Numbers Order in Column

Postby Jerzy » Fri Dec 11, 2009 3:25 am

Joe,

Could you please review the Numbers Order in Columns filter to see why the results of running Preview and Back-test are not the same if we place all the past winning combinations in the package?

I use ver. 1.1.0 of that filter.
Jerzy
 
Posts: 811
Joined: Mon Sep 18, 2006 5:26 am

Postby lottoboy » Fri Dec 11, 2009 9:28 am

Jerzy,

You are close to the door! Go continuely please!

Best,
lb
lottoboy
 
Posts: 738
Joined: Sat Dec 08, 2007 8:37 pm

Postby Joe » Fri Jan 01, 2010 8:18 pm

[quote=Jerzy:1260494743]Could you please review the Numbers Order in Columns filter to see why the results of running Preview and Back-test are not the same if we place all the past winning combinations in the package?

I use ver. 1.1.0 of that filter.
[/quote]preview and back-test functions are different functions, preview function applied to package with latest history differences alone but back test applied to winning combination in WN database with updating each winning numbers accordingly, both cant produce the same result
Joe
 
Posts: 1227
Joined: Mon Sep 18, 2006 6:51 pm
Location: Petaling Jaya, Malaysia

Postby Jerzy » Sat Jan 02, 2010 5:19 am

Joe,

Preview and Back-test functions are different functions and I have no problem with that. However, if both functions test exactly the same winning numbers combinations, they should return exactly the same result. And thats what happens in the standard filters.

Try Preview and Back-test in Odd/Even or in Sum filters.

Insert into the package all your winning numbers combinations and then open one of the standard filters. Select 'All draws', and use any setting of your choice, run Preview and after that Back-test. Both results will be exactly the same.

The problem appears in segment filters where both functions return quite different results while testing the same winning numbers combinations. And I do not know which result is correct. Skipping the last draw from the Winning Number data or removing the last draw winning combination from the package does not rectify the problem.

This problem occurs not only in your filters but also in the Stan's segment filters, for example, Segment Combinations filter.
Jerzy
 
Posts: 811
Joined: Mon Sep 18, 2006 5:26 am

Postby Jerzy » Sat Jan 02, 2010 5:27 am

Stan,

Could you please comment on the problem that I described in my last post addressed to Joe?

In segment filters the Preview and Back-test functions return quite different results when they test exactly the same set of winning numbers combinations. Which result is correct? Or, if both results are only approximations, which result is more reliable?
Jerzy
 
Posts: 811
Joined: Mon Sep 18, 2006 5:26 am

Postby stan » Sat Jan 02, 2010 11:17 am

[quote=Jerzy]
Stan,

Could you please comment on the problem that I described in my last post addressed to Joe?

In segment filters the Preview and Back-test functions return quite different results when they test exactly the same set of winning numbers combinations. Which result is correct? Or, if both results are only approximations, which result is more reliable?
[/quote]

each result reflects a different situation. the back-test shows how the filter performed on past winning numbers.
when applied on winning numbers in the package then it's the same as if numbers of each draw in the database were drawn again after the latest draw. so it has very little statistical value.
inserting winning numbers into the package makes almost no sense. so this feature won't be available in version 5.
Expert Lotto Team
User avatar
stan
Site Admin
 
Posts: 6338
Joined: Thu Sep 23, 2004 1:01 pm

Postby Jerzy » Sat Jan 02, 2010 11:25 am

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

Postby lottoboy » Sun Jan 03, 2010 4:26 am

Stan/Joe,
I agreed with you about Preview applied to package with latest HD and Back-test applied to winning combination with latest HD. (It‘s with HDs of Winning #s not winning #s itself.)

Sure, they are two different functions but the both functions have some relationship in statistical since the winning combinations are the part of the package. And our goal just is to find (or filter up) the Winning combinations from the Package

I have made more than 50 back-tests for pick-5 and pick-6 about their relationships which I have shown you by Posts and E-mails but I have not yet received any reply about that from you.

In fact, the relationships between Pre and BT for Winning Orders in Number Order Filter are interesting and clear but they're very unstable now.

In addition, Number Order Filters (5xx & 6xx) are different from other Filters in EL since its two special features. The first is only one Winning Order is in the 5 positions. The feature of “one order to choose from 5 ordersâ€￾ gives us more options for selecting the Winning Order. The second is the Preview (#s) and BT (%) of Winning Orders should be restricted by the full tickets and % percent

It should produce some very good strategies in the future if you can improve the program further for controlling their relationships according to the statistical law.

Note: My idea above is different from Jerzy's opinion.
lottoboy
 
Posts: 738
Joined: Sat Dec 08, 2007 8:37 pm

Postby lottoboy » Sun Jan 03, 2010 4:51 am

Sorry, this is an error for editing job.
lottoboy
 
Posts: 738
Joined: Sat Dec 08, 2007 8:37 pm

Postby stan » Sun Jan 03, 2010 9:19 am

sorry, but i don't know what you're asking me for...
Expert Lotto Team
User avatar
stan
Site Admin
 
Posts: 6338
Joined: Thu Sep 23, 2004 1:01 pm

Postby lottoboy » Mon Jan 04, 2010 6:15 am

[quote=stan]
sorry, but i don't know what you're asking me for...
[/quote]

My means are as below:

1) How about your opinion to my idea in last posts? I can tell you again and further if you have any question about my idea.

