Numbers rating

[anubina]

Concept:
1 Rate all numbers with an index and sort with this index (the index would be an aggregate of several filters).
2 Get different combinations in groups of 5 in 5 numbers (EuroMillions), with breaks between each combination number.
Optionally 3rd filter combinations obtained.

Gaps:
1 The system improves the more concentration available of the winning numbers.
2 is not thought delete numbers or combinations generated by the system.

Benefits:
1 Allows qualify for an individual filter, up to an aggregate of various filters.
2 What is important is the concentration of winning numbers and not if they have a high rating or low.

Evidence of effectiveness:
* Latest 100 draws.
* 100 random draws.
* Calculate the distance between winning numbers according to the scorecard.

Example (Euromillions):
Number Rated Winner?
1 45 1
2 23
3 87 1
4 2
5 0
...
19 5
20 0
21 1 1
23 2
...
43 100 1
44 90
45 5
46 37
47 4 1
48 18
49 23
50 43


In the example shown, it is noted that different winning numbers have the following values:
Number Rated Winner?
1 45 1
3 87 1
21 1 1
43 100 1
47 4 1


The concentration of numbers sorted by the rating index is very low (there are numbers with values ​​intermediate rating).

To form the rating index should be done with lots of filters.

 

[Frank]

Hi Anubina and welcome to the forum. Thank you for sharing your thoughts on lottery analysis. Whilst I understand the general idea of rating numbers it's not clear how you are doing this using filters. Most people would use filters to eliminate combinations without necessarily gaining any information about specific numbers within those combinations. I rate my numbers by collecting a shortlist from system A , another shortlist from system B etc. then I pool all the balls from all systems within a counting grid and count the duplicates, the most counts for each number gives it a rating. I only use the highest scoring balls. I suppose you could call that a kind of filter.

I don't follow how you are defining a winning number, presumably you test your final shortlist against the last 100 draws, is that what you are saying? That would make them "would have won" numbers had those draws occurred in the future?

I'm assuming, looking at your listing of rated numbers that the last column winner? has a 1 to denote a winner, otherwise blank.
How does this information giving you different "winning numbers" from different past draws help you to combine these winning numbers for betting on a future draw?

 

[anubina]

Hi Anubina and welcome to the forum. Thank you for sharing your thoughts on lottery analysis. Whilst I understand the general idea of rating numbers it's not clear how you are doing this using filters. Most people would use filters to eliminate combinations without necessarily gaining any information about specific numbers within those combinations. I rate my numbers by collecting a shortlist from system A , another shortlist from system B etc. then I pool all the balls from all systems within a counting grid and count the duplicates, the most counts for each number gives it a rating. I only use the highest scoring balls. I suppose you could call that a kind of filter.

I don't follow how you are defining a winning number, presumably you test your final shortlist against the last 100 draws, is that what you are saying? That would make them "would have won" numbers had those draws occurred in the future?

I'm assuming, looking at your listing of rated numbers that the last column winner? has a 1 to denote a winner, otherwise blank.
How does this information giving you different "winning numbers" from different past draws help you to combine these winning numbers for betting on a future draw?

Hi Frank!

I'm here :finger:

Well is very hard explain the concepts in "googlian", then... be patient & compresive.

Yes, Mark 1 means that for that drawing in particular, the number came out the winner , having its own rating.

In the experiment conducted over 100 drawings , a lousy efficiency rating was observed (using simple or compound models).

Perhaps more interesting not delete numbers, but make them appear less periodicity in the combinations ; for example , you can have a combination with 5 numbers with very good rating ( statistical mode ) , but it can happen 3 numbers leaving with good rating, 1 with average rating and 1 with bad rating.

Then the structural filter combination could be say 5 + 0 + 0 + 1 or 3 + 1, although it would be very unusual combinations they had had 5 winning numbers with a poor rating.

Working with SQL ( Oracle ) for ETL and project R for data analysis.

