dev build 110513

[stan]

new development build is now available as application update (menu 'help - check for updates')
it's possible to adjust chart background color in this build and there are also some other minor tweaks.

[Jerzy]

Stan,

The occurrence threshold in Options --> History is a very valuable feature. But it is a bit difficult to use this feature for filtering many levels not knowing what was the real occurrence threshold for previous draws.

Calculating the range of the occurrence threshold for previous draws (or at least for the latest draw) and displaying this information in the Simulation or My Estimates data would help a lot.

Would you mind expanding this new feature?

[stan]

Stan,

The occurrence threshold in Options --> History is a very valuable feature. But it is a bit difficult to use this feature for filtering many levels not knowing what was the real occurrence threshold for previous draws.

Calculating the range of the occurrence threshold for previous draws (or at least for the latest draw) and displaying this information in the Simulation or My Estimates data would help a lot.

Would you mind expanding this new feature?

i'm not sure i'm following you. how is previous history value related to the range of values used as draw simulation estimates hints?

[Jerzy]

Stan,

I am talking about the occurrence threshold, which is accessible from Options --> History --> Estimates. It is the value adjustable by the user. It applies separately to each level but can be applied to all levels. Actually it could be most effective if it is applied to all levels, but it should be done correctly.

Currently, the users enters the occurrence threshold of his choice (ideally it should be informed choice not guessing) and then runs the simulation, enters the range of differences, which is limited by the threshold, to My Estimates and filters the package. Whether the differences that cross the threshold include the winning differences we would know after the draw.

But is the reversed process possible? If we know the latest winning differences in all levels, is it possible to calculate the latest correct occurrence threshold (it would accommodate the latest differences in all levels) and show it in My Estimates window?

It seems to me that currently there is no easy way of calculating the latest occurrence threshold for known differences in a group of levels levels.

If, for example, I apply the occurrence threshold 1.50 to 15.50% and after the draw I know that it was wrong choice I would like to know what value I should enter to avoid similar mistakes in the future.

[teoman]

If, for example, I apply the occurrence threshold 1.50 to 15.50% and after the draw I know that it was wrong choice I would like to know what value I should enter to avoid similar mistakes in the future.