2) If you agree with my idea about Number Order Filter, could you
make some things for the Filter:

a) Could you check both Filters 5xx & 6xx to find some errors or bugs since their work is abnormal now.

b) From my backtests for both games of pick-5 and pick-6 by using Number Order Filters, I found the Sum of Previews for 5/6 winning orders always fluctuates around the value of Full Tickets (such as 575757/13983816). However, the deviations are too big to work normally. The truth in the Filter shows that not only the Previews of Winning Orders have to be restricted by the full tickets under the statistical law but the Filter has some problems. Could you improve the algorithm of the program to try to control this deviation is minimal (as few as possible). In statistical theory, the deviation should be zero. In other words, the Sum of Previews for 5/6 winning orders should equal to the full tickets exactly.

c) Also, from my backtests, I found the Sum of BTs(%) for winning orders always fluctuates around the value of 100%. However, the deviations are too big to work normally. The truth in the Filter shows that not only the BTs of Winning Orders have to be restricted by the 100% under the statistical law but the Filter has some problems. Could you improve the algorithm of the program to try to reduce the deviation as many as possible. The deviation should be zero theoretically.

d) According to the statistical law, the value of Bts (%) should match the values of Preview (#s) proportionally. In other words, the big value of BT should match the big value of Preview. For example, the Preview’s value of BT=29% should be greater than the Preview’s value of BT=17%. If not, or the deviation between the theoretical values and the practical values for the Winning Orders are too big. I think it’s abnormal or something is wrong in the program (code). Could you resolve the problem in the Filter too?

I had asked Joe to resolve these problems above since last year but no answer until now. I’m afraid that these problems come from the main program instead his Plug-in. I think firmly only you can resolve those even it's easy to say but hard to do.
Any question about my problems above please let me know asap.
Thanks.
lottoboy
 
Posts: 738
Joined: Sat Dec 08, 2007 8:37 pm

Postby stan » Tue Jan 05, 2010 3:47 pm

[quote=lottoboy:1262578548]
[quote=stan]
sorry, but i don't know what you're asking me for...
[/quote]

My means are as below:

1) How about your opinion to my idea in last posts? I can tell you again and further if you have any question about my idea.
[/quote]
i don't see any relationship between back-testing and preview for any expert lotto filter. please elaborate on that.

2) If you agree with my idea about Number Order Filter, could you
make some things for the Filter:

a) Could you check both Filters 5xx & 6xx to find some errors or bugs since their work is abnormal now.

i don't have filter's source code and since i didn't follow this thread i don't even know how that filter works.

b) From my backtests for both games of pick-5 and pick-6 by using Number Order Filters, I found the Sum of Previews for 5/6 winning orders always fluctuates around the value of Full Tickets (such as 575757/13983816). However, the deviations are too big to work normally. The truth in the Filter shows that not only the Previews of Winning Orders have to be restricted by the full tickets under the statistical law but the Filter has some problems. Could you improve the algorithm of the program to try to control this deviation is minimal (as few as possible). In statistical theory, the deviation should be zero. In other words, the Sum of Previews for 5/6 winning orders should equal to the full tickets exactly.

same response as above

c) Also, from my backtests, I found the Sum of BTs(%) for winning orders always fluctuates around the value of 100%. However, the deviations are too big to work normally. The truth in the Filter shows that not only the BTs of Winning Orders have to be restricted by the 100% under the statistical law but the Filter has some problems. Could you improve the algorithm of the program to try to reduce the deviation as many as possible. The deviation should be zero theoretically.

d) According to the statistical law, the value of Bts (%) should match the values of Preview (#s) proportionally.

not necessarily

In other words, the big value of BT should match the big value of Preview. For example, the Preview’s value of BT=29% should be greater than the Preview’s value of BT=17%. If not, or the deviation between the theoretical values and the practical values for the Winning Orders are too big. I think it’s abnormal or something is wrong in the program (code). Could you resolve the problem in the Filter too?

i don't have filter's source code
Expert Lotto Team
User avatar
stan
Site Admin
 
Posts: 6338
Joined: Thu Sep 23, 2004 1:01 pm

Postby lottoboy » Wed Jan 06, 2010 4:52 am

Stan,

"i don't see any relationship between back-testing and preview for any expert lotto filter. please elaborate on that."

Please read my posts on 12/10/2009 in the Forum elaboratelly.

The best way is to make some backtest for 5xx or 6xx by using Number Order
Filter and check the Winning Orders from the list of all BT and Preview for all positions (25 positions for 5xx and 36 positions for 6xx) like me have done that. Even though it's time consuming but you can find the relationship between BT and Preview is very clear and strong in the Filter!!!

"c) Also, from my backtests, I found the Sum of BTs(%) for winning orders always fluctuates around the value of 100%. However, the deviations are too big to work normally. The truth in the Filter shows that not only the BTs of Winning Orders have to be restricted by the 100% under the statistical law but the Filter has some problems. Could you improve the algorithm of the program to try to reduce the deviation as many as possible. The deviation should be zero theoretically.

d) According to the statistical law, the value of Bts (%) should match the values of Preview (#s) proportionally. "

not necessarily

I'm confusioned why do you know "the Match proportionally" is not necessary
when you don't see any relationship between back-testing and preview for any expert lotto filter???

Any question about that pls let me know asap. Thanks.
lottoboy
 
Posts: 738
Joined: Sat Dec 08, 2007 8:37 pm

Postby lottoboy » Wed Jan 06, 2010 4:53 am

Joe,

Could you reply my post and resolve these problems in your
Number Order Filter? Thanks.
lottoboy
 
Posts: 738
Joined: Sat Dec 08, 2007 8:37 pm

Postby lottoboy » Tue Jan 12, 2010 4:36 am

Stan/Joe:

Could you give me an answer about my questions? Thanks.
lottoboy
 
Posts: 738
Joined: Sat Dec 08, 2007 8:37 pm

PreviousNext

Return to Plugin Development

Who is online

Users browsing this forum: No registered users and 1 guest