 

[Frank]

Well good luck with the system, hope it brings you success.

By the way, just so you know, this emoticon :finger: is actually a rude gesture. I suggest you don't use it unless you really want to offend someone.

 

[anubina]

Well good luck with the system, hope it brings you success.

By the way, just so you know, this emoticon :finger: is actually a rude gesture. I suggest you don't use it unless you really want to offend someone.

Ups, sorry, I don't see very well.

two types for obtain de agregate of filters (rating):

1.- Proportional.

The sums of the filters and after order the number and re-rating with the order of the sumfilter​

2.- Nominal (the most common).

The rating is the sums of the filter​

The problem with Nominal rating is when sums differents rating (for example with different magnitude)...

Is more effectiveness the proportional system.

Some words for definitions (I have doubts in english):

Effectiveness: % Degree of solution for a problem.
Effective: % Degree of guarantee for the solution.
Efficiency: % Consumption of resource.

 

[sangoma123]

Well good luck with the system, hope it brings you success.

By the way, just so you know, this emoticon :finger: is actually a rude gesture. I suggest you don't use it unless you really want to offend someone.

Hello,Frank:

Two years ago,in 2013:

"Pab, I am sorry.

There has been a GREAT misunderstanding.

I don´t wise that the smiley was doing that gesture.

I was THINKING THAT THE SMILEY WAS DOING OK with both hands.

On the screen of my monitor I can´t saw that the smiley used THAT fingers.

I sit it indeed. I Never did to you that, when always you have helped me.


Regards and SORRY again".

(History repeats again...)

Damn emoticon!!!:

Regards

P.D. Pab,we missing you.I hope you are in a country with good beaches...

 

[Frank]

Yes, it's too small to be seen clearly, and I don't even know why it's available, we are supposed to be a more respectable forum, why would we need it?

I echo your sentiments about PAB , I'd love to know what we did to offend him, he didn't even say goodbye.

 

[anubina]

Is necessary filter numbers? Maybe No.
Is necessary filter draws? Yes.

Then if we use the rating of the numbers (not for filter numbers), then we can use this rating for modulate (key concept) the bets/draws.


But, modulating the bets, then you obtain a core numbers very common in this draws, then you need other tool (smart bets o wheel), then you can optimice the pyramid of potential awards.


The pyramid of potential awards in Euromillions is:

Draws with guarantee prize: ~8%
Draws with potential prize: ~42% (depend of the Stars numbers)

Draws-MainBalls-%total
1-5-~0.00005%
225-4-~0.01%
9,900-3-~0.5%
141,900-2-~7%
744,975-1-~35% (Potential prize, depend of the Stars numbers)
1,221,759-0-~58% --> Target of the filters

A system with >42% of potential draws (with prize) in consistent time is a competitive edge...

 

[anubina]

Hi Anubina and welcome to the forum. Thank you for sharing your thoughts on lottery analysis. Whilst I understand the general idea of rating numbers it's not clear how you are doing this using filters. Most people would use filters to eliminate combinations without necessarily gaining any information about specific numbers within those combinations. I rate my numbers by collecting a shortlist from system A , another shortlist from system B etc. then I pool all the balls from all systems within a counting grid and count the duplicates, the most counts for each number gives it a rating. I only use the highest scoring balls. I suppose you could call that a kind of filter.

I don't follow how you are defining a winning number, presumably you test your final shortlist against the last 100 draws, is that what you are saying? That would make them "would have won" numbers had those draws occurred in the future?

I'm assuming, looking at your listing of rated numbers that the last column winner? has a 1 to denote a winner, otherwise blank.
How does this information giving you different "winning numbers" from different past draws help you to combine these winning numbers for betting on a future draw?

This shorlist of your systemA and systemB is nominal or proportional? has the same numbers of metrics or filters for each system?

Is this correct?
Rating (for me) = Counting grid (for you)
KPI (for me) = system X (for you)
Metric (for me) = shortlist or filter? (for you)

I'm assuming that scoring is the value of the rating for a ball.

