wish list for Expert Lotto 5.0

Postby stan » Thu Oct 22, 2009 9:17 am

[quote=Gatsby]
is it possible to include a button on each filter menu to be used directly as help with the definition of what that filter does or works (as with preview or backtest)?
[/quote]

F1 key does that already, even in version 4.x
Expert Lotto Team
User avatar
stan
Site Admin
 
Posts: 6338
Joined: Thu Sep 23, 2004 1:01 pm

Postby Gatsby » Fri Oct 23, 2009 12:43 am

:red:
I´ve kind of away and trying to catch up....sorry.
Gatsby
 
Posts: 305
Joined: Tue May 01, 2007 3:24 pm

Postby Viking » Fri Oct 23, 2009 6:34 pm

Hi Stan,

Thanks for another new early preview version of EL 5.

I have some thoughts about the segment filter in EL 5 I must share with you. The segment filter is better in EL 5 than in EL 4 since you can adjust how many segments to use. But this filter should be much better than this because after my opinion this filter is on the top of the list for best filters in EL.

As I have told you before I hoped you could implement adjustable values for both min and max difference and not just max as it is now in both EL 4 and 5. I remember you have told me that overlapping segments can be difficult to implement, but forget that if it is difficult. What I want to see and hope you will implement because it is very useful and necessary to get this filter to work at it best is adjustable min difference value without the ability to overlap segments since this is a problem to get to work in EL.

If this min difference value is added to EL 5 there will be possible to filter with overlapping segments using complex filter. You can also use it to jump over some difference values in segments.

Regards,
Viking
User avatar
Viking
 
Posts: 181
Joined: Wed Mar 04, 2009 12:40 pm

Postby stan » Fri Oct 23, 2009 6:52 pm

[quote=Viking]
If this min difference value is added to EL 5 there will be possible to filter with overlapping segments using complex filter. You can also use it to jump over some difference values in segments.
[/quote]

not sure i'm following you. every segment has a min difference value which is the maximum difference of previous segment.
it is possible to create segmentation with adjustable min values but then the segments won't be continuous and there'll be problems when a number falls into a 'gap' between segments. so there might be segment combinations like '0-0-0' when all numbers are out of segments. it's a similar problem to overlapping segments.
i'd prefer adding a new 'history differences' filter instead of abusing segments like this
Expert Lotto Team
User avatar
stan
Site Admin
 
Posts: 6338
Joined: Thu Sep 23, 2004 1:01 pm

Postby Da80th » Fri Oct 23, 2009 8:11 pm

hi Stan,

Here's my wish list for version 5. I am thinking adding Odd/Even option to Segment filter.

For example, to keep thing simple I used only two segment.

Column 1

18,29,31,35,39
10,24,28,36,
11,12,32,38,
5,26,30,37,
13,15,17,,
9,19,21,,
----------------

8
3,20
34
16,33
1,2,14,25
22
23
4,7
27
6


The 1st segment contained 13 ODD + 10 EVEN, total 23 numbers.

And 2nd segment contained 7 ODD + 9 EVEN, total 16 numbers.

The previous drawn numbers (18,29,31,35,39) contained 4 ODD & 1 EVEN.

First - users must estimate how many numbers from previous draw will repeat for next draw.

Second - user must estimate how many numbers will be draw per segment and EL will calculate the probabilities of what are the % of ODD & the % of EVEN per segment.

ONLY when the % per segment are on your favor then you select to keep, otherwise user leave it at default.

IF you estimate this filter correctly then you should be able to eliminate 30% of tickets in package per segment per column.

So instead of using all column 0 to -5 correctly to win jackpot, with this ODD/EVEN Segment filter user only estimate column 0 to -3.

A BIG improvement!


Da80th
Da80th
 
Posts: 382
Joined: Sun Mar 19, 2006 9:11 pm

Postby stan » Fri Oct 23, 2009 10:00 pm

[quote=Da80th]
Here's my wish list for version 5.
[/quote]

please enter it as a 'feature' here: http://expertlotto.com/issues/bug_report_page.php
(you may need to create an account first)

did you back-test this strategy? how does it work when no number is repeated from previous draw?
also if my estimates for the count of repeated numbers is wrong then all my following estimates and results or totally off, aren't they?
Expert Lotto Team
User avatar
stan
Site Admin
 
Posts: 6338
Joined: Thu Sep 23, 2004 1:01 pm

Postby Da80th » Fri Oct 23, 2009 10:51 pm

[quote=stan]

did you back-test this strategy? [/quote]

I did backtested for lest than 10 draws and only for 1st segment. I did it manually so it very confusion.

how does it work when no number is repeated from previous draw?


If no number repeated AND user select "0" that mean EL eliminate ALL the previous drawn numbers in each column.

As above example, 1st segment contained 13 ODD + 10 EVEN, total 23 numbers then EL eliminate ALL previous drawn number (18,29,31,35,39), now the new counts will be 9 EVEN and 9 ODD. The % now become 50/50.

also if my estimates for the count of repeated numbers is wrong then all my following estimates and results or totally off, aren't they?


Yes. Unfortunenately nothing works 100%

Remember user only use this filter when EL give estimate over 75%, another words use this filter when the odds is in your favor.


