high/low columns in the winning numbers table stats

- new features
* added 'match numbers' filter as requested

(started a new thread as the old one was getting a bit long... )

[Maryland]

Stan, In statistics, winning numbers table, I click on "In last 30" menu bar thinking I would get a ascend or descend result, instead my computer locked solid. Even task manager couldn't shut down the program freeze.

[Maryland]

Forgot there is also no browse in import wizard.

[stan]

i don't think this is a bug.
if you have many tickets selected in the wn selection panel (thousands and more), then it may take a few seconds to sort the table.
try 'latest 100' tickets in the wn selection panel.

i'll probably have to restrict the number of wn tickets in this table to the latest 200-500 or so anyway.

[stan]

[quote=Maryland]
Forgot there is also no browse in import wizard.
[/quote]

hmm, it works fine for me. can you email me the error.log file pls?

there's a workaround for this anyway - open the file you wish to import in e.g. notepad and copy&paste its contents to the wizard's window.

btw, sometimes it may take a few seconds for the browse file window to open and sometimes (very rarely though) even to fail to open. but the second attempt to browse for a file should be ok. this is a java problem though, there's nothing i can do about it.

[Maryland]

Your right about to many winning numbers in statistics page. The browse still does not open but I can put it in the default directory that comes up instead of my desktop. I hope I'm saying it right. You click import file then another window opens where you can browse for file but there is no browse option just the default path from Lotto Expert.

[stan]

[quote=Maryland]
Your right about to many winning numbers in statistics page. The browse still does not open but I can put it in the default directory that comes up instead of my desktop. I hope I'm saying it right. You click import file then another window opens where you can browse for file but there is no browse option just the default path from Lotto Expert.
[/quote]

can you pls post/email some screen shots? it's hard to tell what's going on.

[zz]

[quote=stan]

- new features
* added 'match numbers' filter as requested

(started a new thread as the old one was getting a bit long... )
[/quote]

That was really fast ...doe's it mean there are some chances of getting the generator - generate all cosidering an overlap before the final release?

[stan]

[quote=zz]
[quote=stan]

- new features
* added 'match numbers' filter as requested

(started a new thread as the old one was getting a bit long... )
[/quote]

That was really fast ...doe's it mean there are some chances of getting the generator - generate all cosidering an overlap before the final release?
[/quote]

i'll try because i'm curious how this is going to come out

[zz]

deleted - found what was wrong in the setting

[zz]

[quote=stan:1109328862]
[quote=zz]
[quote=stan]

- new features
* added 'match numbers' filter as requested

(started a new thread as the old one was getting a bit long... )
[/quote]

That was really fast ...doe's it mean there are some chances of getting the generator - generate all cosidering an overlap before the final release?
[/quote]

i'll try because i'm curious how this is going to come out
[/quote]

Hi Stan,

Any significant progress ?

[stan]

yes, it's ready.
i didn't upload a new build yet because i'm working on new 'filter statistics' at the moment - it'll be possible to find out how any (compound) filter performed in the past:

1. generate a full wheel
2. pass all generated tickets through a selected filter
3. calculate number statistics (which numbers are most frequent after filtering)
4. take first 6/10/15/20 most frequent numbers and see how many of these numbers were actually drawn in e.g. 20 latest draws.

all of the above will be done in one step only. so basically nothing you couldn't do with the current feature but more convenient.

also it'll serve as a starting point for the predictor - except that we'll try to plug in something more 'intelligent' than a simple yes/no filter

[zz]

[quote=stan]
yes, it's ready.
i didn't upload a new build yet [/quote]


Well, the most important is it is ready....
Thanx Stan


[quote=stan]
1. generate a full wheel
2. pass all generated tickets through a selected filter
3. calculate number statistics (which numbers are most frequent after filtering)
4. take first 6/10/15/20 most frequent numbers and see how many of these numbers were actually drawn in e.g. 20 latest draws.

all of the above will be done in one step only. so basically nothing you couldn't do with the current feature but more convenient.

also it'll serve as a starting point for the predictor - except that we'll try to plug in something more 'intelligent' than a simple yes/no filter
[/quote]


It seems 'filter statistics' will be some really powerful backtesting thingy ... But hope no user may expect you might make an extremely acurate predictor that will work in absolutely all draws. That is not possible nor it will be now or in future...

[stan]

hi everybody,

BETA build 439 is avilable for download. the new build contains the following changes:

bug fixes:
* it's possible to use compound filters in group filters and vice verse (i hope i didn't break something else while fixing as i'm in a hurry a bit - see below)
* it's possible to set '0 to X' filters must pass in filter groups

new features:
* the ticket generator allows to set 'maximum overlap' even when generating non-random (i.e. successive) tickets
* added a new page for filter statistics (back testing). here's how it works:
1. select filter input first. it means setup ticket generator that will provide tickets for the filter.
2. select filter you wish to test. compound and group filters are also available.
3. select winning numbers cut-off date - the winning numbers database will 'pretend' that it contains only previous winning numbers up to and including to the selected draw date when providing data for winning numbers - based filters. more on that below.
4. select for how many latest draws (since the selected cut-off date) you want the stats to be calculated
5. hit the 'calculate' button and then you can examine the table on the right.

the tickets from selected generator are put through the selected filter and statistics are calculated for tickets that passed the filter. the numbers are then sorted from the most frequent ones to the least freguent ones and the table shows how many of the most frequent numbers were actually drawn for a given draw date. e.g. from 10 most frequent numbers only 3 were actually drawn for draw date XYZ. the tooltip for each table cell shows the actual matching numbers.

the 'hits chart' page shows the same data in a visual form. use the 'options' button to see other charts.

you can select a single row in the 'hits table' (i.e. select a particular draw date) and see number statistics for that date in 'numbers table' and 'numbers chart'.

what's missing:
* print/export buttons for each table/chart
* the winning numbers database does not take into account the selected cut off date yet. this date is important for winning numbers related filters only (match wn and wn history filter), using a simple odd/even filter will provide the same statistics for all draws.
the drawback will be that to calculate hits for 'match wn' filter the application will have to make "A * B" calculations where A is the number of latest draws selected and B is the number of tickets the selected generator will produce. (for non-winning numbers-related filters the A is always 1)

note: i won't be online from tonight (thursday) until sunday evening so pls do not expect my immediate response on your comments.

[Maryland]

Stan, could you check out the triplets filter on your sample 6/49. It doesn't filter anything. I tried it on my own 6/49 MD didn't work so I tried it on your sample data base and nada. Thanx