 

[anubina]

( googlian )
Qualifying for me is how to characterize an individual (number , combination or absence) .

Once characterized the individual in a particular draw ( for euromillones ) :
Numbers : 50
Combinations : 2118760
Absences: ~ 247 (draw order )

Study the spectrum of each individual rating , noting the level of rating and concentration.

I repeat this process 100 times , so I get a series of data that expose below :

30% of the draws are not aligned with the statistical mode . I accept this figure as entropy of the system and not spend efforts to achieve positive results in this population.

70% of the draws are aligned with the statistical mode . The level of concentration of individuals of this statistical mode is between 40 % to 70 % or in other words 100 draws, there are 70 drawings which can be removed from 640.000 to 1,300,000 combinations .

 

[anubina]

I have been analyzing euromillones 33 combinations without applying filters, only computing the number density and density combinations of numbers of effective combinations (having> = 1 winning number).

The rating of each combination, you have done according to criteria he used in the filters; that classification applied to the 33 combinations, obtaining a bell curve, and being able to observe the red line, which are observed per rating (x axis) the number of combinations (y-axis); and the green line are effective combinations (which have> = 1 winning number).

By averaging the 33 combinations, the following results were obtained:

1 density combinations between a winning combination and the next winning combination have an average of ~ 29% of total combinations (Euromillions ~ 609,000 combinations).

2nd Of the above combinations, there is a 42% of those combinations that are effective (natural% the game itself).

3 The efficiency of the system calculated as 42% -29% = 13%

For a very efficient system, we have:
42% -1% = 41% -> Very Good, it indicates that the deviations between a rating and one of the draws is very stable, however, between the number of combinations there is a 41% hit> = 1 number.

70% -29% = 41% -> Very Good, it indicates that the deviations between a rating and one of the drawings varies greatly, however, between the number of combinations there is a 70% hit> = 1 number.

72% -1% = 71% -> Excellent, indicating a very stable rating (21,000 combinations) and combinations of those few, you have a 71% of hitting> = 1 issue (14,700 combinations).

Then I'll upload some pictures with bells gaus (characterization of draws)

 

[Icewynd]

Is necessary filter numbers? Maybe No.

One method that I use to filter numbers is a measure of how "due" a number is. This is defined for each number as:

(# of draws/total number of hits for number)-current skip of number

The values range from about +8 to -20. The numbers that hit most often have a value of -3 to +6. I have had good success eliminating numbers with values too high (+7 or +8) and numbers with values that are too low.

Sounds like you are making some progress in the battle against Euromillions!

Good Luck!

 

[anubina]

One method that I use to filter numbers is a measure of how "due" a number is. This is defined for each number as:

(# of draws/total number of hits for number)-current skip of number

The values range from about +8 to -20. The numbers that hit most often have a value of -3 to +6. I have had good success eliminating numbers with values too high (+7 or +8) and numbers with values that are too low.

Sounds like you are making some progress in the battle against Euromillions!

Good Luck!

OK Icewynd, I'm going to characterize the rating with your filter, and launch the 33 draws (the same draws) and when I have the result I post it.

 

[anubina]

I've done a filter testing IceWynd on the following model:

- Game: Euromillions (main balls).
- Number of draws: 33 draws; from #783 03/24/2015 (DD/MM/YYYY) to #815 14/07/2015 (DD/MM/YYYY).
- Filter to be tested: (# of draws / total number of hits for number) -current number of skip.


results:

- Combinations density: 39%
- Density of effective combinations: 43%
- Performance filter: 43% - 39% = 4%
- Calculation Total Time: ~ 2 hours

Add charts last 5 drawings:

EUR815 http://www.mediafire.com/view/d5r7yf15pm9444p/EUR815.png
EUR814 http://www.mediafire.com/view/d5zd6l29a8wejss/EUR814.png
EUR813 http://www.mediafire.com/view/c2erb2kv73lgjm6/EUR813.png
EUR812 http://www.mediafire.com/view/x3szxf0an4zrdn2/EUR812.png
EUR811 http://www.mediafire.com/view/x3szxf0an4zrdn2/EUR812.png

PD. The legend of the axes of the graph are mistakenly interchanged. The X-axis is the rating and the Y-axis is the number of combinations for each rating.