Da80th
Da80th
 
Posts: 382
Joined: Sun Mar 19, 2006 9:11 pm

No Backtest in new V. 5 filters

Postby Jerzy » Sat Oct 24, 2009 3:48 am

Stan,

Thank you for a preview of EL v.5. It looks very impressive.

However, it worries me that the Backtest function, which plays important role in my strategy, is no longer present in new version filters

The Backtest in v.4 allows to see how the particular filter setting performed in the past. It is useful because what happened in the game in the past it will be repeated in the future. Certain settings are more likely to be repeated than others. And some settings are very unlikely, good for filtering tickets out.

Jerzy
Jerzy
 
Posts: 811
Joined: Mon Sep 18, 2006 5:26 am

Postby stan » Sat Oct 24, 2009 8:35 am

[quote=Da80th]
[quote=stan]

did you back-test this strategy? [/quote]

I did backtested for lest than 10 draws and only for 1st segment. I did it manually so it very confusion.

how does it work when no number is repeated from previous draw?


If no number repeated AND user select "0" that mean EL eliminate ALL the previous drawn numbers in each column.

As above example, 1st segment contained 13 ODD + 10 EVEN, total 23 numbers then EL eliminate ALL previous drawn number (18,29,31,35,39), now the new counts will be 9 EVEN and 9 ODD. The % now become 50/50.

also if my estimates for the count of repeated numbers is wrong then all my following estimates and results or totally off, aren't they?


Yes. Unfortunenately nothing works 100%

Remember user only use this filter when EL give estimate over 75%, another words use this filter when the odds is in your favor.
[/quote]

alright then, please enter it as feature request into mantis bug tracker
Expert Lotto Team
User avatar
stan
Site Admin
 
Posts: 6338
Joined: Thu Sep 23, 2004 1:01 pm

Postby stan » Sat Oct 24, 2009 8:41 am

[quote=Jerzy]
However, it worries me that the Backtest function, which plays important role in my strategy, is no longer present in new version filters
[/quote]

back-test is available in complex filter window. i expect that complex filters will be preferred in version 5:
- they are much easier to use than in version 4.x
- they don't block access to package tickets when the filter is being applied
- they're much more powerful than a single filter
- they offer performance evaluation as well
Expert Lotto Team
User avatar
stan
Site Admin
 
Posts: 6338
Joined: Thu Sep 23, 2004 1:01 pm

Postby Jerzy » Sat Oct 24, 2009 10:58 am

OK, Stan. Thank you for your explanation.
Jerzy
 
Posts: 811
Joined: Mon Sep 18, 2006 5:26 am

Postby Viking » Sat Oct 24, 2009 11:35 am

[quote=stan]
not sure i'm following you. every segment has a min difference value which is the maximum difference of previous segment.
it is possible to create segmentation with adjustable min values but then the segments won't be continuous and there'll be problems when a number falls into a 'gap' between segments. so there might be segment combinations like '0-0-0' when all numbers are out of segments. it's a similar problem to overlapping segments.
i'd prefer adding a new 'history differences' filter instead of abusing segments like this
[/quote]

Hi Stan,

I understand that you only want to have continuous segments in EL. I hope you will add a winning number history difference filter instead and also the winning number history difference statistics for just the winning numbers by draw date I described to you in an earlier post.

The reason because I want this filter so badly is because I have tested it for level 0 in EL 4 and it is a powerful filter.

In the WNHD filter you need just one adjustable min and max difference value for each level. When building a complex filter with more WNHD filters together it will be a very powerful filter.

Regards,
Viking
User avatar
Viking
 
Posts: 181
Joined: Wed Mar 04, 2009 12:40 pm

Postby stan » Sat Oct 24, 2009 12:54 pm

[quote=Viking]
I understand that you only want to have continuous segments in EL. I hope you will add a winning number history difference filter instead and also the winning number history difference statistics for just the winning numbers by draw date I described to you in an earlier post.
[/quote]

please add it as a feature request to mantis bug tracker, thanks
Expert Lotto Team
User avatar
stan
Site Admin
 
Posts: 6338
Joined: Thu Sep 23, 2004 1:01 pm

Postby Lexus » Sun Oct 25, 2009 12:45 am

Hi Stan!

Whether introduction jCUDA in EL5 for a productivity raising is possible

References on a theme:

http://www.hoopoe-cloud.com/Solutions/j ... fault.aspx

http://en.wikipedia.org/wiki/Parallel_computing

Or search on CUDA + JAVA
Lexus
 
Posts: 42
Joined: Sun Aug 09, 2009 11:02 am

Postby stan » Sun Oct 25, 2009 12:44 pm

[quote=Lexus]
Whether introduction jCUDA in EL5 for a productivity raising is possible
[/quote]

well, most filtering & stats algorithms are simple integer math so disk i/o operations are the limiting factor.
other algorithms which take a lot of time to compute (e.g. minimizer) cannot be easily parallelized
Expert Lotto Team
User avatar
stan
Site Admin
 
Posts: 6338
Joined: Thu Sep 23, 2004 1:01 pm

PreviousNext

Return to Development versions

Who is online

Users browsing this forum: No registered users and 9 guests

cron