It should appear as statistics to the 'Properties' in function of number of levels (88 levels for 894 draws) chosen by the user, then make averaged the percentages of the latest 50-100 draws and these values ​​can be used for the next draw.
The values ​​below are theoretical.

~~~~Date~~~~~~~~~Numbers~~~~~~~~~Occurrence treshold(Min)~~Occurrence treshold(Max)
15.05.2011 D ~~ 01 11 15 17 40 28 ~~~~~~~~~~~ 2,15% ~~~~~~~~~~~~~~~~~ 33,73%
12.05.2011 J ~~ 05 06 13 16 24 27 ~~~~~~~~~~ 11,02% ~~~~~~~~~~~~~~~~~ 22,37%
08.05.2011 D ~~ 03 06 21 25 38 16 ~~~~~~~~~~ 14,10% ~~~~~~~~~~~~~~~~~ 20,78%

[Jerzy]

Thank you teoman!

Yes, the past occurrence thresholds statistics would be very helpful and probably necessary to complete the issue of filtering on the basis of the occurence threshold.

Regards

Jerzy

[stan]

the threshold limits relate to the most common difference found during the simulation. so if i were to give thresholds for past draws then the minimum and maximum threshold values would be the same.
for example let's assume the difference in draw N is -15. when running draw simulation the most common difference is +23 and has occurrence of e.g. 200000. the occurrence of difference -15 is e.g. 150000. so the actual thresholds for draw N are min=75% and max=75%

the idea is that you don't change the min/max thresholds too often (that's why they are in options window instead of being available in the simulation dialog window). you should enter values to get a section of the simulation histogram which has a high chance of including the actual winning history value but discards differences that occur very seldom (i.e. the left and right side of the histogram chart)

[Jerzy]

Stan,

I would like to show you how I calculate the latest occurrence threshold using a pen, paper and a calculator.

1) I write down the latest WNH Diffs in all levels. In this example I use 21 levels (Australian Saturday Lotto after 14/05/2011 drawings). There were the following winning differences:
+7, +10, +9, -4, -33, +25, 0, -15, +14, +3, -23, +1, -41, +10, -31, +18, +10, +28, +18, -11, -6

2) I hide the last draw in Winning Numbers panel

3) I run the simulation

4) I find out the frequency of winning difference and the highest frequency in each level and calculate the occurence threshold as follows:

L. 0 ( +7 ) 218,622 (+14) 244,515 OT = 89.41%
L. 1 (+10 ) 118,437 ( -6 ) 218,510 OT = 54.20%
L. 2 ( +9 ) 239,555 ( +7 ) 242,995 OT = 98.58%
L. 3 ( -4 ) 174,503 ( +6 ) 221,740 OT = 78,69%
L. 4 (-33 ) 50,895 ( +1 ) 188,782 OT = 26.95%
L. 5 (+25 ) 70,692 ( -5 ) 243,982 OT = 28.97%
L. 6 ( 0 ) 157,917 (+13) 238,630 OT = 66.17%
L. 7 (-15 ) 117,491 ( +7 ) 169,992 OT = 69.11%
L. 8 ( +14) 298,167 (+17) 314.355 OT = 94.85%
L. 9 ( +3 ) 163,682 (+12) 243,971 OT = 67.09%
L.10 ( -23) 24,986 ( +8 ) 256,184 OT = 9.75%
L.11 ( +1 ) 226,363 ( +4 ) 232,063 OT = 97.54%
L.12 (-41) 44,086 ( -6 ) 175,552 OT = 25.11%
L.13 (+10) 201,070 (+16) 228,947 OT = 87.82%
L.14 (-31) 59,165 ( -1 ) 180,210 OT = 32,83%
L.15 (+18) 152,429 ( +8 ) 245,431 OT = 62.10%
L.16 (+10) 215,843 (+21) 246,432 OT = 87.58%
L.17 (+28) 24,324 ( -3 ) 202,218 OT = 12.02%
L.18 (+18) 267,820 (+14) 286,188 OT = 93.58%
L.19 ( -11) 180,085 ( -5 ) 191,934 OT = 93.83%
L.20 ( -6 ) 221,179 ( -3 ) 225,304 OT = 98.16%

The lowest occurrence threshold for negative differences is 9.75% (in level 10). I round it down to 9.50% to make it acceptable to EL.
The lowest occurrence threshold for positive differences is 12.02% ( in level 17). I round it down to 12.00.

It means that min threshold for all 21 levels = 9.50% and max threshold = 12.00%
I adjust the occurrence threshold in Options --> History to these values, enter the values that cross the threshold into My Estimates and run the filter.

Then I uncover the latest draw result in the Winning Numbers panel and check the winnings.

Results:
Accepted 4,433,484 tickets
Rejected 3,711,576 tickets (total 8,145,060 tickets in the package)
Jackpot ticket present
2 nd prizes 6
3 rd prizes 108
4 th prizes 6,115
5th prizes 16,573

It is interesting that the lowest negative difference was -41 (L12) but the min occurrence threshold was created by the difference -23 (L10).

Calculating the occurrence threshold for 21 levels took me appoximately 1.5 hour. If EL5 could do it for me, it would be better.

[teoman]

the idea is that you don't change the min/max thresholds too often (that's why they are in options window instead of being available in the simulation dialog window). you should enter values to get a section of the simulation histogram which has a high chance of including the actual winning history value but discards differences that occur very seldom (i.e. the left and right side of the histogram chart)

Stan, I adjust the occurrence threshold in 'Options' => 'History' => 'Estimates', for example Min 5.75% and Max 12.48% and in the 'Summary History' execute 'Simulation', then enter the values that cross the threshold into 'My Estimates' and run the filter, obtain a tickets number that is not wanted, then I want to again adjust occurrence threshold in 'Options' => 'History' => 'Estimates' and must run again in the 'Summary History' => 'Simulation' and of each time the same complete cycle.
Not it's possible that to a single 'Simulation' to make many adjustments to the occurrence threshold in 'Options' => 'History' => 'Estimates', for find suitable values, without doing each time 'Simulation'?

[stan]

so you're talking about min/max occurrence percentage across all levels, right? i guess it does make sense then
i'll try to come up with something for this then.

[teoman]

so you're talking about min/max occurrence percentage across all levels, right? i guess it does make sense then
i'll try to come up with something for this then.

Yes, after I executed the 'Simulation', I can that adjust occurrence threshold (in 'Options' => 'History' => 'Estimates') of several times until the values ​​are appropriate and I obtain a tickets number which are convenient, but to not more be need to run the 'Simulation' of each time, how many times I'll change the percentage values (in 'Options' => 'History' => 'Estimates'), than when using a different draw.

[Jerzy]

Yes Stan.

Min/max occurrence percentage across all levels would be a fully developed and useful feature.

[PadawanLotto]

for example let's assume the difference in draw N is -15. when running draw simulation the most common difference is +23 and has occurrence of e.g. 200000. the occurrence of difference -15 is e.g. 150000. so the actual thresholds for draw N are min=75% and max=75%

Need more info here on your example.
1. (The difference in draw N is -15.) There are more than one difference in a drawing. Was the -15 the lowest neg occurrence in the drawing?
2. (When running draw simulation the most common difference is +23.) When I simulate using a full package for Rolling Cash 5/39 the most common positive diff is +34 or +33 among the 11 levels. Which level do you get the occurrence from?

[PadawanLotto]

the threshold limits relate to the most common difference found during the simulation. so if i were to give thresholds for past draws then the minimum and maximum threshold values would be the same.
for example let's assume the difference in draw N is -15. when running draw simulation the most common difference is +23 and has occurrence of e.g. 200000. the occurrence of difference -15 is e.g. 150000. so the actual thresholds for draw N are min=75% and max=75%

I didn't now how to find the threshold based on the example above but, I played around a little and found that min 20% and max 20% works great for the Rolling Cash 5/39. I back tested a wide range of drawings and better than 90% passed. The amount of tickets removed was around +/- 50%, nice. Still a lot of filtering has to be done but it's a good start.

Padawan

[chezzmaster]

same here.. im lost.. is there any way to do automate this prosess ?