The green peak indicates the winning combination.

 

[Icewynd]

Interesting Anubina!

Could you explain a little more? What is an "effective combination"? And how are you calculating the rating score for the combination?

It appears from your charts that the winners are either in the 600 range or in the 1500-1650 range. Is that typical? What was the mean?

Good Luck!

 

[anubina]

Interesting Anubina!

Could you explain a little more? What is an "effective combination"? And how are you calculating the rating score for the combination?

It appears from your charts that the winners are either in the 600 range or in the 1500-1650 range. Is that typical? What was the mean?

Good Luck!

Let me explain the first questions, then this afternoon will discuss the test measurements (right now I'm not on the computer test). Also I will make a 2nd test with 100 drawings.

* Could you Explain a little more?

http://www.mediafire.com/view/8a5odv658d7kan9/KPI.png

Copy a basic model of the system I use to characterize the different filters, perform the test and observe the results. Explain different layers:

- Datasource: In this layer, the whole historical data are obtained (until the draw to predict). No treatment, extracting only the portion corresponding historical. Also I generate all possible combinations that could occur in the draw to predict (for example, jump to a specific number if you would come out in the next draw, or absolute or relative frequency).

- Metrics: In this layer, I create the filter condition that I want to measure. On the previous two datasources (historical and combinations), I generate the design proposed by IceWynd filter.

- Measures: In this layer, I can observe the behavior of the filter IceWynd around the historical, and also how this filter will behave in all possible combinations of the draw to predict. Filter made a ponderated sum for each draw and historic predicted for each combination.

- Characterization (KPI Key Performance Indicator): In this layer, I assign to each combination, the weighted value that would correspond both historical analysis and forecast analysis of combinations. Later, I back to ponder this characterization.

- Additionally you could create other KPIs with other filters or combinations thereof, and add them in a balanced way (not applicable in this proposed IceWynd example).


* And how are you calculating the rating score for the combination?

- I'd rather preserve this information. I indicated that the calculation method is always the same, and the only thing that change are the different filters in the Metrics layer.


What is an "effective combination"?

I will first define some words and concept:

- Useful / effective: Value that gives me a process / system. For example, if I have a car and I break the brakes, for me it is very useful workshop vehicle.

- Effective: Measure to determine whether a process, system or utility has enough quality to do the job properly. For example, if I have a car and I break the brakes, I would like when I repaired the brakes are correct and properly halt.

- Efficient: Measure to determine whether a process, system or utility has a proper use of resources (time, money). For example, if I have a car and I break the brakes, I would like when I repaired in no time and money.

If I have no car, no use to me a very effective and efficient workshop, and it is not helpful to me.

If I have a broken car, it does not serve me a very efficient workshop, if not then I repaired the brakes properly.

If I have a broken car, I have to rate among the different workshops to determine which is the more guarantees that performs repair (efficiency) and the best response time and price have.

An effective combination, refers to a combination that at least have a winning number combination. For instance:

Suppose that for the next draw, the winning combination out 5, 16, 27, 33, 46. Then:

1, 2, 3, 4, 5 -> It is effective because it has a number of the winning combination.
1, 2, 3, 4, 6 -> It is not effective because it does not possess any of the winning number combination.

 

[anubina]

The same 33 Draws Euromillions, with corrections in my system and using IceWynd Filter, in the next graph, you can see the bell standard (I use the last draw, very similar to the others), and in blue I put the wining combination:

http://www.mediafire.com/view/8qzcuu385d877j7/33EuromillionsIceWynd.png